ANNEX 1 FINAL REPORT FOR PROJECT FS145003 HISTORICAL DATA …

FS145003 Final report ANNEX 1 v6 of 9

ANNEX 1

FINAL REPORT FOR PROJECT FS145003

HISTORICAL DATA- ANALYSIS FOR OBJECTIVE 4

The aim of this study was to compare the various conditions that are found at the traditional

inspection of fattening pigs from indoor fattening and outdoor fattening (free range) pigs in

the study abattoir.

METHODS

Inspection data for pigs slaughtered at one pig-only abattoir in the east of England from

January 2010 to December 2011 were obtained from the Food Standards Agency (FSA).

There were three datasets for each of the batches slaughtered. Data were available for ante-

mortem inspection (AMI) of live animals, post-mortem inspections (PMI) of carcase

inspection and offal inspection.

The batches were categorised as being either from free range systems (fattened outdoors),

or from indoor fattening systems. This was determined by the batch slap mark. This

indicates the farm of origin. The batch slap marks were compared with a list of the

categories of suppliers of pigs obtained from the abattoir management.

The data were transferred to a spread-sheet (Microsoft Excel) and manually inspected for

errors and anomalies; these were excluded from any analysis by replacing the data as

missing. A few of the conditions identified, namely mastitis and orchitis, were of such low

prevalence, with only one or two batches affected, that this was combined with the “other

condition” categories. The data were transferred to a statistical software package (STATA

version 12) for analysis.

The prevalence of each condition in each batch was determined for the datasets by using

the number of times the condition was identified as the numerator and the batch size as the

denominator.

The proportion of batches affected with a condition was determined by recoding the

presence or absence of a condition in a batch as a binary outcome. The two types of

finishing systems were then compared calculating confidence intervals and using a test for

proportions (Z-test). Confounding by batch size and season was investigated by comparing

odds ratios using the Mantel-Haenszel method.

The batches in which the conditions were found were also compared for the two finishing

systems. In this case the mean prevalence in the batches that were found to have the

condition was compared between the systems using a t-test.

Due to the number of analyses performed in the comparisons (n=82) any differences were

considered statistical significant when P<0.0006 using Bonferroni correction.


RESULTS

The total data available for the analysis contained results from the inspection of 1,220,340

pigs from 7,410 batches from the different rearing and fattening systems

In the analysis period, the mean number of pigs in each batch was 164 and 166 pigs for

indoor and free range fattening systems respectively. There is only slight variation in the

different batch size categories for the different fattening systems (Table A1:1). There was a

degree of seasonal variation in the number of batches of pigs slaughtered, with fewer

batches slaughtered in late spring and in the summer months (Figure A1:1). The proportion

of the batches that were from free range systems was fairly consistent accounting for

approximately a quarter (range 19.5-26.3%) of the batches of pigs slaughtered. Although

batch size and seasonality were considered as possible confounders, when analysed they

had no influence on the associations between the conditions found and the fattening

systems (data not presented).

Table A1:1: Size of the batches of the pigs submitted to slaughter from different fattening systems during 2010 to 2011

Batch size Fattening system Total

Indoor Free Range

n % n % n %

1-50 121 2.1 76 4.6 197 2.66

51-100 652 11.4 247 14.8 899 12.1

101-150 1 366 23.8 270 16.2 1 636 22.1

151-200 1 703 29.7 452 27.2 2 155 29.1

201-250 1 837 32.0 579 34.8 2 416 32.6

251-300 14 0.2 4 0.2 18 0.2

301-350 22 0.4 3 0.2 25 0.3

351-400 13 0.2 9 0.5 22 0.3

401-450 12 0.2 25 1.5 37 0.5

451-500 3 0.1 0 0.0 3 0.0

Total 5 743 100 1 665 100 7 408 100


Figure A1:1: The number of batches of pigs from each of the different fattening systems that were slaughtered at the abattoir during the analysis period (2010-2011).

Ante-mortem inspection data (Table A1:2).

The proportion of batches in which the various conditions were detected was similar for both

indoor and outdoor fattening systems, except that there was a significant increase in tail bite

and lameness in batches from indoor fattening systems. Indoor fattening systems had

26.4% of batches in which tail bite was identified, compared with 5.9% of free range batches,

and 46.5% had lame pigs compared with 24.7% of free range batches. There were no

differences between the two fattening systems in the mean prevalence of conditions within

the batches where the conditions were present.

392 492 522

402 450 436 472 432

573 541 484

549

140

138 154

117 109 130

137 150

141 134

147

168

0

100

200

300

400

500

600

700

800

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Nu

mb

er o

f B

atch

es S

lau

ghte

red

Month of Slaughter

Indoor Free Range


Table A1:2: Conditions found at ante-mortem inspection of slaughter pigs from indoor and free range fattening systems (*statistically significant* difference between systems)

Percentage of the batches affected

with the condition (95% C.I.)

Mean prevalence (%) in batches in

which the condition was present (95%

C.I.)

Indoor Free Range Indoor Free Range

Respiratory signs 6.9

(6.1-7.8)

8.3

(5.5-11.1)

0.8

(0.7-0.9)

0.7

(0.1-1.4)

Abscess 10.8

(9.7-11.9)

6.7

(4.2-9.3)

1.2

(0.7-1.8)

0.6

(0.5-0.8)

Dead or Dying 17.0

(15.7-18.3)

14.8

(11.2-18.4)

0.8

(0.8-0.9)

0.6

(0.5-0.7)

Skin lesions 1.3

(0.9-1.7)

0.8

(0.0-1.7)

2.0

(0.03-3.9)

0.9

(0.0-2.6)

Hernia/Ruptures 45.7

(44.0-47.4)

40.6

(35.6-45.6)

1.1

(1.0-1.2)

0.9

(0.8-1.0)

Trauma 5.4

(4.6-6.1)

4.6

(2.5-6.7)

1.0

(0.9-1.3)

0.8

(0.5-1.0)

Joint disease 8.9

(7.9-9.8)

11.8

(8.6-15.1)

0.9

(0.9-1.1)

1.3

(0.04-2.5)

Lumps/Swellings 1.7

(1.2-2.1)

0.5

(0.0-1.3)

1.1

(0.7-1.6)

0.5

(0.2-0.7)

Tail bite *

26.4*

(24.9-27.9)

*5.9

*

(3.5-8.3)

1.4

(1.2-1.6)

0.8

(0.6-1.0)

Neurological signs 0.2

(0.03-0.3)

1.1

(0.02-2.1)

0.5

(0.5-0.6)

0.5

(0.4-0.6)

Lame *46.5

*

(44.8-48.2)

*24.7

*

(20.3-29.1)

1.2

(1.0-1.3)

0.7

(0.6-0.8)

Stress 1.1

(0.1-1.5)

2.7

(1.0-4.3)

2.4

(0.0-5.8)

0.6

(0.5-0.7)

Other 1.3

(0.9-1.7)

1.3

(0.2-2.5)

0.9

(0.5-1.3)

0.6

(0.2-0.9)

*xx.x* (yy.y-zz.z) – statistically significant values are shown in bold text

Post-mortem inspection

The conditions found during carcase inspection were again similar between the two fattening

systems. There were differences in the proportion of batches with carcases showing oedema

and those with hair contamination (Tables A1:3 and A1:4). In indoor fattened pigs 25.7% of

batches had carcases of pigs with oedema compared with 8.1% for free range batches.


5.9% of free range pig batches had hair contamination compared with 3.9% for batches of

indoor fattened pigs. The prevalence of the conditions in pigs in which the various

conditions were found was again similar for the different fattening systems.

Table A1:3: Pathological conditions found on post-mortem inspection of the carcases of pigs slaughtered from indoor and free range fattening systems (*statistically significant* difference between systems)





C.I.)


Anaemia 5.5

(4.7-6.4)

3.2

(1.8-4.7)

0.8

(0.7-0.8)

0.8

(0.4-1.7)

Jaundice 3.3

(2.7-4.0)

5.4

(3.5-7.3)

0.7

(0.6-0.8)

0.6

(0.5-0.6)

Tumours 0.6

(0.3-0.9)

0.5

(0.1-1.2)

0.6

(0.5-0.7)

0.7

(0.04-1.4)

Oedema *25.7

*

(24.1-27.4)

*8.1

*

(5.8-10.3)

1.1

(1.0-1.2)

1.1

(0.5-1.6)

Poly-arthritis 8.0

(6.9-9.0)

9.7

(7.2-12.2)

0.9

(0.8-1.0)

0.8

(0.6-0.9)

Septicaemia 10.7

(9.5-11.9)

10.1

(7.6-12.6)

0.7

(0.7-0.8)

0.8

(0.6-0.9)

Pyaemia 75.5

(73.9-77.2)

71.1

(67.3-74.9)

1.5

(1.4-1.6)

1.1

(1.0-1.3)

Uraemia 1.0

(0.7-1.4)

1.1

(0.2-1.9)

0.6

(0.5-0.7)

0.6

(0.4-0.8)

Other Conditions 8.4

(7.3-9.4)

12.8

(10.0-15.5)

1.9

(0.8-0.9)

1.2

(0.7-1.6)



Table A1:4: Processing faults found at post-mortem inspection of slaughtered pigs from indoor and free range rearing systems (*statistically significant* difference between systems)





C.I.)


Badly Bled 1.3

(0.8-1.7)

1.6

(0.6-2.7)

1.0

(0.3-1.6)

0.6

(0.4-0.9)

Machine Damage 2.1

(1.6-2.7)

3.4

(1.9-4.9)

3.3

(1.2-5.4)

7.4

(1.3-16.0)

Blood Splash 0.9

(0.7-1.2)

0.4

(0.1-0.7)

0.7

(0.6-0.8)

1.0

(0.5-1.5)

Bile Contamination 93.6

(92.9-94.2)

92.7

(91.4-94.0)

4.2

(4.1-4.2)

4.1

(4.0-4.3)

Faecal Contamination 75.4

(74.3-76.5)

72.6

(70.4-74.8)

2.9

(2.8-3.0)

2.9

(2.7-3.0)

Grease Contamination 3.6

(3.1-4.1)

3.5

(2.6-4.4)

0.8

(0.7-0.9)

0.9

(0.8-1.1)

Hair Contamination *3.9

*

(3.4-4.4)

*5.9

*

(4.8-7.1)

0.8

(0.7-1.0)

1.0

(0.8-1.1)

Other 0.2

(0.1-0.3)

0.2

(0.03-0.4)

16.4

(0.0-38.5)

0.6

(0.0-1.8)


The comparison of the frequency of the pathological conditions found at post-mortem

inspection of offal for the different fattening systems (Table A1:5) revealed that milk spot liver

occurred in more batches of free range fattened pigs compared with indoor fattened ones

(54.8% of batches affected compared with 23.4%). By contrast, pericarditis was found to

occur in more batches of indoor fattened pigs than free range fattened pigs (60.6% of

batches compared with 53.5%). The proportion of batches with other conditions was similar

between the two fattening systems.

When the prevalence of conditions, found at offal inspection in batches where the conditions

occurred, were compared it was found that kidney pathology was significantly higher in

indoor fattened pigs, whereas the within batch mean prevalence of hepatitis, milk spot liver,

peritonitis, pneumonia and other pathology was higher in free range fattened pigs. These

differences were on the whole small except for milk spot liver in which a large difference was

found. There was a mean of 21.1% of pigs affected in free range fattened pigs compared

with 5.9% in indoor fattened pigs.


Table A1:5: Pathological conditions found during post-mortem inspection of offal from slaughter pigs from indoor and free range fattening systems (*statistically significant* difference between systems)




which the condition was present

(95% C.I.)


Abscess 23.7

(22.6-24.8)

24.2

(22.1-26.3)

2.5

(2.3-2.7)

3.3

(2.6-4.0)

Enteritis 24.3

(23.2-25.4)

24.3

(22.3-26.4)

1.7

(1.5-1.8)

1.8

(1.7-2.0)

Endocarditis 5.2

(4.6-5.8)

3.5

(2.6-4.4)

1.0

(0.9-1.1)

0.8

(0.6-0.9)

Hepatitis 35.8

(34.6-37.1)

34.1

(31.8-36.3)

*2.2

*

(2.1-2.4)

*3.1

*

(2.5-3.6)

Kidney Pathology 95.2

(94.7-95.8)

93.8

(92.6-94.9)

*7.2

(7.1-7.4)

*6.7

*

(6.4-6.9)

Milk Spot *23.4

*

(22.3-24.5)

*54.8

*

(52.4-57.2)

*5.9

*

(5.2-6.6)

*21.1

*

(19.5-22.7)

Pericarditis *60.6

*

(59.4-61.9)

*53.5

*

(51.1-55.9)

2.0

(1.9-2.1)

2.0

(1.9-2.2)

Peritonitis 57.7

(56.4-58.9)

58.6

(56.3-61.0)

*1.8

*

(1.7-1.8)

*2.4

*

(2.1-2.6)

Pleurisy 39.9

(38.6-41.2)

38.4

(36.0-40.7)

2.3

(2.2-2.4)

2.6

(2.4-2.8)

Pneumonia 86.6

(85.8-87.5)

88.9

(87.4-90.5)

*8.6

*

(8.4-8.9)

*10.8

*

(10.2-11.4)

Other Pathology 78.2

(77.1-79.3)

76.2

(74.2-78.3)

*4.9

*

(4.8-5.1)

*6.0

*

(5.6-6.4)



DISCUSSION

The data obtained were both extensive, covering all pigs slaughtered over a two year period

from the slaughterhouse, and of high quality with little data being censored; over 99.9% of

the data were used in the analysis.

The possibility exists that conditions detected could have been misclassified or detected with

a different degree of accuracy depending on who was performing the inspection. This is

unlikely to have introduced any significant bias into the analysis as this would likely to have

been non-differential, so equal in the two fattening systems compared.

The classification of the batches of pigs to the type of fattening system was likely to be

accurate as the slaughterhouse had good records of the provenance of pigs; this was in their

commercial interests, as free range pigs command a premium.

The study only involved the inspection data from one abattoir, which is likely to have only

obtained pigs from a discrete geographical region. Potentially this means that the results of

the analysis cannot be generalised to pigs slaughtered in other regions, although most of the

outdoor fattened pigs are produced in the east of England.

The comparison of the different fattening systems revealed that the prevalence of conditions

found at inspection were quite similar between the two. The finding of a higher frequency of

batches with tail bite and lameness at ante-mortem inspection in the indoor fattening

systems was not surprising. Tail biting in pigs is associated with poorer environmental

enrichment and lameness can be associated with poor flooring, so both of these conditions

are associated with housing. When it occurred, however the within batch prevalence was

not statistically significantly different between the two management systems.

Of interest was the very much higher proportion of batches of pigs with oedema in indoor

fattening systems. Oedema can have many causes including liver pathology, cardiac

pathology and kidney pathology none of these were excessively elevated in the comparisons

to explain this large difference. Again, the numbers of pigs affected within batches where

the condition was found were similar between the two fattening systems.

Milk spot liver occurred more frequently in free range fattened pigs, with it being detected in

more than twice the number of batches of free range pigs compared with indoor fattened

ones. In addition, more than a fifth of livers from pigs in which the condition was found were

rejected from free range fattened pigs. Again, this is not unexpected as milk spot liver is

caused by the migration of the larvae of the intestine worm Ascaris suum through the liver.

This migration causes inflammation and haemorrhage which is subsequently repaired by

fibrous material leading to white areas in the liver, the milk spots. The eggs of this worm are

very resistant and so may persist for very many years in the environment, and so are more

likely in pigs reared and fattened outdoors.

Pericarditis was found to occur in more batches of indoor pigs, although this difference is

unlikely to be biologically significant.


CONCLUSION

In conclusion, the prevalence of conditions detected on inspection of pigs submitted to

slaughter from different fattening systems, are, on the whole, quite similar. Most of the

differences can be predicted from the housing and fattening systems used and knowledge

about the influence this has on the occurrence of diseases. This similarity would suggest

that whatever inspection system was in place, it would be likely be equally effective for pigs

from both types of management systems.

FS145003 Final report ANNEX 2 v6 Page 1 of 15

ANNEX 2


ANALYSIS OF CONDITIONS FOUND AT POST-MORTEM INSPECTION FOR

OBJECTIVES 1, 2, AND 3

The aim of these studies was to establish a baseline of the frequency and types of

pathological conditions that can be expected to be identified when using visual-only

inspection in fattening pigs from non-controlled housing conditions; to do the same when

using traditional inspection; and, to compare the baseline values of the frequency and type

of conditions found using the two inspection methods.

MATERIAL AND METHODS

Sample size

The sample size for the study was estimated, in R http://www.r_project.org, a priori as

follows:

It is estimated that approximately 120 commercial herds supply abattoirs with pigs finished

outdoors in the United Kingdom (UK), with an average of 15 batches a year and

approximately 220 pigs per batch (Quality Meat Scotland, personnel communication).

Virtually all outdoor pig finishing units are located in England. The distribution of the

frequency of occurrence of various pathologies is unlikely to be uniform across farms or

batches. For low prevalence conditions it has been assumed that the minimum farm level

prevalence is 25% with at least 0.1% of batches affected. All pigs from each batch included

in the trial would be inspected; therefore within batch sample size calculations are irrelevant.

To allow an estimate of overall pig prevalence (for each sampling period) with a precision of

2% to be made with 95% certainty, assuming the worst case scenario (i.e. a true prevalence

of 50%), then a total of 10 different farms would need to be sampled, with an aim to include

five batches per farm, so that 50 batches in total will be inspected for the study. This would

lead to a total of 11,000 pigs inspected.

Recording of data

During traditional inspection Meat Hygiene Inspectors (MHIs) recorded all conditions found

on a touch screen system. This information was then transferred to the plant records from

where it can be printed out for each batch (identified by the individual slap mark). Data were

collected from the FSA records and transferred, by double entry, to a Microsoft Office Excel

Worksheet by project personnel.

The data for the conditions found at the visual-only inspection point were recorded by MHIs

on a recording sheet that mirrored the FSA records. Identification was again at the batch

level i.e. by the individual slap mark. These data were also double entered on a Microsoft

Office Excel Worksheet by project personnel.

All data collected during the trial were stored in the Microsoft Office Excel Worksheet. Each

batch was identified by a unique code (id), date of slaughter, week of slaughter,

http://www.r_project.org/


slaughterhouse, slap mark, number of pigs slaughtered, number of pigs inspected and all

conditions identified by MHIs by both methods (traditional and visual-only). The number of

pigs inspected was equal to the number slaughtered unless logistical issues occurred during

the recording. For example, if there was no recording available for traditional inspection then

no visual-only inspection was recorded until the system was resolved. This was to ensure

that the carcasses inspected by both inspection methods were the same.

The conditions were initially stored exactly as observed in the abattoir for recording purposes

during the trial. Conditions were then regrouped into categories ready for the statistical

analyses. These groups, or categories, were created following the same criteria used for the

historical analysis of frequency and types of conditions identified by post-mortem inspection

for fattening pigs from controlled and non controlled housing conditions (see Annex 1;

Objective 4).

Statistical analysis

All analyses were performed with R version 2.12.1 from R Foundation for Statistical

Computing. http://www.r_project.org

The data used for the purpose of this study consist of paired observations, from the same

sample population. For all carcasses from the same batch (a group of pigs coming from the

same farm) both traditional and visual inspection methods were applied to the same

carcasses. The frequency of a condition was recorded as the number seen per batch. The

number of pigs in each batch was also provided. The statistical methods used account for

this.

There was only one study abattoir so it was not possible to test if the abattoir influences the

results. The trial took place in five separate weeks, between the end of November 2011 and

the middle of March 2012. This could be considered as the same season; nevertheless the

effect of season was investigated at the level of week.

The number of animals affected, and the frequency (%), or prevalence, of each condition

found by each inspection method was estimated, and its 95% confidence intervals (C.I.)

were calculated. The C.I was calculated for the within batch prevalence using a t-test

approximation to the normal distribution. The distributions of the counts for each condition

were plotted to see if they follow a normal distribution. This was done separately for the two

methods of inspection. The proportion of each condition in each batch for the two inspection

methods was plotted to see if there was an association between the two methods. The

points were coloured by slap mark to visualise if slap mark (farm of origin) might affect the

results. Bland Altman plots were used to analyse the agreement between the two inspection

methods (the Bland Altman plot or difference plot is a method of data plotting used to

analyse the agreement between two different assays/methods).

For those conditions that followed a normal distribution a paired t-test was applied to see if

there were differences between the two proportions and a Pearson‟s correlation test to see if

there was a correlation between the two proportions. The season effect was investigated

using linear models.

The conditions that did not follow a normal distribution were categorised as „absence‟ or

„presence‟ of the condition for each batch. The data were analysed as a binary variable. The



McNemar test was used to see if there were differences between the two methods.

Sensitivity and specificity were estimated for the visual-only method using the traditional

inspection as the standard because the traditional inspection method is recognised and

currently applied in the UK for post-mortem inspection of pigs in abattoirs. The Kappa test

was also estimated to analyse the agreement between the two methods.

Due to the number of analyses performed in the comparisons (n=20) any differences were

considered statistical significant when P<0.00256, using Bonferroni correction.

For the purposes of this report it is, therefore, considered that any estimate is statistically significant if it has a p value < 0.00256.

RESULTS

Farms, batches and carcasses included in the trial:

A total of 62 batches and 11, 086 carcasses were inspected for the purpose of this study.

Twelve different farms were included. Different numbers of batches have been inspected

from each farm during the trial (Table A2:1).

Table A2:1: Summary statistics for the numbers of batches inspected per farm over the study period

Number of batches

inspected per farm

Min Mode Median Mean Max

1

4

4

5.2

19

Descriptive analysis of conditions

Original conditions from the trial are described in the “FSA coding of conditions” column

(Table A2:2). There were 179 different types of conditions found by traditional inspection

recorded and 179 conditions found by visual-only inspection recorded. They were the same

179 conditions (e.g. “bile contamination in trotter found by visual-only inspection” and “bile

contamination in trotter found by traditional inspection”). These conditions were regrouped

into 20 categories, or groups, of conditions. This differs from the preliminary report where 19

groups of conditions were identified. „Pathology in the pluck‟ was previously included in

„other‟ conditions. However, on further examination of the data it has been considered

separately due to the prevalence and the differences in prevalence between traditional and

visual inspection (1.6% and 4.7%) of the condition. The descriptive analysis of the data is

provided below (Table A2:2 and Figure A2:1).


Figure A2:1: Preliminary descriptive analysis of the frequency (%) of the conditions found in pigs from non-controlled housing conditions at post-mortem meat inspection by the two inspection methods.

Table A2:2: Descriptive analysis of the frequency (number [N]; percentage [%] and 95% confidence intervals [C.I.]) of the conditions found in pigs from non-controlled housing conditions at post-mortem meat inspection by the two inspection methods

Conditions in whole carcasses and offal

FSA coding of conditions Visual-only inspection Traditional inspection

N %

95% C.I.

N %

95% C.I.

Generalised Conditions

Suspect pyaemia, Suspect

fever/septicaemia,

Oedema/Emaciation, Anaemia

12 0.1

0.02 – 0.20

55 0.5

0.16 – 0.83

Abscess Abscess in any part of carcase or offal 172 1.6

1.14 – 1.96

235 2.1

1.56 – 2.68

Hepatic Pathology Hepatopathy, Hepatitis/Cirrhosis,

Jaundice

27 0.2

0.92 – 0.40

88 0.8

0.39 – 1.20

Milk Spot Milk Spot 263 2.4

1.33 – 3.41

601 5.4

3.52 – 7.32

Renal Pathology Kidney lesions, kidney pathology,

suspect uraemia

448 4.0

3.28 – 4.80

798 7.2

5.87 – 8.52

Endocarditis Endocarditis 0 0 21 0.2

0.05 – 0.33

Pericarditis Pericarditis 195 1.8

1.30 – 2.22

184 1.7

1.6 – 2.06

Pneumonia Pneumonia with abscess, pneumonia

without abscess

588 5.3

3.57 – 7.04

762 6.9

4.97 – 8.78

Pleurisy Pleurisy 494 4.5

3.30 – 5.61

369 3.3

2.38 – 4.27


Conditions in whole carcasses and offal


N %

95% C.I.

N %

95% C.I.

Enteritis Enteritis, colitis, Pathology in the guts 22 0.2

0.97 – 0.33

123 1.1

0.75 – 1.47

Peritonitis Peritonitis 231 2.1

1.38 – 2.79

221 2.0

1.35 – 2.63

Skin Pathology Skin conditions, dermatitis, erysipelas-

like lesions, bruising

107 1.0

0.65 – 1.28

108 1.0

0.75 – 1.20

Joint Pathology Arthritis 275 2.5

1.57 – 3.39

196 1.8

1.18 – 2.36

Pluck Pathology Pathology in the pluck 181 1.6

0.82 – 2.44

524 4.7

3.47 – 5.98

Other Conditions Other conditions 52 0.5

0.26 – 0.68

127 1.1

0.52 – 1.77

Bile Contamination Bile contamination in any part of

carcase/offal

516 4.7

4.00 – 5.30

574 5.2

4.39 – 5.97

Faecal Contamination Faecal contamination in any part of

carcase/offal

353 3.2

2.16 – 4.21

545 4.9

3.53 – 6.30

Grease Contamination Grease contamination in any part of

carcase/offal

58 0.5

0.30 – 0.75

49 0.4

0.16 – 0.73

Hair Contamination Hair contamination in any part of

carcase/offal

1362 12.3

9.83 – 14.75

419 3.8

1.20 – 6.36

Other Processing Faults Machine Damage, Over scald, Blood

splash, Other processing faults

31 0.3

0.05 – 0.51

23 0.2

0.04 – 0.37

Distributions of the counts for each condition

The distribution of the counts for each condition found by both inspection methods suggest

that generalised condition (Figure A2:2), hepatic pathology (Figure A2:3), endocarditis

(Figure A2:4) and other processing faults (Figure A2:5) do not seem to follow the normal

distribution and so they were categorised into presence or absence of the condition. It is only

possible to apply the paired t-test to compare the frequency of detection of conditions by

both inspection methods when the data follow the normal distribution (Gaussian curve). An

alternative test was used for the four conditions above that did not follow the normal

distribution. The rest of the conditions seem to fulfil the assumption of normal distribution.


Figure A2:2: Distribution of batch counts for the generalised condition for each inspection method

Figure A2:3: Distribution of batch counts for hepatic pathology for each inspection method


Figure A2:4: Distribution of batch counts for endocarditis for each inspection method

Figure A2:5: Distribution of batch counts for other processing faults for each inspection method

Plots of the proportions for each condition

From the plots of the proportion of each condition found by the two inspection methods for

each of the batches, the four conditions previously mentioned (generalised condition (Figure


A2:6), hepatic pathology (Figure A2:7), endocarditis (Figure A2:8) and other processing

faults (Figure A2:9)) again do not seem to show any association between the two inspection

methods. For the rest of the conditions it seems that the slap mark does not influence the

results i.e. no clusters are visually apparent.

Figure A2:6: Plot of the proportion of generalised conditions for the traditional versus visual inspection. The colours represent the slap mark.

Figure A2:7: Plot of the proportion of hepatic pathology for the traditional versus visual inspection. The colours represent the slap mark.


Figure A2:8: Plot of the proportion of endocarditis for the traditional versus visual inspection. The colours represent the slap mark.

Figure A2:9: Plot of the proportion of other processing faults for the traditional versus visual inspection. The colours represent the slap mark.

Paired t test and Pearson’s correlation test for conditions with an observed

normal distribution

There were statistically significant differences between traditional (considered the standard

method, as traditional inspection is accepted and applied for post-mortem inspection in

abattoirs in the UK) and visual inspection methods for the majority of the conditions analysed


(Table A2:3). There were no statistical differences for pericarditis, peritonitis, skin pathology,

joint pathology and grease contamination.

Table A2:3: Summary of the mean of the differences and its 95% confidence interval (C.I.) for the mean of the difference between the percentage of detection of conditions/contaminations between traditional (as the „gold‟ standard) and visual inspection methods – conditions with statistically significant p values shown in bold text

Condition Mean of the difference

(traditional – visual) %

95% C.I. for mean

difference %

P value

Lower

Level

Upper

Level

Abscess 0.67 0.09 1.25 0.02

Milk spots 3.21 1.51 4.91 <0.001

Renal Pathology 3.09 1.82 4.37 <0.001

Pericarditis 0.07 -0.53 0.67 0.82

Pneumonia 1.87 0.40 3.34 0.01

Pleurisy -1.12 -2.16 -0.08 0.04

Enteritis 0.99 0.61 1.37 <0.001

Peritonitis 0.06 -0.73 0.84 0.89

Skin pathology 0.05 -0.24 0.34 0.73

Joint pathology -0.84 -1.86 0.18 0.10

Pluck pathology 3.02 1.88 4.15 <0.001

Bile contamination 0.86 0.08 1.65 0.03

Faecal contamination 1.62 0.80 2.44 <0.001

Grease contamination -0.17 -0.52 0.17 0.31

Hair contamination -8.21 -10.79 -5.63 <0.001

Other conditions 0.75 0.07 1.43 0.03

The highest means of the differences between traditional and visual inspection, where more

affected carcasses were found by the traditional method than the visual method, are for the


following conditions: milk spots (3.21%, 95% C.I. 1.51 – 4.91), renal pathology (3.09%, 95%

C.I. 1.82 – 4.37), pluck pathology (3.02%, 95% C.I. 1.88 – 4.15), and faecal contamination

(1.62%, 95% C.I. 0.80 – 2.44). The highest mean difference however, is for hair

contamination (8.21%, 95% C.I. 5.63 – 10.79). For this condition more affected carcasses

were found with the visual inspection method than the traditional method.

For the majority of the conditions there is a statistically significant correlation between the

findings by the different inspection methods (Table A2:4). The strongest correlations are for

the faecal contamination (0.81, 95% C.I. 0.70 – 0.88) and pneumonia (0.71, 95% C.I. 0.56 –

0.82). If two variables (in our case both types of inspection methods) show strong correlation

then they are variables that tend to vary together, they change in similar way; i.e. when

traditional inspection detects a high number of one particular condition, the visual-only

inspection will also detect a high number of that condition.

Table A2:4: Correlation estimate, 95% confidence interval (C.I.) and p value for the condition analysed – conditions with statistically significant p values shown in bold text

Condition (traditional-visual) Correlation estimate (95%C.I.) P value

Abscess 0.57

(0.38 - 0.72)

<0.001

Milk spots 0.57

(0.38 - 0.72)

<0.001

Renal Pathology 0.46

(0.24 - 0.64)

<0.001

Pericarditis 0.16

(-0.09 - 0.39)

0.215

Pneumonia 0.71

(0.56 - 0.82)

<0.001

Pleurisy 0.54

(0.34 - 0.70)

<0.001

Enteritis 0.29

(0.05 - 0.51)

0.021

Peritonitis 0.32

(0.08 - 0.53)

0.010

Skin pathology 0.53

(0.32 - 0.69)

<0.001

Joint pathology 0.53

(0.32 - 0.69)

<0.001


Condition (traditional-visual) Correlation estimate (95%C.I.) P value

Bile contamination 0.48

(0.26 - 0.65)

<0.001

Faecal contamination 0.81

(0.70 - 0.88)

<0.001

Grease contamination 0.23

(-0.02 - 0.46)

0.069

Hair contamination 0.47

(0.25 - 0.64)

<0.001

Pluck pathology 0.59

(0.40 - 0.73)

<0.001

Other conditions 0.15

(-0.10 - 0.39)

0.244

McNemar test for conditions with an observed abnormal distribution

There was a statistically significant difference between the two inspection methods for two of

the four conditions where a non-normal distribution was observed (Table A2:5).

If the traditional inspection method is assumed to be the „gold‟ standard method, there is not

a good agreement between the two inspection methods for these conditions as the kappa

values were low for all of them. This is mainly because the sensitivity was low for all of them

(Table A2:6).

Table A2:5: MacNemar results for the categorised conditions

Condition P value

Generalised conditions <0.001

Hepatic pathology 0.045

Endocarditis 0.0015

Other processing faults 0.606


Table A2:6: Sensitivity (SE), specificity (SP), 95% confidence interval (C.I.), Youden‟s J statistic, Kappa value and the interpretation for the conditions analysed as binary variables.

Condition SE (95%C.I.) SP (95% C.I.) Youden’s J

statistic

Kappa

value

Interpretation

of kappa value

Generalised

conditions 26.1

(11.1 - 45.7)

100

(88.9 - 100)

0.2608 0.31 Low association

Hepatic

pathology 8.7

(1.5 - 29.5)

76.9

(60.3 - 88.3)

-0.1441 -0.16 Low association

Endocarditis 0

(0.0 - 30.1)

100

(91.1 - 100)

0 0 Low association

Other

processing

faults

40

(13.7 - 72.6)

82.7

(69.2 - 91.3)

0.227 0.20 Low association

No seasonal effect was found and the Bland Altman plots (not presented here) for the

conditions analysed suggest that linear regression does not apply to these data due to the

dispersion of the data.


DISCUSSION

The trial was carried out in one study abattoir. For logistical reasons a second abattoir in

south west of UK was not included. Conditions in pigs might be different depending on the

locations of the farms of origin. This may particularly be the case for abattoirs in the south

west of England, where bovine TB is present and outdoor pigs might be at a higher risk of

contact with TB than indoor pigs. However, most pig farms are located in the east of England

and due to the large number of outdoor pigs processed in the study abattoir we have

covered different areas of production in the UK. The study farms were located in the East of

England (in Norfolk and Suffolk) and in the South West (in Wiltshire). It will be unlikely that

we can generalise the finding to the whole of the UK unless we do further research into the

coverage of the study population achieved. As batches of carcasses included in the trial

have been identified (by the slap mark which identifies the farm of origin) and we know the

spatial location of the study farms, we could do some further studies of geographical

coverage of the study population and the need to include different areas in future studies, if

spatial location data for the denominator population were available. Industry experts have,

however, indicated that such data are “not easily found.”

The fact that the trial was carried out in only one abattoir might have biased the result by the

layout of the slaughter line. In the opinion of one of the authors, with extensive experience in

various abattoirs in the UK, the layout is not that different to the majority of the abattoirs in

UK, if not better. The effect of any such bias is, therefore, considered to be minimal.

The trial took place in five separate weeks, between the end of November 2011 and the

middle of March 2012. Season does not appear to have any effect on the difference

estimates. This is not unexpected as the weeks included in the trial were all during the winter

and a seasonal effect in the conditions was very unlikely to be present. If different seasons

were included in the trial some effect might potentially be expected; however, no such effect

of seasonality was found in the historical analysis for Objective 4 (Annex 1). The main

reason for the selection of winter season for the set up of the trial was the fact that the

presence of certain conditions, such as respiratory conditions (pneumonia, pleurisy), are

believed to be higher during the winter months. The trial was not designed to compare the

seasonality of conditions, but to study the effect of both post-mortem inspections when those

conditions peak.

When we compared visual-only inspection with traditional inspection it can be seen that

there are only slight differences between the frequencies with which each condition was

found. It is possible and not unexpected that conditions may be missed by visual-only

inspection. Although these differences are statistical significant they are less than 3.5% for

the majority of the conditions. This implies that at batch level there is not a biologically

significant difference between visual-only and traditional inspection, as the actual number of

carcases with these conditions that were missed (milk spot, renal pathology, and pluck

pathology) was very low. The detection of these conditions by visual-only inspection might

be improved by the alteration of the current layout of the slaughter line to maximise visual

access to all parts of carcases and offal (Objective 5; Annex 5). The majority of conditions

where traditional inspection exceeded visual-only inspection are conditions in offal; access to

the whole offal was substantially reduced by the visual-only inspection method. If access is

improved to all parts, then it will increase the ability to find these conditions. The same could


be said for conditions located in the back of the carcase and any contamination affecting the

rear of carcase that is not accessible for visual-only inspection. However, endocarditis will

only be detected by incision of the heart.

For those conditions with a statistically significant correlation between the findings by the

different inspection methods, such as faecal contamination and pneumonia, with an increase

in the prevalence of these conditions the difference between visual and traditional inspection

will remain the same.

For the purpose of this study we have assumed that the traditional inspection method is the

gold standard method (as it is currently accepted and applied in all abattoirs in UK for the

purpose of hygiene inspection). Ideally, a detailed post-inspection examination of carcases

and offal would be undertaken to establish the validity of this assumption.

CONCLUSION

We have found statistically significant differences between the number of carcasses found to

be affected by several conditions when examined by traditional and visual-only inspection.

For the majority of these conditions, traditional inspection finds more cases. Nevertheless,

the difference estimates are low, mostly under 3% and the biological significance is minimal

to negligible. Differences in conditions detected by both methods are very small and the

impact of those differences on the public health is expected to be negligible to very low

(Annex 4 Objective 8).

Some of the differences might be resolved with alterations to the abattoir line to allow a

better and full visual access to all parts of carcase and offal during the post-mortem

inspection. For further discussion and recommendations about implementation see Annex 5

Objective 5).


ANNEX 3


METHODS AND ANALYSIS FOR MICROBIOLOGICAL INVESTIGATIONS -

OBJECTIVES 6 & 7

The aims of this study are:

to establish the baseline of microbial cross contamination (total aerobic plate

count, Enterobacteriaceae count, and Salmonella isolation) after visual-only post-

mortem inspection and after traditional post-mortem inspection of fattening pigs

from non-controlled housing conditions;

to identify any difference in the risk of cross contamination in carcases visually

inspected compared to those traditionally inspected through total aerobic plate

count, Enterobacteriaceae count, and Salmonella spp. isolation;

to establish the baseline of microbial cross contamination with Yersinia spp. after visual-only post-mortem and after traditional post-mortem inspection of fattening pigs from non-controlled housing conditions.

to identify any difference in the risk of cross contamination in carcases visually inspected compared to those traditionally inspected through Yersinia spp isolation.

MATERIALS AND METHODS

Sample size

The sample size for the study was estimated a priori for each of the different elements to

be investigated. Sample sizes were calculated with R. http://www.r_project.org.

1. Sample size for detection of total aerobic plate count and Enterobacteriaceae count

Due to the variation of microbiological contamination across abattoirs (Zwefel et al.,

2005) we had to make some assumptions in order to calculate the appropriate sample

size for this study. We assumed the worst scenario where 50% of carcases have

evidence of cross-contamination with presence of Enterobacteriaceae and we aimed to

detect a difference of at least 10 percentage points in the proportion of carcases

presented with microbiological contamination between traditional and visual inspection,

with a power of 80% and a significance level of 95%. The estimated sample size was

412 swabs for each group (carcases traditionally inspected and carcases visually

inspected) with 824 swabs in total.



2. Sample size for isolation of Salmonella spp.

The baseline prevalence of Salmonella in carcases varies across abattoirs but it remains

low (0% to 9.64% (Unpublished data BPEX/FSA WP4); 13.5% (95% C.I. 9.9 – 18.1%)

from surface swabbing of carcases in the UK during the EU survey carried out in 2007

(Anonymous, 2008)). Assuming a baseline prevalence of 15% of the carcases presented

with Salmonella and still aiming to detect a difference of 10 percentage points, with a

power of 80% and a significance level of 95%, we need to process 160 swabs for each

group (carcases traditionally inspected and carcases visually inspected), 320 swabs in

total.

A total 824 samples would suffice to cover all microbial counts and isolation. The same

swabs will be used to test for the total aerobic plate count, Enterobacteriaceae, and

Salmonella.

3. Sample size for Yersinia spp.

Yersinia is not included in the compulsory microbiological checks in the EU and,

therefore there is uncertainty regarding Yersinia prevalence in carcases in the United

Kingdom (UK). It is believed that the prevalence of Yersinia spp., in carcases in the UK,

is as low if not lower than the prevalence of Salmonella (EFSA, 2012).

The sample size for this study was estimated based on the same assumption used for

Salmonella spp. Assuming a baseline prevalence of 15% of the carcases presented with

Salmonella and still aiming to detect a difference of 10 percentage points, with a power

of 80% and a significance level of 95%, we need to process 160 swabs for each group

(carcases traditionally inspected and carcases visually inspected), 320 swabs in total.

This is less than half of the total number of swabs to be collected in the trial.

Sample collection

Swabs were collected after traditional inspection and visual-only inspection of batches of

fattening pigs from non controlled housing conditions slaughtered in the study premises.

Swabs had to be collected and sent to the diagnostic laboratory (Epidemiology Research

Unit [ERU], Inverness) before 11 am every day during the trial, so batches included in

the study for collection of conditions found at traditional inspection and at visual-only

inspection (but expected to be slaughtered after 10 am that day were not included in the

sampling frame, to avoid any delay with the transport of samples to the laboratory.

Swabs were collected in five separate weeks, within the period 28th November 2011 to

16th March 2012. Swabs were collected every day from Monday to Friday during those

weeks when logistically possible. There were only three days of sampling during the first

week due to industrial action and issues with the courier.

The aim for the study was to collect approximately 200 swabs weekly, 40 swabs daily

(20 swabs after traditional inspection and 20 swabs after visual-only inspection) over

four weeks of the full trial to achieve the sample size required for the Enterobacteriaceae


count (n=800). One in every two swabs would then be selected with a random

systematic method, every day once the samples arrived at the laboratory to be used for

the isolation of Yersinia (n= 400).

In the pilot study week (week 1) 160 samples were collected and processed for Yersinia

spp. only, to study the capacity of the laboratory to process the samples for the rest of

the trial. These samples were not processed for total aerobic plate count,

Enterobacteriaceae count or Salmonella spp. isolation.

In the first two weeks of the full trial (weeks 2 & 3), samples were taken at the abattoir

after traditional inspection with a different pattern, due to logistical issues; however, this

is not expected to affect the analysis. One in every two swabs was selected on arrival at

the laboratory, as described above, to be used for the isolation of Yersinia (n= 199). All

samples were processed for total aerobic plate count, Enterobacteriaceae count and

Salmonella spp. isolation.

All the samples collected (n= 400) in weeks 4 & 5 were processed. All samples were

processed for total aerobic plate count, Enterobacteriaceae count, Salmonella spp. and

Yersinia spp. isolation.

Table A3:1: Distribution of swabs collected at the abattoir and swabs processed for Yersinia isolation during the trial

Samples collected Samples processed for Yersinia isolation

After visual inspection

After traditional inspection

After visual inspection

After traditional inspection

Week 1 80 80 80 80 Week 2 100 100 50 49 Week 3 100 100 50 50 Week 4 100 100 100 100 Week 5 100 100 100 100 Total 480 480 380 379

Sampling methodology

1. Swabs were taken according to the FSA protocol described at UKmeat.org

[http://www.ukmeat.org/RedMeatCarcasses.htm]

2. Swabs were taken immediately after visual-only inspection and immediately after

traditional inspection in the abattoir line (see Figure A3:1); one member of the

project team took samples after the visual-only inspection and one after the

traditional inspection. These were the same two project personnel for the

duration of the trial, but positions were changed every day to avoid any bias due

to different sampling technique.

http://www.ukmeat.org/RedMeatCarcasses.htm


Figure A3:1: A pictorial representation of the microbiological swabbing points on the abattoir line.

3. Swabs were taken from different sides of the carcases to avoid sampling the

same area of the same animal after visual-only inspection and after traditional

inspection. Swabs after visual-only inspection point were always taken on the left

side of the carcass. Swabs after traditional inspection point were always taken on

the right side of the carcass. The sides used at each collection point were not

changed during the trial for two reasons: firstly, it was logistically challenging to

attempt to swab on the right side after the visual-only inspection in the study

abattoir. Secondly, the potential bias due to a particular side being more

contaminated due to handling of carcases was assessed by the authors and the

conclusion was that nearly half of the plant staff and MHIs were observed to be

left-handed (therefore the handling of carcases were expected to be balanced

between both sides in the overall number of carcases, as staff and MHIs change

positions routinely), so no bias was expected to influence the estimates.

4. Each swab was immediately identified (labelled) with date, sampling point

(visual-only or traditional) and a unique reference number in order to identify at

what point of the inspection time the sample has been taken at (beginning,

middle or end of the line) This formed the basis for the categories of the ‘line

position’ variable.

5. Approximately 200 samples were collected per week (100 after visual-only

inspection and 100 after traditional inspection), if logistically possible.

6. Daily sampling was spread through the line to enable to collect samples at the

beginning, middle and end of the line.

7. Carcases were selected for swabbing with a systematic random sampling

strategy. In order to reach the number of samples required for each day and

ensure that samples were evenly spread through the line, carcases were

selected with a routine depending on the number of pigs slaughtered that day,


the number of samples required for that day and the line speed (e.g. if 400 pigs

are slaughtered, 20 samples (at each point) required, and 200 carcases per hour

line speed: 1 carcase every 6 minutes or 1 carcase every 20 were swabbed to

reach the numbers for the day).

8. Once all samples are collected and properly identified for the day, they were

placed in the box with frozen freezing blocks and sent to the ERU laboratory in

Inverness to arrive within 24 hours.

Recording of data

The expected number of batches of outdoor pigs and total number of pigs to be

slaughtered every day was requested from the Official Veterinarian or the plant manager

every morning before starting. The sampling strategy was determined, as above, to

obtain the required daily number of swabs. Recalculation of the sampling strategy was

done if any logistical issue occurred that jeopardised the completion of all samples

before 11am (deadline for transport of samples to the laboratory) once the slaughter

started, for example, a delay in the slaughtering or arrival of pigs. Any lot expected to be

slaughtered after 10am was excluded from the calculation.

Date of sampling, number of sample (VTP001-VTP160 for the pilot week and 1-800 for

the rest of the weeks) and point of collection (visual-only or traditional) were recorded

every day on datasheets. Number of lots and carcases included in the sampling strategy

was recorded in a datasheet.

Sample processing

Samples were processed in the ERU laboratory 24 hours after collection in the abattoir.

Counts and isolation were carried out by two of the project team (see Tables A3:2a & b).

Total aerobic plate count, Enterobacteriaceae count, and Salmonella isolation were

carried out using the following methods:

1. Total aerobic plate count was carried out according to the BS EN ISO 4833:2003

(British Standard Institution, 2003a)

2. Enterobacteriaceae count was carried out according to the BS EN ISO 21528-

2:2004 (British Standard Institution, 2004)

3. Salmonella: Isolation of Salmonella was carried out using the BS EN ISO

6579:2002 (British Standard Institution, 2002 and Health Protection Agency,

2011b)

Yersinia isolation was carried out using the following methods:

1. British Standards Institution (2003b) BS EN ISO 10273:2003 Microbiology of food

and animal feeding stuffs - Horizontal methods for the detection of presumptive

pathogenic Yersinia enterocolitica. London: BSI


2. Health Protection Agency (2011c). Detection of Yersinia species. Microbiology

Services. Food, Water & Environmental Microbiology Standard Method F33:

Issue 1.0.

Detail of laboratory technique

10mls of maximum recovery diluent (MRD, Oxoid Ltd, Thermo Fisher Scientific, UK) was

added to each swab bag and processed in a stomacher for 60 seconds. 10ml of liquid

was retrieved from each bag and placed in a sterile, labelled universal container (neat

sample). This became the neat sample used to inoculate enrichment broths for

Salmonella and Yersinia culture and for enumeration of bacteria. 1ml of the neat sample

was added to 9ml MRD (Oxoid Ltd) and mixed with vortex for 5 seconds. This process

was repeated for preparing the decimal dilutions needed for both enumeration

techniques.

Total aerobic plate count

17ml standard plate count agar (APHA, Oxoid Ltd) was added to 1ml of each dilution in

a sterile Petri dish, incubated at 30°C ±1°C for 72 (±3) hours. Those plates with 15-300

colonies present were counted and these results used to calculate the total count

present on the original swab, as follows:

Number of bacteria = Σc/ (n1 + 0.1n2) d where

Σc = the sum of colonies counted

n1 = the number of dishes retained in the first dilution

n2 = the number of dishes retained in the second dilution

d = the dilution factor corresponding to the first dilution

The result obtained was corrected to two significant figures and multiplied by ten to

reflect the volume of diluent added to the swab.

Enterobacteriaceae counts

17ml violet red bile glucose agar (VRBGA, BO0197, Oxoid Ltd) was added to 1ml of

each dilution in a sterile Petri dish, the surface was overlaid with a further 10mls of

VRBGA. The completely set plates were incubated at 37°C (±1°C) for 24 (±2) hours.

Plates with <150 colonies were retained for counting. Confirmation of the identity as

Enterobacteriaceae was carried out on up to five colonies by performing an oxidase test

and a glucose fermentation test (Oxoid Ltd).

Count per ml of neat sample= (No of colonies confirmed/ No of colonies tested)

(Presumptive count/ Volume tested x dilution)

The result obtained was corrected to two significant figures and multiplied by ten to

reflect the volume of diluent added to the swab.


Salmonella culture

1ml of the neat sample was added to 10ml buffered peptone water (BPW, Oxoid Ltd)

and incubated at 37°C (±1°C) for 18 (±2) hours. After pre-enrichment selective

enrichment broths were inoculated, 1ml BPW was transferred to 10ml Muller Kauffmann

tetrathionate-novobiocin (MKTTn, Oxoid Ltd ) broth and incubated at 37°C (±1°C) for 24

(±1) hours and 0.1ml BPW was transferred to 10mls Rappaport Vassiliadis Broth with

soya (RVS, Oxoid Ltd) and incubated at 41.5°C (±1°C) for 24 (±1) hours.

Following incubation, the selective enrichment broths were subcultured to both brilliant

green agar (BGA, Oxoid Ltd) and xylose lysine deoxycholate agar (XLD, Oxoid Ltd)

plates using a 10µl loop, streaking for single colonies. Following incubation at 37°C

(±1°C) for 24 (±3) hours, the plates were examined for typical colonies.

Up to five suspect colonies representative of each broth and agar combination were,

when present, subcultured to blood (BA, Oxoid Ltd) and MacConkey (Mac8, Oxoid Ltd)

agars to produce pure growth of well isolated colonies, following incubation at 37°C

(±1°C) for 24 (±3) hours representative colonies from each subculture were tested for

autoagglutination and also agglutination with polyvalent O and H antigens (Oxoid Ltd).

Biochemical confirmation was carried out on colonies that produced any sign of

agglutination (n=1) or were phenotypically similar to Salmonella but failed to agglutinate.

Triple sugar iron agar slants (TSI, Lab M, Bury, UK) were inoculated along with a urease

broth (BioConnections, Wetherby, UK). Both were incubated at 37°C (±1°C) for 21 (±3)

hours. Confirmation of identity was continued using API 20E (bioMerieux UK Ltd)

screening kit, following the manufacturer’s instructions and BA and Mac purity plates

were set up.

During the sampling at the abattoir, we swabbed approximately 1 metre (100cm) of the

pig carcase with the length of the swab sponge (10cm). That gives us a total surface

swabbed of 10cm x 100cm= 1,000 cm2

(http://www.ukmeat.org/RedMeatCarcasses.htm). We divided the total counts for aerobic

plate and Enterobacteriaceae by 1,000 cm2 to obtain the number of bacteria per cm2.



Table A3:2a: Summary of microbiological procedures in the laboratory for total aerobic plate count, Enterobacteriaceae count, and Salmonella isolation

Day 1

Day 2

Day 3

Day 4

Day 5

Day 6

Reference

Salmonella

Set up enrichment

Selective

enrichment

Selective isolation

media

Slide agglutinations,

purity plates

Set up

identifications

Identification as

probable Salmonella

BS EN ISO 6579:2002

Total aerobic plate counts

Set up serial dilutions and

duplicate counts

Counts

BS EN ISO 4833:2003

Enterobacteriacea

e counts

Set up serial dilutions and

duplicate counts

Counts

Purity plates

Set up

confirmations

Identification as

Enterobacteriaceae

BS EN ISO

21528-2:2004

Table A3:2b: Summary of microbiological procedures in the laboratory for Yersinia isolation

Day 1 Day 14 Day 15 Day 16 Day 17

Yersinia Set up cold enrichment

Selective isolation media

Purity plates

Set up identifications, motility tests

Identification as probable Yersinia

Roberts and Greenwood (2003). Practical Microbiology (3rd

Edition). Blackwell Publishing. Oxoid


Yersinia culture

Table A3:3: Enrichment broth and incubation conditions for Yersinia culture

Volume

of

sample

Enrichment

broth

Temperature Time Alkali treatment

(0.5% KOH in

0.5% saline)

Agar

0.1ml 9.9ml ITC 25°C (±1°C) 48 hours

(±2 hr)

No SSDC, CIN

1ml 9ml PSB 25°C (±1°C) 5 days static

incubation

Yes CIN

1ml 9ml TBW 9°C (±2°C) 14 days Yes CIN

The neat sample was added to enrichment broths and incubated as shown in Table A3:3.

After incubation, 10µl of the irgasan ticarcillin potassium chlorate broth (ITC broth base;

ticarcillin; potassium chlorate; Sigma-Aldrich Co Ltd) were subcultured onto both Cefsulodin

irgasan novobiocin agar (CIN, Oxoid Ltd) and Salmonella/Shigella agar with sodium

deoxycholate and calcium chloride agar (SSDC AGTC Bioproducts Ltd), both agars were

incubated 30°C (±1°C) for 21 (±3) hours, examined for typical colonies and reincubated at

the same temperature for a further 24 hours before re-examination.

10µl of the incubated peptone sorbitol broth (PSB, Sigma-Aldrich) broth was subcultured

directly onto CIN agar, and a further 0.5ml was mixed with 4.5ml potassium hydroxide

solution (KOH) and 10µl was immediately subcultured to CIN agar. The agar was incubated

30°C (±1°C) for 21 (±3) hours, examined for typical colonies and reincubated at 30°C for a

further 24 hours before re-examination.

The incubated Tris buffered peptone water (TBW, Oxoid Ltd, TRIS, Sigma-Aldrich) were also

alkali treated. One ml TBW was mixed with nine 9ml KOH, and immediately 10µl was

subcultured to CIN agar. The agar was incubated 30°C (±1°C) for 21 (±3) hours, examined

for typical colonies and reincubated at 30°C for a further 24 hours before re-examination.

Up to five suspect colonies from each agar/alkali treatment combination were subcultured to

blood (BA) and MacConkey (Mac) agars to produce pure growth of well isolated colonies.

Tests to determine the production of oxidase, urease and indole were set up along with a

Kligler’s agar slant. On Kligler’s agar Yersinia produce an acid butt with no gas or hydrogen

sulphide production and an unchanged slant. Some lactose positive strains have been

reported, none were identified in this study. Yersinia produce urease, may or may not

produce indole and do not produce oxidase.

Confirmation of identity was continued using API 20E screening kit, following the

manufacturer’s instructions and BA and Mac purity plates were set up.

It is noted that Yersinia plasmids may be spontaneously lost on culture above 30°C and

during prolonged passage. In this study, no culture occurred above 30°C and Yersinia was

stored at -80°C as soon as they were identified as such by API20E.


All swabs collected at the study abattoir during the trial the weeks 1, 4 and 5 were used to

isolate Yersinia. For the weeks 2 and 3, the isolation of Yersinia was restricted to one every

two samples with a systematic random sampling approach. A total of 759 samples were

processed at the end of the study for the purpose of isolation of Yersinia.

Data management

Results from the laboratory were recorded in a spreadsheet. Variables included in the

dataset were: abattoir, date of collection, individual id number (swab reference), inspection

method (collection point: after visual-only inspection or after traditional inspection), total

aerobic plate count, Enterobacteriaceae, Salmonella, day (4 to 23), line position and week.

The ‘Day’ variable is assigned from 1 to 23 for all days we collected swabs during the trial,

including the pilot week where swabs were collected (three days collection that week) for

Yersinia isolation only (not included in this report). This was an approximation to batch as

the batch was not recorded for each swab.

’Week’ was used as an approximation of season (i.e. the week where we collect swabs

during the trial.) Data from week 2 to 5 are included in this report as week 1 was used for

Yersinia isolation only.

‘Line position’: daily swabs were grouped in four groups, according to their position along the

slaughter line (from 1= swabs collected from carcases at the beginning of the line to 4=

swabs collected from carcases at the end of the line). We could expect the contamination to

increase in the carcases during the day, so carcases at the end of the day might be more

contaminated than carcases at the beginning of the day.

Outcome variables assessed in this study were total aerobic plate count, Enterobacteriaceae

plate count, Salmonella spp. isolation and Yersinia spp. isolation.

The variables were analysed to see if there was any difference between the two inspection

methods in terms of microbiological counts.

Statistical analysis

All analyses were performed with R version 2.12.1 from R Foundation for Statistical

Computing. http://www.r_project.org

The data used for this study were individual observations, sampled by systematic random

strategy from the same study population: carcases from fattening pigs from non-controlled

housing conditions. Samples were not intentionally taken from the same carcase after both

inspection methods (visual-only and traditional) and measures were in place to avoid

sampling the same area of the carcase if the same carcase was sampled at both inspection

points by chance. Only one study abattoir was used, so it was not possible to test if the

abattoir or the FSA team influences the results. The trial took place in five separate weeks

from the end of November 2011 to the middle of March 2012, which could be considered the

same season; nevertheless the effect of season was investigated at the level of week.

There was no recording of the batch (farm of origin of the pigs) from which the samples

were taken, Individual batches processed every day were recorded (with an average of two

batches used for sampling per day), so variable date is used as an approximation of farm of



origin, to study the possible effect of farm of origin on microbial contamination. Samples

were identified in four groups every day (from 1= at the beginning of the line to 4= at the end

of the line) to identify the location of carcases on time during the day. This was taken into

account in the analysis to study the effect of the position on the line on the results. Potential

bias of personnel collecting the samples was addressed during the sampling by rotating the

swab collection position daily. Potential bias of MHIs handling the carcases (different

methodologies of handling and inspection could increase or decrease the risk of

contaminating the carcase after traditional inspection) was considered and addressed as

MHIs were rotated every 20 minutes and different MHIs handled the carcases swabbed for

the purpose of this study.

Total aerobic plate count and Enterobacteriaceae count

Distributions of the counts were plotted in a histogram to see if they followed a normal

distribution. This was done separately for the two methods of inspection. Total aerobic plate

count and Enterobacteriaceae count variables were transformed into log10 total aerobic plate

count and log10 Enterobacteriaceae count to try to fulfil the normality assumption. The mean

of the microbial counts with their 95% confidence interval (C.I.) were calculated for both

inspection methods. For total aerobic plate count the estimates calculated were the mean of

the logs. For the Enterobacteriaceae count the estimates were the mean of the values, as

the log scale has a large number of zeros.

The student t-test for comparison of the mean for aerobic plate count was applied to analyse

the differences between the mean of the count after visual-only inspection and the mean of

the count after traditional inspection. For the Enterobacteriaceae count there were a large

number of zeros, so the variable was categorised as a binary variable: presence/absence.

For the categorised variables (presence/absence) a chi-squared test was used to see if

there was an association between these variables and the inspection method. The samples

with more than zero Enterobacteriaceae plate count were transformed to log10 and a student

t-test was used to compare the means between the inspection methods.

Linear models were also applied to investigate the association between the inspection

method and the microbial contamination of carcases. Other variables included in the models

were inspection method, week, date and line position, in order to study the effects of season,

farm of origin, and position of carcase in the line. The variables were selected to enter in

the multivariable model if P<0.15.

Yersinia spp. isolation

The outcome variable was categorized as a binary variable: presence/absence. For the

univariable analyses a logistic regression model (Wald test) was used to test if there was an

association between each of the variables individually and the presence of Yersinia. The

variables tested in the univariable analyses were: inspection method, week, date, and line

position. The aim is to study the effects of inspection method, season, farm of origin, and

position of carcase in the line.

Any variables with a p<0.15 in the univariable analysis were selected to be used in the

multivariable analysis. A logistic regression model (Wald test) was also used for this, to test

if there was an association between the selected variables and the presence of Yersinia.


All

Due to the number of analyses performed in the comparison (n=4) we adjusted the

significance level, using Bonferroni correction (n =4).

For the purpose of this study, therefore, we will consider statistical significant any estimate

with a p<0.0127 for all tests performed.

RESULTS

Total aerobic plate count, Enterobacteriaceae count and Salmonella spp.

isolation

Farms, batches, and carcases available for sampling

A total of 800 samples were collected during the trial for the purpose of this study. 400

samples were collected after visual-only inspection and 400 samples after traditional

inspection (see Table A3:4). A total of 44 batches from ten farms and 7,931 carcases were

included in the sampling frame. The number of carcases, batches and farms are lower than

the number used for conditions during the same four weeks, due to logistic requirements.

Samples had to be dispatched before 11:00 am every day (in order to arrive to laboratory

within 24 hours), so pigs from batches included in the condition study, but not expected to be

slaughtered before 10:00 am were not included in the selection for sampling. This is not

expected to significantly affect the results.

The number of total swabs collected for the purpose of the study was slightly lower than the

proposed sample size for total aerobic plate count and Enterobacteriaceae. This should not

affect the results, as the prevalence actually found in the carcases during the trial was lower

than the worst scenario assumption, therefore a lower number of swabs will suffice to reach

the same power and significance level (80% and 95% respectively).

TableA3:4: Summary statistics for the number of swabs collected per day over the subset of the study period applicable to total aerobic plate count and Enterobacteriaceae counts and Salmonella spp. isolation.

Number of swabs collected per day


Visual-only

inspection

20 20 20 20 20

Traditional

inspection

9 20 20 20 24

Total 29 40 40 40 44

Descriptive analysis of microbial contamination in carcases after both inspection

methods

A description of the mean of all outcomes analysed in the study is presented in Tables 3 and

4. For the total count plate count the estimates are presented as the log10 of the actual

results. For the Enterobacteriaceae count the large number of zeros made it impossible to

present the estimates as their log values; the actual estimates are presented instead.


Table A3:5: Descriptive analysis of total aerobic plate count on carcases during the trial after both inspection methods.

Mean of the log10

of total aerobic plate count

Mean Standard deviation

95% C.I.

After traditional inspection 1.497 1.03 1.396 – 1.598 After visual-inspection inspection 1.421 0.95 1.327 – 1.514

Table A3:6: Descriptive analysis of Enterobacteriaceae count on carcases during the trial after both inspection methods.

Mean of Enterobacteriaceae count

Mean Standard deviation

95% C.I.

After traditional inspection 0.359 3.19 0.046 – 0.673 After visual-inspection inspection 0.060 0.64 -0.003 – 0.124

Salmonella counts were zero for all the samples. No Salmonella spp. was isolated from any

of the swabs collected for the purpose of the study. No analysis has, therefore, been carried

to identify the difference between the risks of cross contamination with Salmonella in

carcases from fattening pigs from non-controlled housing conditions in the study abattoir.

Distribution of the microbial counts in carcases after both inspection methods

Distribution of the count of each microbial contamination of carcases, total aerobic plate

count and Enterobacteriaceae count, were plotted to see if they follow the normal

distribution.

The plots (Figure A3:2 and A3:4) of the variable total aerobic plate count and

Enterobacteriaceae count suggested that a log transformation should be applied to both of

the data sets (see Figures A3:3 and A3:5).


Figure A3:2: Histogram of the total aerobic plate count

Figure A3:3: Histogram of the log transformation of the aerobic plate count


Figure A3:4: Histogram of Enterobacteriaceae count

Figure A3:5: Histogram of the log of Enterobacteriaceae count


Microbial contamination of carcases (total aerobic plate count and Enterobacteriaceae

count) was plotted separately by inspection methods to see if there was any suggestion of a

difference between them. No difference was apparent in the total aerobic plate counts

(Figure A3:6).

Figure A3:6: Distribution of log of total aerobic plate count for each inspection method

Due to the large number of zeros (565 samples) in the Enterobacteriaceae count, we

categorised this variable into presence/absence and plotted the counts just for the group

where it was present (i.e. Enterobacteriaceae count > 0). A slight difference is apparent in

the Enterobacteriaceae count between both inspection methods; the microbial contamination

with Enterobacteriaceae being lower after visual-only contamination (Figure A3:7).


Figure A3:7: Distribution of log of Enterobacteriaceae count for each inspection method

Univariable analyses

There is no statistical difference in the total aerobic plate counts between the inspection

methods, using the student t-test to compare the two means (Table A3:7).

Table A3:7: Results of student t-test for the log of total aerobic plate count

Inspection method Mean Difference between the

mean (95%C.I.)

P value

Traditional 1.497 0.08 (-0.06, 0.21) 0.2772

Visual 1.421

The variable Enterobacteriaceae count (Figure A3:4) had a high number of samples with

zero counts. After categorisation of this variable as absence/presence of Enterobacteriaceae

(Table A3:8) and using a chi-squared test to test if there was an association between the

inspection method and the Enterobacteriaceae detection, there was no association between

the categorized Enterobacteriaceae variable and the inspection method (p=0.76).

Table A3:8: Enterobacteriaceae presence/absence by inspection method

Inspection method

Traditional Visual

Enterobacteriaceae

count

Absence 280 285

Presence 120 115

When the subset of counts for Enterobacteriaceae contamination when present (i.e. >0,

n=235) were compared, there is a statistical significance difference in the


Enterobacteriaceae contamination of carcases between inspection methods.

Enterobacteriaceae contamination is lower after visual-only inspection compared to

Enterobacteriaceae contamination after traditional inspection (Table A3:9).

Table A3:9: Student t-test for the comparison between the two inspection methods for Enterobacteriaceae counts where Enterobacteriaceae = presence.


mean (95%C.I.)

P value

Traditional -1.14 0.43 (0.22, 0.63) <0.001

Visual -1.57

Multivariable analyses

The results from the use of a linear model to test if the other variables (week, date and line

position) influence the total aerobic plate count indicate that the variable date influences the

outcome (results not shown). This seems to imply that the batch/farm or origin is the only

variable that influences the results in terms of aerobic plate count.

We also used linear models to test the category ‘presence of Enterobacteriaceae (n=235

samples)’, with the Enterobacteriaceae count as outcome and the inspection methods,

weeks, date and the line position as fixed factors. The only statistical significant variable was

the inspection method, implying that the inspection method is the only variable that

influences the results in terms of Enterobacteriaceae count. For this type of variable (more

than half of the samples had zero counts) a zero-inflated binomial model fits the data better.

Yersinia spp. isolation

Farms, batches, and carcases available for sampling

A total of 960 samples (see Table A3:10) were collected during the trial for the purpose of

this study: 160 in the trial week (week 1) and 800 in weeks 2 to 5, of which 480 samples

were collected after visual-only inspection and 480 samples after traditional inspection (see

Table 1). These swabs came from a sampling frame of 54 batches from 12 farms consisting

of 9,633 carcases. Only approximately half of the collected swabs were processed for the

isolation of Yersinia during the weeks 2 and 3. This brings the total number of swabs

processed down to 759 (380 after visual-only inspection and 379 after traditional inspection).

The number of total swabs collected for the purpose of the study was larger than the

proposed sample size for Yersinia (320: 160 after traditional inspection and 160 after visual-

only inspection). This should be more than sufficient to reach the same power and

significance level (80% and 95% respectively).


Table A3:10: Summary statistics for the number of swabs collected per day over the full study period.

Number of swabs collected per day


Visual-only

inspection

13 20 20 21 40

Traditional

inspection

9 20 20 21 40

Total 22 40 40 42 80

Descriptive analysis of microbial contamination with Yersinia in carcases after both

inspection methods

Table A3:11: Distribution of Yersinia isolated from carcases swabs after both inspection methods during the trial.

Yersinia

enterocolitica

n (percentage*)

Yersinia

pseudotuberculosis

n (percentage*)

Yersinia

fredericksenii

n (percentage*)

Traditional Visual-only Traditional Visual-only Traditional Visual-

only

Week 1 5 (6.25%) 2 (2.5%) 1 (1.25%) 1 (1.25%)

Week 2

Week 3 1 (2%)

Week 4 1 (1%)

Week 5 1 (1%)

Total 6 (1.6%**) 4 (1.1%**) 1 (0.3%**) 1 (0.3%**)

*% of samples collected that week

** % of the total samples collected during the trial

During the pilot study (first week in November 2011) 160 carcase swabs were tested for

Yersinia; seven Yersinia enterocolitica, one Yersinia pseudotuberculosis, and one Yersinia

fredericksenii were isolated from the carcase swabs received in that week.

For the rest of the trial (four weeks between 16th January 2012 and 16th March 2012) 599

carcase swabs were tested for Yersinia; a total of three Yersinia enterocolitica were isolated

from those carcase swabs.

The distribution of Yersinia isolated from the samples after both inspections during the trial is

shown in Table A3:11. During the first week of the trial, nine Yersinia were isolated from the


carcase swabs: seven Yersinia enterocolitica, five (6.25% of samples collected that week)

after traditional inspection and two (2.5% of samples collected that week) after visual-only

inspection; one Yersinia pseudotuberculosis (1.25% of samples collected on that week) after

visual-only inspection; and one Yersinia fredericksenii (1.25% of samples collected on that

week) after traditional inspection. No Yersinia of any type was isolated in the second week of

the trial. During the third week, one Yersinia enterolitica (2% of samples collected on that

week) was isolated after traditional inspection. During the fourth week, one Yersinia

enterolitica (1% of samples collected on that week) was isolated after visual-only inspection.

During the fifth week, one Yersinia enterolitica (1% of samples collected on that week) was

isolated after visual-only inspection.

In total, twelve (1.6% of the total samples collected during the trial) Yersinia were

isolated from the swabs: seven (0.9% of the total samples collected after during the trial)

after traditional inspection, six of them Yersinia enterolitica and one Yersinia

fredericksenii; five (0.6% of the total samples collected after during the trial) after visual-

only inspection, four of them Yersinia enterocolitica and one Yersinia

pseudotuberculosis.

Univariable analyses

For Yersinia detection, the outcome variable was categorized as presence or absence,

regardless of the type of Yersinia. From now on, we consider a sample being positive for

Yersinia any sample positive for Yersinia enterolitica, Yersinia pseudotuberculosis or

Yersinia fredericksenii.

Table A3:12: Detection of Yersina for the inspection methods

Inspection method

Traditional Visual

Yersinia isolation Absence 372 375

Presence 7 5

The majority of the samples were negative (Table A3:12).

There is no statistical difference (p=0.55) in the Yersinia detection in carcases between the

inspection methods (Table A3:12).

Carcases sampled at the end of the line had more risk of having Yersinia than at the

beginning of the line (p=0.09).

Carcases sampled in week 1 had more risk of presence of Yersinia than in other weeks

(p<0.01).

Carcases sampled at two dates (28th and 29th November 2011) had more risk of presence of

Yersinia than in the rest of the dates (p<0.03).


Multivariable analyses

The variables at the univariable analyses selected to enter the multivariable model were

week, line position and date.

In the final multivariable model there was a statistically significant association with the

presence of Yersinia for the variable date, specifically the 29th of November 2011 (p<0.001).

Two batches (i.e. two different farms) were processed on that day; one of the farm provided

pigs for the rest of the trial and the other one only sent animals on that day. Otherwise, there

was no statistical difference (p=0.55) in the Yersinia detection in carcases between the

inspection methods.


DISCUSSION

The EU Regulation No 1441/2007 (Anonymous, 2007) imposes microbiological criteria for

foodstuffs. Chapter 2 of this Regulation defines the hygiene status of the process by the

level of microbial contamination. If we apply these microbiological criteria to the results of the

samples in the study, then as both the mean of the Enterobacteriaceae count and the mean

of the total aerobic plate count were below the lower limit it was considered to be

‘satisfactory’ (Table A3:13). This abattoir is considered a satisfactory abattoir as far as

hygiene is concerned when measured by microbiological checks of carcases after

inspection.

Table A3:13: Swabbing criteria for pig carcasses [http://www.ukmeat.org/RedMeatCarcasses.htm]

Total aerobic plate Count

Enterobacteriaceae count

Salmonella

Unacceptable mean log /number of positives is equal to or above

4.3 2.3 5/50

Acceptable mean log is below 4.3 2.3

Satisfactory mean log / number of positives is below

3.3 1.3 5/50

The same level of total aerobic plate (TAP) counts was observed with each inspection

method; however, lower levels of Enterobacteriaceae counts were observed after visual-only

inspection. The trial was carried out in only one abattoir. The study abattoir was particularly

clean, with a very good protocol of hygienic measures implemented by both the plant

personnel and the FSA personnel. Plant personnel and FSA personnel had a very good

routine of washing and disinfecting hands and knives on the line. This was the norm and was

not just due to the occurrence of the trial. This emphasis on good hygiene practice would

result in a lower level of any cross contamination of any carcases. The microbiological

outcomes of the study (TAP and Enterobacteriaceae counts) were not a surprise for the

authors, when they compare the study abattoir to other abattoirs that they have visited in the

past in the UK. If, as the results of the study demonstrate, a reduction of contamination with

Enterobacteriaceae can be achieved by the use of visual-only inspection in an abattoir with a

very low prevalence of contamination in their carcases, it would also be expected to be

observed in abattoirs with a higher initial prevalence of contamination i.e. visual-only

inspection results in a lower level of any cross contamination of carcases.

By using only one study abattoir, we have reduced the study to only one FSA team. As

explained above, the hygiene routine of this team is very good. A complete ‘hands-off’

visual-only system was followed by all MIHs carrying out the trial; this should identify the

largest apparent gain that can be made between the two inspection systems.

No Salmonella spp. was isolated from any of the 800 swabs processed during the trial.

According to the assumption made prior to the start of the trial, based on the prevalence of

Salmonella observed in carcases in UK, (up to 9.64% or 13.5% (depending on the reference

- Unpublished data BPEX/FSA WP4 and Anonymous, 2008, respectively) we expected to

isolate some Salmonella spp. The FSA defines the level of Salmonella in an abattoir as

unacceptable [[http://www.ukmeat.org/RedMeatCarcasses.htm] when Salmonella spp. is

detected in more than five out of 50 samples (10%). We collected 800 samples, so even if




we isolated Salmonella in up to 16 of the samples, the abattoir would have remained within

the acceptable range. Swabs were taken in the side of the carcase, as we tried to identify

any cross contamination. The study abattoir follows a particularly good hygiene practice and

that could contribute to the why we did not isolate any Salmonella. It would also be useful to

know more about the disease/infection status of the production units involved in the study

and how they compare to both the general non-controlled housing fattening pig production

unit population and the controlled housing fattening pig production unit population, with

respect to Salmonella disease/infection status i.e. are these particularly clean pigs coming

into an abattoir with good hygiene/cross-contamination control? Such data are currently not

available to us.

We would not expect to have experienced any difference in the results had we included all

outdoors batches received and processed at the abattoir. Only five batches arrived at the

abattoir after 10 am and were, therefore, excluded from the sampling frame. They were not

constantly from the same producers, so no bias is expected.

The trial has been carried out in only one study abattoir, in the East of England. For logistical

reasons a second abattoir in the south west of the UK was not included in the trial. The

prevalence of microbiological agents, such as Salmonella and Yersinia in pigs might be

different (Anonymous, 2008), depending on the locations of the farms of origin. This

hypothesis gains some support from the statistically significant difference in Yersinia

contamination presence that was observed, after adjusting for an approximation of farms of

origin (date variable) for one particular date: the 29th of November 2011. On that particular

day two different farms were recorded as delivering animals to the abattoir. One of the farms

sent animals on most of the days in the following weeks for the rest of the trial. The other

farm sent animals exclusively on that day. It is possible that this one farm was the source of

the Yersinia that was detected.

However, most of pig farms are located in the east of England and due to the large number

of outdoor pigs processed in the study abattoir we expect to have covered different areas of

‘outdoor’ pig production in the UK. To confirm this we would need (as mentioned in Annex 2)

either demographic and spatial location data to determine the geographical distribution of

‘outdoor’ pig production in the UK and to compare it with the production units included in the

trial, or (for a more crude approximation) an indication of what proportion of the total

processing of ‘outdoor’ pigs from non-controlled housing conditions is done at the study

abattoir. Such data are not readily available. The difference in Enterobacteriaceae

contamination between inspection methods was statistically significant after adjusting for an

approximation of farms of origin (date variable). Again, further research into the coverage of

the target population that was achieved would assist in the assessment of the validity of this

result and whether it can be extrapolated to the non-study population within the UK.


CONCLUSIONS

Carcases from pigs from non-controlled housing conditions in the study abattoir after visual-

only inspection have a similar level of contamination with Yersinia spp. than after traditional

inspection, i.e. the level of contamination after both inspections is the same. However, a

lower level of contamination with Enterobacteriaceae was found after visual inspection than

after traditional inspection. The total aerobic plate counts are similar, i.e. the level of

contamination after both inspections is the same; and no Salmonella spp. was isolated from

any carcases. These results suggest that we were able to reduce the microbial load i.e. the

contamination of carcases by changing the post-mortem inspection method to a visual

system where handling of carcases by FSA personnel was minimised.

Total aerobic plate count and Enterobacteriaceae count are commonly used as an

approximation of any microbiological contamination in carcases. So we can say that the risk

of any microbiological contamination in carcases is lower after visual-only post-mortem

inspection compared to traditional post-mortem inspection.

We can conclude that as this result has been observed in a particularly clean abattoir,

results are expected to be observed in any abattoir with level of contamination as low as and

higher than the study premises.


ANNEX 4


REPORT OF THE RISK ASSESSMENT FOR OBJECTIVE 8

This purpose of this work was to undertake a formal, mostly qualitative, risk assessment,

based on guidelines described by the Codex Alimentarius Commission (CAC), on the impact

of the implementation of visual-only inspection for fattening pigs from non-controlled housing

conditions and the impacts that this could have in terms of public health, animal health,

animal welfare and the allocation of resources

We have used a modified CAC risk assessment approach to qualitatively assess the

potential change in risks to human (public health via a food-borne route), animal health and

animal welfare of a change in the meat inspection method, from the traditional method

currently employed to a visual-only (‗hands-off‘) methodology, for fattening pigs from non-

controlled housing management systems i.e. raised outdoors from weaning to slaughter. We

have not considered the component of public health risk that is due to occupational

exposure. We have used data from previous work, scientific literature, publically available

information and our own field study to inform the risk assessment. Based on this information

we have taken a cautious approach and considered a worse case scenario.

Of the five public health hazards we have assessed (endocarditis, granulomatous lesions,

Salmonella spp., Yersinia spp., and the hygiene process indicators - total aerobic plate count

and Enterobacteriaceae count) only two have a revised risk on a change in inspection

method. The risk for hazards associated with endocarditis lesions changes from negligible to

non-negligible, i.e. very low, while from the results of the Enterobacteriaceae count analysis

it is possible that the risk of cross-contamination between carcases is reduced. Only two

animal health hazards were identified and assessed (endocarditis and granulomatous

lesions). Again, endocarditis has a revised risk on a change in inspection method from

negligible to non-negligible i.e. very low. Despite the revised risk classification for public and

animal health, attributable to the reduced detection of endocarditis lesions by visual-only

inspection compared to traditional inspection for outdoor pigs, the fact still remains that

outdoor pigs from non-controlled housing conditions present at least the same, if not less

(Hill et al., 2011), of a risk than indoor pigs from controlled housing conditions. Visual

inspection is acceptable for pigs from indoor, controlled and integrated management

systems (Anon., 2004b); therefore there is no reason relevant to the public health risk

presented to exclude outdoor pigs purely on grounds of the management system from which

they originate. This is also the case for the animal health risk. Action by producers is unlikely

to be taken on the basis of information received about endocarditis lesions from post-mortem

data feedback. Action would be taken in response to clinical signs in live pigs with

associated production losses and their economic impact for the causal agents associated

with endocarditis lesions (Strepococcus spp, including S. suis & E. rhusiopathiae).

One of the arguments for a move from a traditional palpation and incision inspection system

to a visual-only based one is that it could reduce cross contamination of carcases that would

occur via the hands and knives of meat inspectors. In our field study, no difference was

found in the isolation of Yersinia spp. or Salmonella spp., total aerobic plate count or the


presence/absence of Enterobacteriaceae; however, when present the Enterobacteriaceae

count was lower on carcases that had been visually inspected than traditionally inspected,

implying less contamination. The abattoir used for the field study had a particularly good

hygiene process. It is possible that a change in the inspection method from traditional to

visual would lead to a similar result in any abattoir with a level of contamination as low as or

higher than the study premises. If the level of contamination is lower, then it could be

hypothesised that the potential for cross-contamination would be lower; however, we cannot

draw that directly as a conclusion from our study.

The primary benefit of a visual-only system of inspection that encompassed pigs from non-

controlled housing conditions would be, in the United Kingdom (UK), the ability to implement

such an inspection system for all pigs. At present although it is theoretically possible to do so

for pigs from controlled housing conditions, in terms of the regulatory process, such systems

have not been implemented because slaughterhouses process fattening pigs from different

management systems.

From this risk assessment, based on current evidence, we do not consider that there is any

appreciable additional risk to public health, animal health or animal welfare from visual-only

inspection of fattening pigs from non-controlled housing conditions in the UK over and above

that which currently exists with traditional inspection.


TABLE OF CONTENTS

Background ........................................................................................................................... 7

The problem ....................................................................................................................... 7

Project FS145003 ............................................................................................................... 7

Risk Analysis ........................................................................................................................ 7

Objective 8.......................................................................................................................... 8

Risk Assessment .................................................................................................................. 9

Previous work ..................................................................................................................... 9

Define the question ........................................................................................................... 11

‘How would the risk profile change if fattening pigs from non-controlled housing conditions

are visually inspected, in terms of: ........................................................................................ 11

What is the baseline against which any change in the risk profile is to be considered? ...... 11

Definitions ........................................................................................................................................... 11

1. What is the relative risk posed by visual-only inspection of ‘outdoor’ pigs over

traditional inspection of ‘outdoor’ pigs? ................................................................................. 12

2. What is the absolute risk of visual-only inspected outdoor pigs? ................................. 12

Risk pathway .................................................................................................................... 13

Hazard Identification ......................................................................................................... 15

Hazard identification for Question 1: ................................................................................. 15

Is the sensitivity of detecting a condition affected by the inspection method? ..................... 15

Previous work ..................................................................................................................................... 15

Hazard identification 1PAH: ................................................................................................................ 16

Hazard identification 1AW: ................................................................................................................. 17

Study FS145003 Objective 3 .............................................................................................................. 18

What conditions in free-range pigs were observed in our study at statistically different

frequencies by the two inspection methods? ........................................................................ 18

Do they pose a potential hazard to public health? ................................................................ 20

Do they pose a potential hazard to animal health? ............................................................... 20

Do they pose a potential hazard to animal welfare? ............................................................. 21

Hazard identification 1PAH & 1AH & 1AW: ........................................................................................ 22

Do our study findings support the animal health associated public health hazards identified

in previous work? .................................................................................................................. 22

Hazard identification 1PAH: ................................................................................................................ 22

Hazard identification for Question 2: ................................................................................. 22

Is the level of carcass contamination affected by the inspection method? ........................... 22

A – Identify the micro-organisms or the microbial toxins of concern with food (CAC, 1999) 22

Previous work ..................................................................................................................................... 22

Hazard identification 2A-PHCM: ......................................................................................................... 23

Study FS145003 Objectives 6 & 7...................................................................................................... 25

B – Identify other non-microbial carcass contaminants that might be affected by the

inspection method. ................................................................................................................ 25


Carcass meat (internal) contaminants ............................................................................... 25

Hazard identification 2B –PH CI: ........................................................................................................ 26

Carcass external contaminants ......................................................................................... 26

Study FS145003 Objective 3 .............................................................................................................. 26

Hazard identification 2B –PH CE:....................................................................................................... 27

Hazard Characterisation ................................................................................................... 27

Endocarditis in pigs ........................................................................................................... 27

Endocarditis - Streptococcus spp. including Streptococcus suis ....................................................... 27

Endocarditis - Erysipelothrix rhusiopathiae ........................................................................................ 28

Granulomatous lesions - Rhodococcus equi ...................................................................................... 28

Granulomatous lesions - Mycobacterium spp. ................................................................................... 29

Salmonella spp. .................................................................................................................................. 29

Yersinia spp. ....................................................................................................................................... 29

Total aerobic and Enterobacteriacae counts ...................................................................................... 30

Exposure Assessment ...................................................................................................... 30

What is the current baseline exposure assessment with the traditional inspection method? ......................................................................................................................................... 31

What is the contribution to the actual public health exposure of organisms associated with

endocarditis and granulomatous lesions in pork meat? ........................................................ 31



Granulomatous lesions - Rhodococcus equi ...................................................................................... 31

Granulomatous lesions - Mycobaterium spp.: .................................................................................... 31

Endocarditis and granulomatous lesions public health exposure assessment summary .................. 32


carcass microbial contamination of pork meat? .................................................................... 32

Salmonella spp. .................................................................................................................................. 32

Y. enterocolitica* ................................................................................................................................. 33

Total aerobic and Enterobacteriacae counts ...................................................................................... 34

Microbiological public health exposure assessment summary .......................................................... 34

What is the baseline for animal health? i.e. disease prevention and control? .................... 34

What contribution does information derived from post-mortem inspection data have in the

prevention and control of disease in the outdoor pig population? ........................................ 34

Endocarditis – general ........................................................................................................................ 34



Granulomatous lesions - General ....................................................................................................... 34

Granulomatous lesions - Rhodoccus equi .......................................................................................... 35

Granulomatous lesions - Mycobaterium spp: ..................................................................................... 35

Salmonella spp. .................................................................................................................................. 35

Y. enterocolitica* ................................................................................................................................. 35

Risk Characterisation ........................................................................................................ 36


Is there a change in the risk profile if the inspection method is changed from traditional to visual-only? ....................................................................................................................... 36

How many pigs from non-controlled housing conditions are currently processed annually in

Great Britain? ........................................................................................................................ 37

Risk assessment discussion ............................................................................................. 42

Risk assessment conclusions ........................................................................................... 44

Risk mitigation measures .................................................................................................. 45

Impacts ............................................................................................................................. 46

What would be the impact of the introduction of visual-only inspection of fattening pigs from

non-controlled housing conditions in the UK on…? .............................................................. 46

Impact on public health ..................................................................................................... 46

Impact on occupational exposure ..................................................................................... 46

Impact on animal health .................................................................................................... 46

Impact on animal welfare .................................................................................................. 47

Impacts on resources ....................................................................................................... 47

Other impacts ................................................................................................................... 48

Benefits ............................................................................................................................ 49

References ......................................................................................................................... 52


TABLES AND FIGURES

Figure A4:1: The components of risk analysis ....................................................................... 8

Table A4:1: Comparison of the components of two risk assessment approaches ................. 8

Table A4:2: Differences in visual-only inspection compared to traditional inspection in outdoor pigs during FSA trial FS145003.............................................................................. 14

Table A4:3: Conditions for which detection might be affected by the inspection method and their relevance to a potential change in risk adapted from Hill et al., (2011) ........................ 15

Table A4:4: Conditions for which detection might be affected by the inspection method and their relevance to a potential change in risk to humans adapted from the Danish Way (Alban et al., 2008). ........................................................................................................................ 16

Table A4:5: Descriptive analysis of the frequency of the eight conditions (with a statistically significant difference between the two inspection methods) found in pigs from non-controlled housing conditions at post-mortem meat inspection by the two inspection methods; number (n), percent of total carcases inspected (%) and 95% confidence interval (95% C.I.) .......... 19

Table A4:6: Summary of the eight conditions where there was a statistical significant difference between visual-only and traditional inspection methods...................................... 19

Table A4:7: Hazard identification – conditions that detection of which could be or were affected by the inspection method and their relevance to a potential change in risk ............ 24

Table A4:8: The descriptive analyses of the frequencies (number=n [%] and where appropriate 95% confidence interval, C.I.) found in the study abattoir in FS145003 by the two inspection methods for the hazards identified...................................................................... 37

Table A4:9: Student t-test for the comparison between the two inspection methods for Enterobacteriaceae counts where Enterobacteriaceae = presence and for the log10 of total aerobic plate counts. ........................................................................................................... 37

Table A4:10: The predicted guesstimates of the total annual numbers of hearts/carcases potentially missed if visual-only inspection was implemented as per the trial conditions and prevalence in all ‗outdoor finished‘ pigs was the same as per the trial conditions. ............... 38

Table A4:11: Summary of the Risk Characterisation for Public Health for the five hazards for which detection could be or was affected by the inspection method .................................... 40

Table A4:12: Summary of the Risk Characterisation for Animal Health (AH) for the two major hazards for which detection could be or was affected by the inspection method ................. 41

Table A4:13: Potential impacts on resources of the introduction of visual-only inspection for fattening pigs ...................................................................................................................... 47

Figure A4:2 Public and animal health hazards that arise from animal health conditions and their relative risks comparing inspection methods ............................................................... 50

Figure A4:3: Public health (foodborne) hazards from carcass microbial contaminants and their relative risks comparing inspection methods ............................................................... 51


BACKGROUND

The problem

The focus of traditional methods of meat inspection is on the detection of gross lesions or

flaws in the carcass. Such methods of meat inspection are not always suitable for detecting

some important food-borne pathogens. Modernisation of post-mortem inspection regulations

in the European Union (EU) means that, provided that certain requirements are met,

carcases of fattening pigs reared under controlled housing conditions in integrated

production systems since weaning need only undergo visual inspection (EC Regulation

854/2004). This does not involve palpation or incisions and may reduce the risk of cross

carcass contamination. Uptake by the UK pig industry has been low because

slaughterhouses accept a mixture of indoor and outdoor reared pigs throughout the day and

the latter still have to be inspected by traditional means.

Project FS145003

Our project – ‗Trial of visual inspection of fattening pigs from non-controlled housing

conditions‘ - included a field trial of visual versus traditional meat inspection of fattening pigs

from non-controlled housing conditions. Batches of such pigs (termed ‗outdoor‘ or ‗free-

range)‘ were inspected by both visual and traditional methods and carcass sponge samples

were taken from a subset of pigs on the slaughter line after each of the inspection points.

The conditions found by each inspection method were recorded and compared (Objectives

1, 2 & 3). The microbiological work included total aerobic plate counts; Enterobacteriaceae

counts, and the isolation of Salmonella and Yersinia (Objectives 6 & 7). Obstacles to the

implementation of a risk-based visual-only inspection system were also investigated

(Objective 5). In addition, historic data from a full calendar year of traditional meat inspection

in the abattoir was analysed, in order to establish a baseline for the frequency of conditions

recorded by this method for pigs from different management systems (Objective 4).

The results from these investigations have been used to inform a risk assessment. Objective

8 is to undertake a formal, mostly qualitative, risk assessment, based on guidelines

described by the Codex Alimentarius Commission (CAC), on the impact of the

implementation of visual-only inspection for fattening pigs from non-controlled housing

conditions and the impacts that this could have in terms of public health, animal health,

animal welfare and the allocation of resources. This work is reported here.

RISK ANALYSIS

Risk analysis consists of four components – hazard identification, risk assessment, risk

management and risk communication. The purpose of this work is predominantly risk

assessment i.e. to address the first two parts: hazard identification and risk assessment

(Figure A4:1). Communication is required throughout the process; however whether the risk

is acceptable is, ultimately, a risk management decision. Risk management includes the

consideration of: the identified and assessed risks versus the potential benefits and the

costs; possible options, risk mitigation and reduction measures; implementation and

subsequent monitoring needs.


Figure A4:1: The components of risk analysis

There are two recognised, formalised approaches to risk assessment. These are the World

Organisation for Animal Health (OIE, 2004) Terrestrial Animal Health Code and that of the

Codex Alimentarius Commission (CAC, 1999). The stages and terminology used in these

systems, and their relationships are described below (Table A4:1).

Table A4:1: Comparison of the components of two risk assessment approaches

OIE approach Codex Alimentarius Commission (CAC) approach

Hazard identification Hazard identification

Risk Assessment

Release assessment Hazard characterisation

Exposure assessment Exposure assessment

Consequence assessment

Risk estimation Risk characterisation

Objective 8

We were asked to use a CAC approach to the Risk Assessment. This has needed to be

modified in order to address the non-public health aspects of the investigation.

Hazards have been identified then characterised. Hazard characterisation is a qualitative or

quantitative description of the severity and duration of adverse effects that may result from

the ingestion of a micro-organism or its toxin in food. A dose-response assessment should

be performed if the data are obtainable (CAC, 1999). We give a qualitative description of the

characteristics of each organism, identified as a hazard, when ingested by humans. In

addition, we describe the general characteristics of the hazard in pigs. Exposure

assessment estimates the level of microbiological pathogens or toxins and the likelihood of

their occurrence in foods at the time of consumption (CAC, 1999). For public health hazards,

whether they arise from animal health conditions or contaminants such as microbiological

organisms, the extent of actual or anticipated human exposure will only change if the risk

profile of the final (chilled) pork carcass (FCPC) changes significantly and if that change is

not combated by existing measures in the food chain after the final (chilled) pork carcass

stage. We have investigated the contribution of pork meat to the total public health exposure

to organisms that are associated with endocarditis and granulomatous lesions and carcass

microbial contamination. For hazards of concern to animal health we have investigated the

contribution that information derived from post-mortem inspection data makes to the

prevention and control of disease in the outdoor pig population. We have then characterised

Hazard identification Risk assessment

Risk management

Risk communication


the risk that exists when outdoor pigs are inspected by traditional methods. This entails the

integration of hazard identification, hazard characterisation and exposure assessment to

obtain a risk estimate (CAC, 1999). We have then compared this to the risk that exists when

outdoor pigs are inspected by a visual-only method.

Subsequent to the risk assessment we have then explored potential methods for risk

mitigation; taken a look at the impacts and at the potential benefits of implementing a visual-

only method of inspection for outdoor pigs, from non-controlled housing conditions and

drawn our overall conclusions.

RISK ASSESSMENT

Previous work

An assessment of the risk for humans associated with supply chain meat inspection – the

Danish Way (Alban et al., 2008) - did not investigate the difference in production

management systems; just that between the inspection methods. From this assessment, it

was concluded that:

the omission of incisions into the mandibular lymph nodes and the routine opening of

the heart for swine carcases do not seem to be associated with an increased risk for

human health;

there could be a positive effect on the working environment and no negative effect on

animal health.

A qualitative risk assessment (Hill et al., 2011) of the comparative risks to public and animal

health from visual inspection of indoor and outdoor pigs concluded that the risk was

negligible for all pigs. However, there were insufficient data to assess if there would be a

reduction in food-borne risk due to reduced microbiological carcass contamination. Both of

these studies utilised versions of the OIE risk assessment approach.

For a recent EFSA opinion (EFSA 2011b), on meat inspection of swine, a qualitative risk

assessment of foodborne hazards resulted in the identification of four hazards of public

health significance from the slaughter of pigs. The risk assessment was conducted using

data on prevalence on chilled carcases, incidence and severity of disease in humans, and

source attribution of hazards to pork, with the chilled carcases as the target. Salmonella spp.

were considered of high relevance currently in the EU, while Yersinia enterocolitica,

Toxoplasma gondii and Trichinella spp. were considered to be of medium relevance. The

risk reduction measures indicated for Salmonella spp. and Y. enterocolitica would also help

in the control of a number of other microbial hazards. It was recommended that inspection by

palpation and incision should be omitted in pigs subjected to routine slaughter as the risk of

microbiological contamination is higher than the risk associated with any potential reduction

in the detection of conditions currently targeted by these techniques. However; adoption of

inspection without palpation and incision is suggested as one part only of a wider

‗comprehensive pork carcass safety assurance‘ system. The scope of the work was to

evaluate meat inspection in a public health context; animal health and welfare issues were

considered with respect to the possible implications of adaptations/alterations to current

inspection methods, or the introduction of novel inspection methods that might be proposed.

Other issues that relate to the fitness of the meat for human consumption, transmissible


spongiform encephalopathies and the impact of changes to meat inspection procedures on

the occupational health of abattoir workers and inspectors, or to controls related to any

biological hazards at any meat chain stage beyond the abattoir, and the implications for

environmental protection were not included. Two methodologies (qualitative and

quantitative) were used; the former relied on expert opinion and a review of the literature,

and the latter used a three stage epidemiological modelling approach. During current

systems of meat inspection, the probability of detection of conditions is often low, particularly

for non-typical cases. This would be reduced further in the proposed modified system of pig

meat inspection although the magnitude of the difference would vary with the

disease/condition. It was recommended that this reduced detection probability could be

mitigated by follow-up inspection i.e. when abnormalities are seen with visual inspection then

further palpation and/or incision should be conducted.


Define the question

For the risk assessment the overarching questions are:

‘How would the risk profile change if fattening pigs from non-controlled housing conditions

are visually inspected, in terms of:

1. Public health

2. Animal health, and

3. Animal welfare

The impact on allocation of resources is then a consequence of risk management options

and decisions that arise from these risk assessments rather than part of the primary risk

assessment.

The fundamental question is:

What is the baseline against which any change in the risk profile is to be considered?

Hill et al. (2011) approached the problem as follows:

Indoor pigs were defined as ‗pigs raised specifically for slaughter indoors since

weaning on quality-assured farms‘. Outdoor pigs were defined as ‗pigs raised

specifically for slaughter outdoors since weaning on quality-assured farms‘.

Pigs reared indoors since weaning (with appropriate Food Chain Information (FCI)

and from integrated production systems) can be visually inspected; therefore the risk

that these pigs pose to public health and animal health and welfare must be

acceptable. If ‗outdoor pigs‘ pose the same or less of a risk, in respect of specified

hazards, then visual inspection of outdoor pigs must also be acceptable.

Their questions were, therefore:

1. What is the relative risk posed by visual-only inspection of outdoor pigs over visual-

only inspection of indoor pigs?

2. What is the absolute risk of visual-only inspected outdoor pigs?

Study FS145003 has not been designed to compare ‗indoor‘ and ‗outdoor‘ pigs; rather, it

compares the two inspection methods when applied to fattening pigs from non-controlled

housing conditions.

Definitions

The following definitions apply to FS145003 and this risk assessment:

1. Fattening pigs - ‗a pig raised specifically for slaughter.‘

2. Non-controlled housing conditions or ‗outdoor‘ – fattening pigs raised entirely

outdoors from weaning to slaughter. These are often referred to at the abattoir as

‗free-range‘ or ‗organic‘ pigs. Within this study the terms ‗outdoor‘ and ‗free range‘ are

used interchangeably and are used to refer to pigs that have been raised as

described above. As far as the authors are aware, all such commercial fattening pigs

in Great Britain (GB) are also born in non-controlled housing conditions. They need


to be distinguished from outdoor pigs that are raised outdoors and enter controlled

housing conditions at some point to be finished before they are sent to the abattoir. It

was assumed, in the design of the study, that fattening pigs raised entirely outdoors

from weaning to slaughter would be more likely to be different to pigs raised totally

indoors than fattening pigs raised partially outdoors and partially indoors; therefore

they would potentially present the greatest risk, if inspection systems were to be

changed.

3. Visual-only inspection - this is an inspection of the carcass without any palpation or

incision of any part of the carcass. It is an inspection modified from the requirements

of the European Hygiene Regulation (EC) 854/2004 (Anon., 2004b), in which the

carcases are not handled, palpated or incised, they are only examined visually (See

Table 2).

4. Traditional inspection1 - this is an inspection of the carcass where different parts of

the carcass are visualised, palpated and/or incised (See Table 2). This is derived

from the requirements of the European Hygiene Regulation (EC) 854/2004 (Anon.,

2004b).

The assumption for FS145003 is that fattening pigs from non-controlled housing conditions

inspected by traditional methods pose an acceptable risk to public health and animal health

and welfare. Thus, if the same type of pigs pose the same or less of a risk (in respect of

specified hazards) when inspected by visual-only methods, then these risks must also be

acceptable.

The first question then changes, from that asked by Hill et al., (2011) as follows:

1. What is the relative risk posed by visual-only inspection of ‘outdoor’ pigs over

traditional inspection of ‘outdoor’ pigs?

2. What is the absolute risk of visual-only inspected outdoor pigs?

There are two main criteria that will determine whether the risk profile will change with the

inspection method. These are:

1. Whether the sensitivity of detecting a condition is affected by the inspection method;

if not, then there will be no change in risk;

2. Whether the level of carcass contamination is affected by the inspection method; if

not, then there will be no change in risk.

The absolute risks remain as stated in Hill et al., (2011):

To public health – determined by the relationship between the burden of

contaminated meat entering the food chain and the rates of human illness

attributable to that contaminated pig meat.

To animal health – this may alter if there is any change in the information that is

available to be reported back to the pig farmers.

1 Meat hygiene inspector (MHI) is the UK term for Official Auxiliary


The latter will be determined by the complex relationship between the detection of a

condition and the outcome of any action taken due to the dissemination of that information in

terms of subsequent ‗cases saved‘. In addition, the absolute risk to animal health depends

on any other reporting that arises from the post-mortem inspection data e.g. the identification

and reporting of notifiable diseases.

Risk pathway

The risk pathway and, therefore the risks, on farm, during transport, ante-mortem and during

the slaughter process up until the carcass is presented for inspection are assumed to remain

constant for all outdoor pigs, regardless of the inspection method used.

In addition, the risk pathway and the processes for the final (chilled) pork carcass in the rest

of the food chain are assumed to remain constant. We are, therefore, considering at this

point only the portion of the risk pathway from slaughter to the relative risk of the final

(chilled) pork carcass.


Table A4:2: Differences in visual-only inspection compared to traditional inspection in outdoor pigs during FSA trial FS145003

Traditional Inspection Visual-only Inspection Head visual visual

Tongue visual visual

Submaxillary lymph nodes visual, incise visual Mouth visual visual Fauces visual visual Throat visual visual Lungs visual, palpate, incision¹ visual Trachea visual, incision¹ visual Main bronchi branches visual, incision visual Oesophagus visual visual Bronchial and mediastinal lymph nodes

visual, palpate visual

Pericardium visual visual Heart visual, incision visual Diaphragm visual visual Liver visual, palpate visual Hepatic and pancreatic lymph nodes


GIT and mesenteric visual visual Gastric and mesenteric lymph nodes

visual, palpate, incision² visual

Spleen visual, palpate² visual Kidneys visual, incision² visual Renal lymph nodes incision² visual Pleura and peritoneum visual visual Genital organs visual³ visual Udder visual visual Supramammary lymph nodes

visual, incision², 3, 4 visual

Umbilical region (young) visual, palpate, incision² visual Joints (young) visual, palpate, incision² visual Based on the legal requirements contained in Regulation (EC) 854/2004.

¹When for human consumption ²When necessary ³Unless penis discarded 4Sows


Hazard Identification

This step is common to both risk assessment methodologies.

Hazard identification for Question 1:

Is the sensitivity of detecting a condition affected by the inspection method?

Previous work

This question has been considered in depth by Hill et al., (2011). A shortlist was produced of

pig diseases and conditions that could potentially be affected by a change from traditional to

visual inspection methods. It did not include conditions where detection would be unchanged

or that clearly did not pose a risk to human or animal health. A preliminary risk assessment

led to the identification of two conditions for which a change in risk to public or animal health

might potentially occur: porcine tuberculosis (pTb) and endocarditis. The conditions that

were shortlisted are shown in Table A4:3.

Table A4:3: Conditions for which detection might be affected by the inspection method and their relevance to a potential change in risk adapted from Hill et al., (2011)

Condition Preliminary risk assessment outcome

Comment

Endocarditis – usually bacterial (Streptococcus spp, Erysipelothrix rhusiopathiae, occasional parasitic or mycotic lesion)

Change (^) in human and animal health/welfare risk may arise.

Full risk assessment required because of potential serious human health issues, (although very low baseline) and lower sensitivity using visual inspection

Further investigation of outdoor-indoor pig prevalence advised; possibility that indoor prevalence is higher than outdoor. This is not an issue when comparing outdoor v. outdoor; the assumption is that inspection method will not change actual prevalence.

Necrosis in lymph nodes – possibly due to infection with Mycobacterium bovis, Mycobacterium avium, or Rhodococcus equi

Change (^) in human and animal health/welfare risk may arise.

Full risk assessment required because of the importance of Mycobacterium infections (although very low apparent prevalence) and unlikely to be detected using visual inspection

Although apparent prevalence is low in pigs; it is considered to be higher in outdoor pigs. This is not an issue when comparing outdoor v. outdoor; the assumption is that inspection method will not change actual prevalence.

Lungworm lesions No change in human health risk

Unlikely to change (^) animal health/welfare risk as initial sensitivity of detection is low anyway

Very unlikely to be detected using visual inspection

Pericarditis (acute & chronic) No change in human health risk

Unlikely to significantly change (^) animal health/welfare risk

Likely to be detected visually; possibly small reduction in sensitivity of detection through visual inspections.

Human health risk only if due to a subset combination of inadequately cooked meat and an acute case; most are not pathogens relevant to human health.

It was assumed that indoor


Condition Preliminary risk assessment outcome

Comment

prevalence was > outdoor.

Tape worms (cysticercosis) No change in human health risk

Unlikely to change (^) animal health/welfare risk as likely to be detected visually when carcass split; possibly small reduction in sensitivity of detection through visual inspections.

Human health risk is due to consumption (if meat is inadequately cooked)

In the assessment of the risk for humans associated with supply chain meat inspection – the

Danish Way (Alban et al., 2008) - the following potential hazards were considered (Table

A4:4), of which two were further assessed. These two were granulomatous lesions and

endocarditis.

Table A4:4: Conditions for which detection might be affected by the inspection method and their relevance to a potential change in risk to humans adapted from the Danish Way (Alban et al., 2008).

Mandibular lymph nodes Hearts

Granulomatous lesions due to:

Tuberculosis lesions Pericarditis

Rhodococcus equi Epicarditis

M. avium paratuberculosis Apostematous myocarditis (abscess in the heart)

Visceral larval migrans

Parasites Endocarditis

Neoplasm

Fungi

Grey text = identified but not relevant; black text = identified and considered to be relevant hazards

It has not been considered necessary to repeat hazard identification for animal and public

health conditions, given the extent of the work conducted in these two studies and the

similarities of their findings. Porcine tuberculosis (pTb) lesions are usually considered as a

subset of granulomatous lesions.

The prevalence of conditions such as endocarditis and granulomatous lesions are proxy

measures for specific organisms that may be a public health hazard. In the Danish Way

(Alban et al., 2008) these hazards were identified as Streptococcus spp. & Erysipelothrix

rhusiopathiae and Rhodoccus equi and Mycobaterium spp, respectively

Hazard identification 1PAH: the animal health associated public health hazards are,

therefore, considered to be endocarditis and granulomatous lesions to include porcine

tuberculosis lesions.

Hazard identification of conditions that might pertain to animal welfare was not part of the

Danish Way (Alban et al., 2008) risk assessment, whereas Hill et al., (2011) considered only

the welfare aspects of the diseases/conditions identified as hazards within the exposure and


consequences part of the risk assessment (all negligible to very low). Lesions of importance

for animal welfare can be identified both during ante and post-mortem inspection. The

following were identified in the ‗Overview on current practices of meat inspection in the EU‘

(Alban et al., 2011) as procedures by which animal welfare might be assessed in pigs:

Inspection of the surface of the skin as well as identification of fractured bones

(among others injuries caused during transport of the animals);

inspection of the gastrointestinal tract, the mesenterium, and the related lymph nodes

(Lnn. gastrici, mesenterici, craniales or caudales) (Gastric ulcer, peritonitis related to

hernia);

inspection and palpation of the umbilical area as well as the joints in young animals.

In case of doubt, incisions into the area around the umbilical area are made, and the

joints are opened (hernia, umbilical infection). This lesion might be considered of

relevance for animal welfare if a high prevalence is observed.

inspection of the genitalia (except from penis if it has already been condemned).

(Late pregnancy / torsion of testicle / hernia).

Not all of these will be applicable to fattening pigs (e.g. late pregnancy) and the majority are

already only assessed by visual inspection during the current traditional inspection protocol

(see Table A4:2).

The probabilities of detection of typical cases during ante and post-mortem inspection of five

welfare conditions were modelled in the external scientific report to EFSA (COMISERV,

2011) on the contribution of meat inspection to animal health surveillance in swine. These

conditions were lameness, arthritis and bursitis, tail biting and/or amputation, bruising and

skin lesions, and Dark, firm and dry (DFD) meat. The probabilities of detection did not

change significantly between conventional and visual-only inspection methods.

Hazard identification 1AW: no additional conditions have been identified as hazards for

animal welfare.


Study FS145003 Objective 3

The results from this comparison of the baseline values of the frequency and type of

conditions found using visual-only inspection and traditional inspection procedures in

fattening pigs from outdoor housing conditions, obtained from the data collected during the

field trial, help to inform:

1. whether the hazards identified in the previous work (above) are relevant to the study

situation;

2. whether additional hazards are identified, specific to the study situation.

A trial was run in one abattoir over five separate weeks during the period from the end of

November 2011 until the middle of March 2012. In this trial 11,086 carcases of fattening

pigs, from 62 batches and 12 farms, from non-controlled housing conditions were inspected

using both post-mortem inspection methods (traditional and visual-only inspection). The

type, frequency and distribution of conditions detected by each of the post-mortem

inspection methods was established and then compared. The effect of season and farm of

origin as potential confounders were investigated in the statistical analyses, as was the

effect of farm of origin as a cluster. No such effects were found. The analyses were

corrected for multiple significance testing.

What conditions in free-range pigs were observed in our study at statistically different

frequencies by the two inspection methods?

At a level of statistical significance of P < 0.00256 there were differences in the frequencies

found by the two inspection methods for eight of the categories of conditions (Tables A4:5 &

A4:6). These were milk spots, renal pathology (including kidney lesions, kidney pathology

and suspect uraemia), enteritis (including enteritis, colitis and pathology in the guts), pluck

pathology (as stated), faecal contamination of the carcass and hair contamination of the

carcass and, when analysed as absence/presence, for generalised conditions (suspect

pyaemia, suspect fever/septicaemia, oedema/emaciation and anaemia) and endocarditis.

The frequencies observed were higher with the visual method of detection for hair

contamination (visual > traditional).

The frequencies observed were higher with the traditional method of inspection for

milk spot, renal pathology, enteritis, pluck pathology and faecal contamination, and

for the absence/presence of generalised conditions and endocarditis (traditional >

visual).


Table A4:5: Descriptive analysis of the frequency of the eight conditions (with a statistically significant difference between the two inspection methods) found in pigs from non-controlled housing conditions at post-mortem meat inspection by the two inspection methods; number (n), percent of total carcases inspected (%) and 95% confidence interval (95% C.I.)

Conditions in whole carcases and offal


N % 95% C.I.

n % 95% C.I.

Generalised Conditions Suspect pyaemia, suspect fever/septicaemia,

oedema/emaciation,anaemia

12 0.1 0.02 – 0.20

55 0.5 0.16 – 0.83

Milk Spot Milk spot 263 2.4 1.33 – 3.41

601 5.4 3.52 – 7.32

Renal Pathology Kidney lesions, kidney pathology,

suspect uraemia 448 4.0

3.28 – 4.80 798 7.2

5.87 – 8.52

Endocarditis Endocarditis 0 0 21 0.2 0.05 – 0.33

Enteritis Enteritis, colitis, pathology in the guts 22 0.2

0.97 – 0.33 123 1.1

0.75 – 1.47

Pluck Pathology Pathology in the pluck 181 1.6 0.82 – 2.44

524 4.7 3.47 – 5.98

Faecal Contamination Faecal contamination in any part of carcass/offal

353 3.2 2.16 – 4.21

545 4.9 3.53 – 6.30

Hair Contamination Hair contamination in any part of carcass/offal

1362 12.3 9.83 – 14.75

419 3.8 1.20 – 6.36

Table A4:6: Summary of the eight conditions where there was a statistical significant difference between visual-only and traditional inspection methods

Condition Mean of the difference (traditional

– visual) %

P value

Milk spots 3.21 <0.001

Renal Pathology 3.09 <0.001

Enteritis 0.99 <0.001

Pluck pathology 3.02 <0.001

Faecal contamination 1.62 <0.001

Hair contamination -8.21 <0.001

Generalised conditions <0.001

Endocarditis 0.0015

We did not expect to see a difference in detection rates for these generalised conditions

(suspect pyaemia, suspect fever/septicaemia, oedema/emaciation and anaemia) between

the two inspection methods as Meat Hygiene Inspectors (MHIs) currently use predominantly

visual inspection to identify them. On further examination of the data, the only one of the four

component conditions that had a statistically significant difference between the two


inspection methods was suspect pyaemia (absence/presence; McNemar test p=0.00018,

mean of the differences 0.65, 95% C.I. 0.23-1.1, p=0.0031). In most of the batches where

MHIs detected suspect pyaemia, there was only one positive carcass. However, there was

one batch with 11 suspect pyaemia (out of 79) positive carcases recorded and another one

with six suspect pyaemia (out of 427) positive carcases recorded. These numbers are

unexpected, especially so when the general health of pigs arriving to the abattoir is as good

as it was observed to be in the study. The Meat Hygiene Service (MHS) reported that only

0.35% of carcases slaughtered in British abattoirs were rejected as a whole (whole carcass

rejected) in 2005 (NADIS, 2012), mostly due to pyaemia, septicaemia and oedema. Pyaemia

was the major cause, responsible for 55% of the total rejection of carcases in the UK during

that year i.e. 0.19% of all pig carcases; seen in 19 in 10,000 carcases or almost two in 1000

carcases. Suspect pyaemia prevalence in carcases in abattoirs in UK observed in the British

Pig Executive (BPEX) monitoring scheme (BPHS) is lower than 1% of carcases inspected

(BPEX, 2008). In addition, in the analyses of historical data from the study abattoir the mean

prevalence of pyaemia in batches in which the condition was present in free range pigs was

only 1.1% (95% C.I. 1.0 – 1.3). In the opinion of the authors the most likely explanations for

the unprecedented numbers (over 10%) recorded at the traditional inspection point in these

two batches are due to recording error and, possibly, misclassification. This conclusion is

supported by further examination of the data. Neither ‗Generalised conditions‘ nor suspect

pyaemia are, therefore, considered to be a potential hazard. This position is supported as

Hill et al., (2011) did not include it in their first round of hazard identification.

Do they pose a potential hazard to public health?

Endocarditis cannot be detected by visual-only inspection; therefore, it has to remain as a

possible animal health associated public health hazard.

Milk spot was not considered by Hill et al., (2011) to be a hazard as it could be detected

visually, did not require palpation and incision and was not thought to be a public health

hazard when consumed. Apart from endocarditis, none of the other pathological conditions

identified above (renal pathology, enteritis, and pluck pathology) were considered in their

first round of hazard identification, with the possible exception that 'Nephropathy (abnormal

kidney)‘ would be included in our classification of ‗renal pathology‘. Although detection of

nephropathy requires both a visual and incision approach, they (Hill et al. 2011) identified no

public health risk.

A difference in the frequencies of faecal contamination recorded between the visual and

traditional inspection methods was observed in our study. It is possible that this could be

considered an additional public health hazard. It would however be most likely to be

considered as a proxy measure for microbiological contamination, which we have studied

directly in this field trial (see Hazard identification for Question 2) and so the question of

whether this is a potential hazard will be considered there.

Do they pose a potential hazard to animal health?

Although there are statistical differences between the amounts found by the two inspection

methods, with traditional inspection frequencies being higher than visual inspection

frequencies for seven of the categories of conditions investigated, this does not

automatically mean that there is a potential hazard to animal health. The difference observed


may not mean anything in terms of animal health, the disease process, the number of cases,

actual prevalence or the potential to ‗save‘ cases if control strategies are implemented

subsequent to the condition being recorded in the abattoir, i.e. the difference may not be

biologically significant, or clinically relevant.

Faecal contamination is not a condition that would have a direct or indirect relationship with

animal health. This leaves milk spots, renal pathology, enteritis, pluck pathology and

endocarditis. Although the pigs were individually inspected by both methods, the number of

pigs with a particular condition was recorded at batch level e.g. 3/50 and the frequency

calculated (%). For milk spots, renal pathology, enteritis and pluck pathology, the difference

in frequency found by the two inspection methods was calculated (traditional minus visual)

for each batch and the mean of the differences estimated. These mean differences are very

small (less than 3.5%). This, combined with the fact that for three of the conditions (milk

spots, renal and pluck pathology) there is a within-batch mean prevalence (when present) of

more than 5%, and they have a statistically significant correlation between the findings by

the two inspection methods and that for all four conditions there is a high batch level

prevalence (more than 50% of batches affected) implies that at batch level there is not a

biologically significant or clinically relevant difference between visual-only and traditional

inspection i.e. the reduced number of individual affected pig carcases found and recorded

when using visual inspection compared to traditional inspection is unlikely to have a

significant impact on the availability of relevant pathological data to feed back to the

producers for these four conditions (milk spots, renal pathology, enteritis and pluck

pathology).

Some of the differences observed within our trial between the frequencies found by visual

and traditional inspection may have been due to the total ―hands-off‖ nature of the study

design because access to the whole offal was substantially reduced. This would have

affected the ability to detect milk spot, renal pathology, enteritis and pluck pathology by

visual inspection. Even if they are not considered to be significant hazards, an improvement

in accessibility to all parts of the carcass should increase the ability to find these conditions

and decrease the differences.

Endocarditis occurs in low numbers, at a low mean prevalence within batches and a

comparatively low percentage of batches are found to be affected with the condition. In the

analysis of historical data from this abattoir the mean within-batch prevalence in batches of

free-range pigs, in which the condition was present, was 0.8%. The percentage of batches

affected with endocarditis was 3.5% (95% C.I. 2.6 – 4.4). If a condition occurs at a low

within-batch prevalence and the detection of its presence in a batch is reduced when the

inspection method is changed from traditional to visual, then this reduction in the number of

batches being classified as affected will have the greatest apparent impact on a condition

that affects a lot of batches i.e. the apparent number of affected batches will be reduced.

This may have a significant impact on the availability of relevant pathological data to feed

back to the producers and, subsequently, on animal health. Whether it actually has a

significant impact on animal health is dependent on whether producers take action to

mitigate or control the causes of the conditions in their herds after receiving such

information.

Do they pose a potential hazard to animal welfare?


None of the conditions identified in our study would have direct effects on animal welfare due

to any reduction in their identification. They may have indirect effects due to any implications

for animal health.

Hazard identification 1PAH & 1AH & 1AW: no additional animal health associated public

health hazard has been identified from our study.

No additional animal health associated animal health hazards have been identified.

No further animal health associated animal welfare hazards have been identified.

Do our study findings support the animal health associated public health hazards identified in

previous work?

The original animal health associated public health hazards that were identified in the earlier

studies were endocarditis and granulomatous lesions to include porcine tuberculosis lesions.

Endocarditis cannot be detected by visual-only inspection; therefore, it has to remain as a

possible animal health associated public health hazard. We were unable to investigate

whether there was a difference in the frequency of granulomatous lesions to include porcine

tuberculosis lesions identified by visual and traditional methods of inspection as none were

observed in the study abattoir during the study period. So, it too has to remain as a possible

animal health associated public health hazard.

Hazard identification 1PAH: the original animal health associated public health hazards

are, therefore, still considered to be endocarditis and granulomatous lesions to include

porcine tuberculosis lesions.

Hazard identification for Question 2:

Is the level of carcass contamination affected by the inspection method?

A – Identify the micro-organisms or the microbial toxins of concern with food (CAC, 1999)

Previous work

The recent EFSA opinion on the public health hazards to be covered by the inspection of

meat (swine) looks at the European Union (EU) situation as a whole (EFSA 2010b). The

BIOHAZ committee identified 15 biological hazards for which there is evidence that they

occur or may occur in pigs in Europe and that they can be transmitted via food to humans.

The main transmission pathways for Brucella suis, Erysipelothrix rhusiopathiae, Leptospirae

and Streptococcus suis were not considered to be foodborne (Anon., 2000) and so were not

included. There is a lot of uncertainty about the public health risk related to consumption of

bacteria that are resistant to antibiotics; however it was considered to be significantly lower

than the 15 pathogens listed. It was therefore concluded not to consider this characteristic of

microbiological hazards in isolation, but as one of pathogenicity attributes when ranking

them, especially for Salmonella and MRSA. A simple preliminary qualitative risk assessment

using a ranking algorithm was used to produce a ‗short list‘ for fresh (chilled) pork carcases.

Information on source attribution based on evidence, the literature and expert opinion

suggesting epidemiological links between human cases and pork, was considered as a

moderating factor in the final ranking of hazards. There was one final high risk (Salmonella

spp.) and several of a final medium risk ‐ Y. Enterocolitica*; Trichinella spp.; Toxoplasma

gondii; Sarc. suihominis and T. solium cysticercus. The latter two were excluded from further


consideration as there was no information on occurrence in carcases and human cases in

EU. The actual relevance to humans in the EU is, therefore, unknown and it was not

currently considered relevant in the EU pig population. However, they are to be monitored in

future. Those pathogens ranked as ‗low risk‘ were excluded from further consideration.

Trichinella spp. is detected by diagnostic testing that uses enzymatic digestion of muscle.

This procedure would not be altered by an alteration to the carcass inspection method. It is

not possible to transfer the micro-organism from one carcass to the other by manual

handling at inspection and so it is not further considered as a hazard relevant to our

situation. Current post-mortem inspection cannot macroscopically detect Toxoplasma gondii;

it will not be transferred by manual handling and so it too is not further considered as a

relevant hazard.

Hazard identification 2A-PHCM: this leaves the two microbial hazards identified as

Salmonella spp. and Y. enterocolitica*

*In this EFSA Opinion Yersinia enterocolitica is defined as ―...human enteropathogenic

Y. enterocolitica with biotype/serotype combinations that have their main reservoirs in pigs,

in particular biotype 4/serotype O:3, biotype 2/serotype 9, but also biotype 2/serotype O:5,

27.‖

In addition to the EFSA Opinion referred to above, there is control and surveillance of the

hygiene of the process to reduce the risk of pathogens on the meat. This includes aerobic

colony count, Enterobacteriaceae and Salmonella according to Regulation (EC) No

2073/2005 of 15 November 2005 on microbiological criteria for foodstuffs, Annex 1, Chapter

2.1 9 (Anon., 2005).


Table A4:7: Hazard identification – conditions that detection of which could be or were affected by the inspection method and their relevance to a potential change in risk

Potential Hazard Hazard area Ref: mentioned in Comment Retain

Public Health

Animal Health

Animal Welfare

Hill et al., (2011)

Alban et al., (2008)

FS145003 field study

EFSA (2010b)

Endocarditis * * * * *

Cannot be identified by visual inspection – has to be considered a public health hazard. Not currently included in reported trends analyses for industry , therefore not considered to be highly significant for animal health.

*

Granulomatous lesions including porcine tuberculosis

* * * * None found; cannot be excluded

Perceived public health importance of Mycobacterium infections

*

Milk spots * * *

Statistically significant finding tradition > visual inspection frequency recorded for free range pigs; mean of differences very small; unlikely to be of clinical/biological significance at batch level. Can be mitigated by improving visualisation/accessibility of offal.

Renal pathology * Part *

Enteritis * *

Pluck pathology * *

Faecal contamination

* * As above – in addition, considered a proxy measure

for general microbial contamination of the carcass that has been measured directly in the study.

Hair contamination * *

Statistically significant finding visual > traditional inspection frequency recorded for free range pigs;

Generalised conditions

* * * Statistically significant finding tradition > visual

inspection absence/presence frequency recorded for free range pigs in FS145003. Considered to be an anomaly of reporting and misclassification error

Salmonella spp. * * * *

No Salmonella spp. isolated in FS145003 *

Y.enterocolitica * * * *

No statistical difference detected in frequency of isolation of Yersinia spp from carcases inspected by the two methods n FS145003

*

Total aerobic colony count and Enterobacteriaceae

* Direct measure of general microbial contamination of

the carcass: assessment of hygiene of process. *


Study FS145003 Objectives 6 & 7

The results from the microbiological checks carried out during both visual-only and traditional

inspection of carcases during the field trial are able to inform this assessment with respect to

total aerobic plate counts, Enterobacteriaceae counts and the presence of Salmonella and

Yersinia and a comparison of their frequency between the inspection methods.

This aspect of the field trial is not designed to identify additional microbiological hazards.

In the same trial as previously mentioned in one abattoir over five separate weeks during the

period from the end of November 2011 until the middle of March 2012, microbiological

checks were carried out after visual-only inspection and after traditional inspection. Carcass

sponge sampling was used as described in Regulation 2073/2005 (Anon., 2005).

Total aerobic plate counts, Enterobacteriaceae counts and culture for Salmonella were

performed on 400 swabs taken after traditional inspection and 400 after visual-only

inspection. These were taken during the second, third, fourth and fifth four weeks of the trial

between 16th January 2012 and 16th March 2012. Samples could only be taken from

batches and pigs that were expected to be slaughtered before 10am to enable despatch to

the laboratory and receipt within 24 hours. Ten farms, 44 batches and 7,931 carcases from

outdoor pigs were included in the sampling frame. Only five batches were not included due

to their late arrival at the abattoir. Seven hundred and fifty-nine swabs were cultured for

Yersinia: 379 after traditional inspection and 380 after visual-only inspection. These came

from all five weeks of the study. Twelve farms, 54 batches and 9,633 carcases from outdoor

pigs were included in the sampling frame.

Potential biases and confounders were considered during the sampling design, or accounted

and adjusted for during the statistical analysis. The analyses were corrected for multiple

significance testing and P < 0.0127 was considered to be the level of statistical significance.

There was:

no statistical difference in the mean of the log10 of total aerobic plate contamination of

carcases after the two inspection methods;

a statistically significant difference in the mean of the log10 of the Enterobacteriaceae

count when present (i.e. >0); the level of contamination of carcases was lower after

visual-only inspection compared to traditional inspection;

no statistical difference in the proportion of Yersinia contamination of carcases found

after the two inspection methods;

and, no Salmonella spp. were isolated from any sample in the study.

B – Identify other non-microbial carcass contaminants that might be affected by the

inspection method.

Carcass meat (internal) contaminants

In the aforementioned recent EFSA Opinion (EFSA 2010b), carcass meat contaminants of

public health significance, such as chemical and medicinal residues, were considered. The

CONTAM Panel considered all substances listed in Council Directive 96/23/EC and

evaluated the outcome of the residue monitoring plans for the period 2005-2009. The


individual contaminants were ranked into four categories denoted as high, medium, low, and

negligible potential concern, after consideration of a number of criteria. Although the present

meat inspection system facilitates the implementation of the sampling for such contaminants,

their presence or absence is not determined by the current inspection procedures; further

laboratory analysis is required. Issues other than those of public health significance but that

still compromise fitness of the meat for human consumption (Anon., 2004b) e.g. sexual

odour (‗boar taint‘) will also not be significantly affected by a change in inspection procedure

from traditional to visual.

Hazard identification 2B –PH CI: no hazards to public health from carcass contaminants within the carcass have, therefore, been identified.

Carcass external contaminants

The frequency of presentation of any external carcass contaminants for outdoor pigs should

be independent of the inspection method, unless there is the possibility of cross-

contamination by the handling required for traditional inspection. In the historical analysis of

conditions recorded at ante and post-mortem inspection at the study abattoir (FS145003

Objective 4), five types of contamination were recorded as post-mortem processing faults.

These were blood splash, bile, faecal, grease and hair contamination. Blood splash occurs

before or during the stunning and subsequent slaughter process when blood leaves the

blood vessels due to an excessive blood pressure or trauma of the vessels. It can be seen in

small spots or bigger infiltration of whole muscle. It is not a public health issue, just an

aesthetic observation, but may indicate deficiencies with the stunning procedure i.e. have

welfare implications. Faeces and bile contamination could lead to contamination from

bacteria within them. The faeces or bile could be from the same carcass or result in cross-

contamination from another one. Grease refers to machinery grease; although it is food-

grade it could be considered as a potential hazard. Hair contamination is probably only an

aesthetic and food quality (rather than safety) issue because pig carcases go through the

scalding tank and burner and any remaining hair should be hygienic.

Study FS145003 Objective 3

At a level of statistical significance of P < 0.00256 there were statistical differences in the

frequencies found by the two inspection methods for two of the categories of conditions that

might have some relevance to external carcass contaminants with respect to microbiological

hazards (Tables A4:5 & A4:6). These were hair contamination of the carcass and faecal

contamination of the carcass.

The frequencies observed were higher with the visual method of detection for hair

contamination (visual > traditional).

The frequencies observed were higher with the traditional method of inspection for

faecal contamination (traditional > visual).

The presence of hair contamination (but see above) and faecal contamination could

conceivably be associated with presence of microbiological organisms. It is possible that

these both could be considered an additional public health hazard, i.e. as proxy measures

for microbiological contamination and the consequent potential for cross—contamination of

carcases. These aspects are studied directly in this field trial by measuring total aerobic


plate counts and Enterobacteriaceae counts. To avoid ‗double-counting‘ in the risk

assessment they have not been considered separately from these counts.

Hazard identification 2B –PH CE: the total aerobic plate counts and Enterobacteriaceae

counts will provide direct measures for the hygiene of the process, therefore indirect

potential hazards such as hair contamination and faecal contamination are not considered

individually.

Hazard Characterisation

Hazard characterisation is: a qualitative or quantitative description of the severity and

duration of adverse effects that may result from the ingestion of a micro-organism or its toxin

in food. A dose-response assessment should be performed if the data are obtainable (CAC,

1999).

We give a qualitative description of the characteristics of each organism, identified as a

hazard, when ingested by humans. In addition, we describe the general characteristics of the

hazard in pigs.

Endocarditis in pigs

Endocarditis is an inflammation of the internal lining, or endocardium, of the heart. It can

occur after a bacterial infection elsewhere in the body. If the bacteria circulate in the blood

stream (a bacteraemia) then they may reach the heart and cause an inflammatory response

there. Often this is seen as lesions on the valves. If these lesions are seen in the pig‘s heart

then it indicates that they have been, and may still be, infected with the bacteria.

Endocarditis - Streptococcus spp. including Streptococcus suis

Streptococci are common organisms in all animals including people. They may be present

as commensal organisms or can cause a range of human infections ranging from sore throat

to severe, life threatening infection such as meningitis and septicaemia. Long term outcomes

of infection can include endocarditis, imbalance and long term deafness. Fatal cases of

S. suis are rare, but not unknown. Patients who are immuno-suppressed or have had their

spleen removed (asplenic) are known to be at greater risk from disease. Infection may,

rarely, lead to toxic shock syndrome (TSS), which is difficult to treat and can cause multi-

organ failure (Health Protection Agency (HPA) accessed July 2012

http://www.hpa.org.uk/Topics/InfectiousDiseases/InfectionsAZ/StreptococcusSuis/GeneralInf

ormation/)

Streptococcus suis is carried in the tonsils of pigs, and pig-to-pig spread is mainly by nose-

to-nose contact or by aerosol over short distances. It may be transmitted to the sucking

piglet from the sow or from other piglets. Streptococcal meningitis in sucking piglets is

sporadic in individual piglets. It may be worse in sucking pigs when the organism has been

introduced into the herd for the first time, or where it is secondary to infection with PRRS.

Clinical signs are quick in onset; the piglet may be found dead, convulsing, or shivering, lying

on its side and paddling, with rapid side to side eye movements (nystagmus). S. suis is

associated with a variety of other conditions including septicaemia (infection of the blood),

polyserisitis (inflammation of the lining of the abdominal and chest cavities), arthritis,

endocarditis (infection of the heart) and pneumonia. It has also been isolated from cases of

rhinitis and abortion. The pattern and relative importance of the different syndromes vary in

different countries. S. suis is sub-divided into at least 34 serotypes. They vary in their

http://www.hpa.org.uk/Topics/InfectiousDiseases/InfectionsAZ/StreptococcusSuis/GeneralInformation/



pathogenicity and the diseases they cause, both between and within types. Some types

appear to be non-pathogenic and have been isolated mainly from healthy pigs, some are

mainly associated with lung lesions, and some have been isolated from other animal species

as well as pigs. Clinical outbreaks in pigs are usually associated with trigger factors such as

overcrowding, poor ventilation or weaning. It can be treated with penicillin.

Endocarditis - Erysipelothrix rhusiopathiae

Erysipeloid is rare; in humans it is usually a mild, but pruritic (itchy, intense burning

sensation) skin condition caused by the bacterium Erysipelothrix rhusiopathiae. The bacteria

are introduced accidentally from infected animals through pre-existing skin wounds. Human

erysipeloid is largely an occupational disease of slaughterhouse workers, agricultural

workers, and those in the meat-handling and fishing industries. Occasionally it can present

with systemic symptoms such as fever, muscle aches and headaches as well as a skin

lesion. Arthritis of the fingers, septicaemia and endocarditis are rare, but not unknown long

term effects. It can be treated with penicillin. There are also effective vaccines available to

prevent disease.

Pigs are the major reservoir. E. rhusiopathiae colonizes the pharynx of pigs and is shed in

the faeces, urine or oronasal secretions of 30-50% of healthy swine. It can also be isolated

from faeces, soil, water etc. in an infected animal's environment. Clinical presentations

include acute septicaemia and ‗diamond skin disease‘. Mortality can be quite high. Chronic

forms of infection include endocarditis and arthritis

http://www.vetmed.wisc.edu/pbs/zoonoses/Erysipelas/erysipelasindex.html.

Granulomatous lesions - Rhodococcus equi

The first case of Rhodococcus equi infection in a human was reported in 1967. While still not

commonplace, R. equi has been isolated increasingly, especially as an opportunistic

pathogen. Most human infections have been associated with immune system dysfunction,

for example due to organ transplantation or infection with the human immunodeficiency virus

(HIV). Concurrently, increasing recognition of R. equi as a pathogen has led to improved

laboratory identification of infections. Infections often manifest as pulmonary involvement or

soft tissue abscesses and they are associated with significant mortality. Treatment can be

challenging, requiring early diagnosis and prolonged combination antibiotic therapy,

sometimes in combination with surgical therapy. Necrotizing pneumonia is the most common

manifestation of R. equi infection. Extrapulmonary R. equi infections have included wound

infection, subcutaneous abscess, brain abscess, thyroid abscess, retroperitoneal abscess,

peritonitis, meningitis, pericarditis, osteomyelitis, endophthalmitis, lymphadenitis,

lymphangitis, septic arthritis, osteitis, bloody diarrhoea, and fever of unknown origin, among

others. Bacteraemia and dissemination of infection follow from the primary infection site,

which is usually the lung http://emedicine.medscape.com/article/235466-overview#a0104

R. equi is an important cause of pneumonia in foals. It can infect wild boar and domestic pigs

(Makrai et al., 2008) and is commonly found in soil. From the literature reviewed by Hill et al.,

R. equi infections do not seem to have an apparent effect on the health, welfare or

performance of pigs, being asymptomatic, i.e. there are no clinical signs. No additional

information has been found.

http://www.vetmed.wisc.edu/pbs/zoonoses/Erysipelas/erysipelasindex.html

http://emedicine.medscape.com/article/235466-overview#a0104


Granulomatous lesions - Mycobacterium spp.

Human tuberculosis (TB) caused by M. bovis cannot be distinguished from that caused by

the closely related M. tuberculosis (the human TB bacterium) on clinical signs. The course

and extent of the disease is the same, as is the treatment in most cases. Symptoms of

respiratory TB include weight loss, night sweats, fever and a persistent cough which may

contain blood or pus. The treatment of TB in people is long and involves a combination of

several drugs (http://www.hpa.org.uk/webc/HPAwebFile/HPAweb_C/1259151943662)

From the literature reviewed by Hill et al.,(2011) Mycobacterium spp infections do not seem

to have an apparent effect on the health, welfare or performance of pigs, being

asymptomatic, i.e. there are no clinical signs. They are considered a dead-end host (no

further transmission occurs) for M. bovis, although there is some evidence that M. avium

may be excreted in the faeces raising the possibility of pig-to-pig transmission. No additional

information has been found.

Salmonella spp.

Salmonella bacteria are widespread in human and animal populations. There are a large

(approximately 2500 strains) number of Salmonella serotypes that cause food poisoning,

typhoid and paratyphoid fevers in humans. Clinical signs of food poisoning include

diarrhoea, stomach cramps and sometimes vomiting and fever, which usually last for four to

seven days and is self-limiting i.e. it clears up without treatment. However, it can be more

serious and cause dehydration in young children, the elderly and people whose immune

systems are not working properly.

http://www.hpa.org.uk/Topics/InfectiousDiseases/InfectionsAZ/Salmonella/

In pigs, outbreaks of septicaemic salmonellosis are rare. Salmonellosis can occur at any age

but is most common in growing pigs over eight weeks of age. Severe S. Choleraesuis

infection occurs typically at around 12 to 14 weeks. Disease is dose dependent, that is, a

relatively large number of organisms are required before clinical signs occur. The Salmonella

serotypes that are most likely to cause clinical disease in pigs are S. Choleraesuis, and

S. Typhimurium and, to a lesser extent, S. Derby. Other "exotic" salmonella serotypes may

infect pigs and be shed in the faeces for limited periods but they usually remain sub-clinical.

They may be shed in faeces for several weeks or months with no clinical disease.

S. Choleraesuis and S. Derby may be carried for long periods by sows, the former

sometimes causes clinical disease in sows (fever, depression, septicaemia, pneumonia,

meningitis arthritis and diarrhoea). They multiply mainly in the intestines of young growing

pigs but also in some sows. Pigs may become long-term sub-clinical carriers, the organisms

surviving in the mesenteric lymph nodes draining the intestine. Many such carriers do not

shed the bacteria in faeces unless they are stressed. Pigs may be intermittent or continuous

faecal shedders of other serotypes but the carrier state is usually short, weeks or a few

months and is self limiting. S. Typhimurium and S. Derby are more likely to cause milder

disease, the main sign of which is usually diarrhoea.

Yersinia spp.

In humans, yersiniosis is usually self-limiting and does not require treatment, although

antimicrobial therapy may be prescribed. Most human illness is caused by Y. enterocolitica.

Y. enterocolitica is a relatively infrequent cause of diarrhoea and abdominal pain. Clinical

signs vary with age and immune system status. Infection with Y. enterocolitica occurs most

http://www.hpa.org.uk/webc/HPAwebFile/HPAweb_C/1259151943662



often in young children where common clinical signs in children are fever, abdominal pain,

and diarrhoea, which is often bloody. They typically develop four to seven days after

exposure and may last one to three weeks or longer. In older children and adults, right-sided

abdominal pain and fever may be the predominant signs, and may be confused with

appendicitis. In a small proportion of cases, complications such as skin rash, joint pains, or

spread of bacteria to the bloodstream can occur.

(http://www.cdc.gov/ncidod/dbmd/diseaseinfo/yersinia_g.htm). Infection is more common in

the winter.

In pigs, asymptomatic carriage in the tonsils is high. Clinical disease in pigs due to Yersinia

infection is extremely rare. As a cause of clinical disease it is not listed as a Veterinary

Investigation Disease Analysis (VIDA) code for Animal Health and Veterinary Laboratories

Agency (AHVLA)

Total aerobic and Enterobacteriacae counts

These do not constitute hazards in their own right. Rather they are measures of the hygiene

process on the slaughter line and the potential for cross-contamination to occur. Such cross-

contamination could have implications for the risk present, in terms of increased numbers of

carcases affected, for other microbiological agents. As such an indicator they cannot be

independently characterised.

Exposure Assessment

Exposure assessment estimates the level of microbiological pathogens or toxins and the

likelihood of their occurrence in foods at the time of consumption (CAC, 1999).

For public health hazards, whether they arise from animal health conditions or contaminants

such as microbiological organisms, the extent of actual or anticipated human exposure will

only change if the risk profile of the final (chilled) pork carcass (FCPC) changes significantly

and if that change is not combated by existing measures in the food chain after the FCPC

stage.

The exposure measured by the prevalence of each agent or condition on or in carcases at

the FCPC stage is a proxy measure for human exposure from pork meat, whereas the actual

exposure is that just prior to consumption. In FS145003 we are measuring the observed

frequency slightly earlier in the risk pathway close to the inspection points. We are assuming

that:

any further steps in the immediate pathway to the FCPC stage are independent of

the inspection method and would not result in a significant change (reduction or

increase) in prevalence of each agent or condition on or in carcases;

all processes from here up to immediately prior to consumption remain constant, i.e.

are independent of the inspection method;

the frequency of these conditions or agents found during traditional inspection of

outdoor pigs constitutes an acceptable exposure.

http://www.cdc.gov/ncidod/dbmd/diseaseinfo/yersinia_g.htm


What is the current baseline exposure assessment with the traditional inspection

method?


endocarditis and granulomatous lesions in pork meat?


The literature for Streptococcus suis was reviewed by Hill et al., (2011). It is found in pigs in

the United Kingdom (UK). Human infection with S. suis is rarely reported and only about 150

cases have been reported from the world literature. S. suis infection is rare among humans

in England and Wales with only two to three human cases being reported each year. The

last fatal case in the UK occurred in a farm worker in 1999, due to S. suis type 14. During

2004 a total of 112 isolates were found in pigs in England and Wales, of which 46% (51)

were S. suis type 2, the serotype most commonly associated with human infection

(http://www.hpa.org.uk/Topics/InfectiousDiseases/InfectionsAZ/StreptococcusSuis/GeneralIn

formation/). People in direct contact with pigs or pig products are considered at risk. Human

infection is thought to occur mainly via cuts or abrasions when handling infected carcases.

The main transmission pathways for S. suis were not considered to be foodborne (Anon.,

2000). The bacteria are unable to survive proper cooking processes (Leps and Fries, 2009),

so the contribution to human exposure made by pork meat is considered to be negligible.


The domestic pig is the most important reservoir of E. rhusiopathiae, the causal agent of

swine erysipelas, although it does cause disease in other species (Wang et al., 2010).

Humans become infected through exposure to infected or contaminated animals or animal

products and it is considered predominantly as a disease of occupational exposure. The

main transmission pathways for E. rhusiopathiae were not considered to be foodborne

(Anon., 2000). The people with the highest risk of exposure include butchers, abattoir

workers, veterinarians, farmers, fishermen, and fish-handlers (Reboli and Farr, 1989). It is

estimated that 30–50% of healthy swine harbour the organism in their tonsils and other

lymphoid tissues. However, given that between 2001 and 2006 the Health Protection Agency

(HPA 2007) reported only nine cases of bacteraemia in England, Wales and Northern

Ireland caused by this ‗uncommon pathogen‘ and subsequently the most recent report

covering 2006 to 2010 (HPA, 2011a) reports twenty cases {from one to seven/year), the

contribution to human exposure made by pork meat is considered to be negligible.


The primary ‗at risk‘ group of humans is immunocompromised people, such as those with

HIV-AIDS or transplant recipients. However, given that between 2001 and 2006 the Health

Protection Agency (HPA 2007) reported only three cases in England, Wales and Northern

Ireland of bacteraemia caused by this ‗uncommon pathogen‘ [one in 2001 and two in 2004],

and subsequently the most recent report covering 2006 to 2010 (HPA, 2011a) reports two

cases in 2008 and one case in 2009, the contribution to human exposure made by pork

meat is considered to be negligible.

Granulomatous lesions - Mycobaterium spp.:

M. bovis was the only Mycobacteria spp. identified by Hill et al. (2011) as being of a potential

risk to humans from pigs. It has not been thought necessary to re-evaluate this, given the

short time span since the completion of their work. From the enhanced TB surveillance data

for the countries of the UK




http://www.hpa.org.uk/Topics/InfectiousDiseases/InfectionsAZ/Tuberculosis/TBUKSurveillan

ceData/EnhancedTuberculosisSurveillance/TBEnhanced01country/ the number of human

case reports for the UK in 2010 was 8,482, a rate of 13.6 (95% C.I. 13.3 – 13.9) per 100,000

compared with 8,917 [14.4 (95% C.I. 14.1 – 14.7) per 100,000] for 2009. In 2009 there were

35 isolates of M.bovis in humans with 23 of these being in people over 65 years of age. The

likely contribution of pigs to this number of cases is negligible (Hill et al., 2011). This is re-

inforced by the fact that only 29 porcine tuberculosis cases have been found in eight million

carcases inspected in 2010 (Food Standards Agency (FSA) personal communication). In

addition, in a 13 year study only one human M. bovis case was determined to be acquired

from an animal source in the United Kingdom (de la Rua-Domenech et al., 2006).

Endocarditis and granulomatous lesions public health exposure assessment

summary

The food-borne contribution made by pork meat to human exposure to Streptococcus spp.

including S. suis is considered to be negligible.

The food-borne contribution made by pork meat to human exposure to E. rhusiopathiae is

considered to be negligible.

The food-borne contribution made by pork meat to human exposure to R. equi is considered

to be negligible.

The likely contribution of pigs to the number of cases of M. bovis in humans is negligible.


carcass microbial contamination of pork meat?

Salmonella spp.

Salmonella is a major cause of food-borne illness in humans. In 2010, 9,133 human isolates

were reported to the HPA centre for Infections in England and Wales. This is a substantial

decrease from the 17,163 reported in 2001. Of these isolates, 27% were S. Enteritidis, 21%

(1959) were S. Typhimurium, 6% were typhoidal salmonellas and 46% were ‗other

serotypes‘. From the 2009 reported human isolates the rate per region varied from 11.3 –

25.3 per 100,000 of the population.

(http://www.hpa.org.uk/Topics/InfectiousDiseases/InfectionsAZ/Salmonella/). As with the

majority of self-limiting diseases, it is likely that this figure under-represents the total burden

of illness attributable to Salmonella spp. within the population. Farm animals and foods of

animal origin are important sources of human Salmonella infections.

How much of the human cases can be attributed to foodborne exposure from pork is not

easy to quantify it is likely to be a proportion of the S. Typhimurium cases and some of the

‗other serotypes.‘ Hill et al., (2003) estimated that the risk of illness from S. Typhimurium

originating from pig-meat (whether via cross-contamination from hands or under-cooking) is

low. The expected number of cases of S. Typhimurium originating from pig-meat per year

was estimated from the mean risk of illness per serving for each pig-meat product by each of

the two exposure routes (i.e. undercooking and cross-contamination) and the level of

consumption of those products. This gave an estimate of 1,687 cases per year in England

and Wales. When they compared the estimated expected number of cases to the actual

number of reported cases in England and Wales in 2001 (and adjusted for under-reporting

using data from the UK IID Study), it suggested that approximately 30% of all

http://www.hpa.org.uk/Topics/InfectiousDiseases/InfectionsAZ/Tuberculosis/TBUKSurveillanceData/EnhancedTuberculosisSurveillance/TBEnhanced01country/

http://www.hpa.org.uk/Topics/InfectiousDiseases/InfectionsAZ/Tuberculosis/TBUKSurveillanceData/EnhancedTuberculosisSurveillance/TBEnhanced01country/



S. Typhimurium cases are attributable to pig-meat. However, this estimate should be treated

with caution since good quality consumption data were not available. As 30% of 21% is

approximately 6-7% of the total number, if applied to the rates from 2009 this would give

rates of from less than one to two per 100,000 of the population i.e. risk is low.

The Biological Hazard Panel has assessed the public health risks from Salmonella in pigs

and the impact of possible control measures. The assessment suggested that pigs and pork

may be responsible for 10 % to 20 % of all human cases of salmonellosis in the EU in 2009,

but with differences between countries (EFSA, 2010b).

Y. enterocolitica*

Pigs are considered to be the major animal reservoir for Yersinia strains that cause human

illness, even if other animal species, e.g. cattle, sheep, deer, small rodents, cats and dogs

may also carry pathogenic serotypes. Infection can be acquired from contaminated food and

water (organisms can multiply in food at 4°C).Transmission pathways include direct contact

with infected animals; person to person spread and there is a particular association with raw

pork and pork products. Preventive advice includes the avoidance of raw or undercooked

pork

http://www.hpa.org.uk/Topics/InfectiousDiseases/InfectionsAZ/Yersinia/GeneralInformation/.

Y. enterocolitica is listed as an uncommon pathogen involved in bacteraemia by the HPA

and was isolated from 6 to 15 cases per year between 2001 and 2006 in England, Wales

and Northern Ireland (HPA, 2007). Y. frederiksenii, Y. intermedia, Y. pseudotuberculosis and

Y. rohdei have also been isolated on occasion. During this period (2001-2006) the HPA

recorded between 11 and 34 laboratory reports in England and Wales per year of

Y. enterocolitica. This figure was 32 in 2009 with a provisional figure of 19 reports for 2010.

(http://www.hpa.org.uk/Topics/InfectiousDiseases/InfectionsAZ/Yersinia/)

Based on data from the Foodborne Diseases Active Surveillance Network (FoodNet), which

measures the burden and sources of specific diseases over time, approximately one culture-

confirmed Y. enterocolitica infection per 100,000 head of population occurs each year.

(http://www.cdc.gov/ncidod/dbmd/diseaseinfo/yersinia_g.htm).

In the EU in 2007, yersiniosis was the third most frequently reported human zoonosis with a

total of 8,792 confirmed cases (EFSA, 2009a) with a notification rate of 2.8/100,000 per

head of population. Y. enterocolitica was the most common species reported in human

cases by MSs and was isolated from 93.8% of all confirmed cases. By 2010, there was a

total of 6,776 confirmed cases (EFSA, 2012) with a notification rate of 1.58/100,000 per

head of population and Y. enterocolitica was isolated from 91.0% of all confirmed cases.

Fosse et al., (2008) estimated that the mean rate of human clinical cases in Europe per

annum attributable to Y. enterocolitica was 2.8/100,000 per head of population per annum

with a median of 0.943/100,000. The likely contribution of pigs to this number of cases is not

quantified, and we consider it to be high, even though the incidence of human cases in the

UK is low.

The majority of surveys in pigs at the abattoir determine the prevalence in tonsils; however

Gürtler et al., (2005) also investigated iliocaecal lymph nodes and carcases. In a population

of fattening pigs where the prevalence of Y. enterocolitica in tonsils at the abattoir was

http://www.hpa.org.uk/Topics/InfectiousDiseases/InfectionsAZ/Yersinia/GeneralInformation/

http://www.hpa.org.uk/Topics/InfectiousDiseases/InfectionsAZ/Yersinia/

http://www.cdc.gov/ncidod/dbmd/diseaseinfo/yersinia_g.htm


38.4%, the prevalence on the carcass surfaces before chilling was 0.3% and this fell to 0%

after chilling.

Total aerobic and Enterobacteriacae counts

These do not constitute hazards in their own right. Rather they are measures of the hygiene

process on the slaughter line and the potential for cross-contamination to occur. Such cross-

contamination could have implications for the risk present, in terms of increased numbers of

carcases affected, for other microbiological agents. As such an indicator, a baseline

exposure assessment cannot be independently made.

Microbiological public health exposure assessment summary

The risk of illness from S. Typhimurium originating from pig-meat (whether via cross-

contamination from hands or under-cooking) is low.

The likely contribution of pigs to the number of human cases of yersiniosis is not quantified,

and we consider it to be high, even though the incidence of human cases in the UK is low.

What is the baseline for animal health? i.e. disease prevention and control?

What contribution does information derived from post-mortem inspection data have in the

prevention and control of disease in the outdoor pig population?

Endocarditis – general

Endocarditis is not scored or reported back as a condition to producers in either the British

Pig Executive (BPEX) British Pig Health scheme (BPHS) or the Wholesome Pigs Scotland

(WPS) scheme.


During 2004 a total of 112 isolates were found in pigs in England and Wales, of which 46%

(51) were S. suis type 2, the serotype most commonly associated with human infection

(http://www.hpa.org.uk/Topics/InfectiousDiseases/InfectionsAZ/StreptococcusSuis/GeneralIn

formation/). In the VIDA report of yearly trends for pigs between 52 to 92 cases of S. suis

were diagnosed per year from laboratory submissions in GB in the period 2004 to 2011

(http://vla.defra.gov.uk/reports/docs/rep_vida_pigs04_11.pdf).


It is estimated that 30–50% of healthy swine harbour the organism in their tonsils and other

lymphoid tissues. In the VIDA report of yearly trends for pigs between four and 21 cases of

E. rhusiopathiae were diagnosed per year from laboratory submissions in GB in the period

2004 to 2011. (http://vla.defra.gov.uk/reports/docs/rep_vida_pigs04_11.pdf).

Outbreaks of death due to endocarditis has caused 15% mortality in a population of 500

growing pigs (http://www.nadis.org.uk/bulletins/erysipelas.aspx).

Granulomatous lesions - General

Granulomatous lesions are not scored or reported back as a condition to producers in either

the BPHS or WPS scheme. In the historical analyses of FSA data from the study abattoir

such lesions were not classified as a separate category, therefore, if they did occur, they

would be part of the ‗other‘ category. No cases were recorded in the study period itself by

either method of inspection. Granulomatous lesions per se are not considered as a recorded

condition in post-mortem inspection of animals at abattoirs in UK. They are lesions



http://vla.defra.gov.uk/reports/docs/rep_vida_pigs04_11.pdf


http://www.nadis.org.uk/bulletins/erysipelas.aspx


considered by MHIs to help them reach a diagnosis (e.g. TB) and are not recorded in any

official system. If such a diagnosis is reached then the information is used by the authorities

as sentinels for TB in cattle. Cattle herds on and around the pig farm will be tested.

Granulomatous lesions - Rhodoccus equi

In the VIDA report of yearly trends for pigs, no cases of Rhodococcus equi were diagnosed

per year from laboratory submissions in GB in the period 2004 to 2011.

(http://vla.defra.gov.uk/reports/docs/rep_vida_pigs04_11.pdf).

Granulomatous lesions - Mycobaterium spp:

In the VIDA report of yearly trends for pigs, between zero and three cases of tuberculosis

were diagnosed per year from laboratory submissions in GB in the period 2004 to 2011.

(http://vla.defra.gov.uk/reports/docs/rep_vida_pigs04_11.pdf). Only 29 porcine tuberculosis

cases have been found in eight million carcases inspected in 2010 (FSA personal

communication). However, there is no information available about the actual prevalence in

the pig population

Salmonella spp.

An EU-wide Salmonella baseline survey was conducted in 2008. Analyses at country-level

demonstrated a strong positive association between the prevalence of Salmonella-positive

breeding holdings and the prevalence of Salmonella-positive production holdings,

suggesting a vertical dissemination of Salmonella between the holdings. The overall EU

apparent prevalence of Salmonella-positive holdings with breeding pigs was 31.8% and

33.3% of the production holdings were estimated to be positive for Salmonella. This

prevalence varied from 0% to 55.7% among the Member States. The estimated EU

prevalence of production holdings positive for S. Typhimurium and S. Derby was 6.6% and

9.0%, respectively. In the UK the prevalence of Salmonella-positive production holdings was

44% (95% C.I. 37.8-50.9%, n=191) with a prevalence of 6.6% (5.3-7.9%) for S. Typimurium

and 11% (95% C.I. 5.3-7.9%) for S. Derby, respectively (EFSA, 2009b).

In the VIDA report of yearly trends for pigs, between 67 and 120 cases of salmonellosis due

to S. Typhimurium were diagnosed per year from laboratory submissions in GB in the period

2004 to 2011. This compares to between zero and one cases of salmonellosis due to

S. Choleraesuis and 20 to 39 cases of salmonellosis due to Salmonellae not otherwise

specified (http://vla.defra.gov.uk/reports/docs/rep_vida_pigs04_11.pdf).

Y. enterocolitica*

From 2003 to 2005 the prevalence in the tonsils of 630 pigs from 45 farms in England of

pathogenic Y. enterocolitica was 44% and Y. pseudotuberculosis was 18% with 60% of pigs

carrying enteropathogenic Yersinia. Y. enterocolitica was detected on 69% of farms and

Y. pseudotuberculosis on 78%. The prevalence of each Yersinia spp. varied by season,

region and type of management system (Ortiz Martinez et al., 2010). The prevalence of

pathogenic Y. enterocolitica has also been found to vary between countries from 93% to

32% (Ortiz Martinez P., 2010; Ortiz Martinez et al., 2011). Clinical disease in animal

reservoirs is rare. As a cause of clinical disease it is not listed as a VIDA code for AHVLA

laboratories.





In a report from EU Member States (EFSA and ECDC, 2011a), on average, 4.8 % of pig

meat units were found positive for Y. enterocolitica in the reporting group and a high

prevalence was reported by two Member States in slaughter batches of pigs.

Risk Characterisation

Risk characterisation is the integration of hazard identification, hazard characterisation and

exposure assessment to obtain a risk estimate (CAC, 1999).

Risk in a qualitative, or semi-qualitative, assessment is usually expressed in categories. In

this assessment we use a modified system from Hill et al., (2011) and define the categories

as follows:

Negligible Risk or the frequency/outcome is so low as to not merit consideration

Very low Risk or the frequency/outcome is almost negligible but cannot be excluded from consideration due to uncertainty or other extenuating circumstances

Low Risk or the frequency/outcome is small/infrequent but still worth considering intervention/mitigation

Medium Occurs frequently or associated with a modest outcome

High Occurs often and/or associated with a significant outcome

Very high Almost certain to occur and/or associated with a serious outcome

Here we characterise the baseline risk with the current traditional method of inspection for

both public health and animal health (Tables A4:11 & A4:12) and then examine if that risk

profile is likely to alter if the inspection method is visual-only as implemented in our

study/field trial.

Is there a change in the risk profile if the inspection method is changed from

traditional to visual-only?

As stated at the beginning, there are two main criteria that will determine whether the risk

profile will change with the inspection method. These are:

1. Whether the sensitivity of detecting a condition is affected by the inspection method;

if not, then there will be no change in risk;

2. Whether the level of carcass contamination is affected by the inspection method; if

not, then there will be no change in risk.


From study FS145003 we have the frequencies of the hazards by the inspection method

(Table A4:8).

Table A4:8: The descriptive analyses of the frequencies (number=n [%] and where appropriate 95% confidence interval, C.I.) found in the study abattoir in FS145003 by the two inspection methods for the hazards identified.

Hazard

Identified

Frequency n [%] and 95% C.I. by inspection method

Traditional Visual

Endocarditis (1PAH) 21 [0.2]

0.05 - 0.33

0 [0]

Granulomatous lesions to include porcine tuberculosis lesions (1PAH)

0 0

Salmonella spp. (2A-PHCM) 0 0

Yersinia spp.* (2A-PHCM) 7 [1.8]

0.8 - 3.8

5 [1.3]

0.5 - 3.1

Enterobacteriaceae 120 [30]

26 – 35

115 [29]

25 – 33

Table A4:9: Student t-test for the comparison between the two inspection methods for Enterobacteriaceae counts where Enterobacteriaceae = presence and for the log10 of total aerobic plate counts.


mean (95%CI)

P value

Enterobacteriaceae counts where

Enterobacteriaceae = present

Traditional -1.14 0.43 (0.22, 0.63) <0.001

Visual -1.57

log 10 of total aerobic plate count

Traditional 1.497 0.08 (-0.06, 0.21) 0.2772

Visual 1.421

How many pigs from non-controlled housing conditions are currently processed annually in

Great Britain?

It is estimated that approximately 120 commercial herds supply abattoirs with pigs finished

outdoors in the UK, with an average of 15 batches a year and approximately 220 pigs per

batch (Quality Meat Scotland (QMS), personal communication). This would lead to an

estimate of approximately 400,000 ‗outdoor finished‘ pigs processed per year. Virtually all

outdoor pig finishing units are located in England. If we assume that these figures above

have some degree of accuracy then the 11,086 trial pigs form approximately 3% of the total

annual British outdoor finished pig throughput. Hill et al., (2011) estimated that 5-10% of the

total UK pig production is reared outdoors after weaning. Slaughter statistics indicate that 7,

312, 000 head of pig were killed in 2010 (Lewis, 2010). Not all of these would be fattening

pigs, however the number of boars and sows killed is no longer available. Based on previous


years it would be between 100,000-200,000. This would give an estimate of 350,000 to

700,000 for the total annual British outdoor finished pig population.

If we assume:

700,000 ‗outdoor finished‘ pigs processed are per year;

that the frequency of the conditions found in the trial pigs is the same as in the

general population of British outdoor finished pigs (a huge assumption);

that a total ‗hand-off‘ visual-only method of inspection were implemented with a

similar visual access and line speeds to the trial abattoir, at a similar point on the line

with respect to carcass trimming and dressing procedures abattoirs;

then, based on the observed overall frequency from the study (Table A4:10), there is a

potential for the total number of undetected hearts/carcases to be as follows:

Table A4:10: The predicted guesstimates of the total annual numbers of hearts/carcases potentially missed if visual-only inspection was implemented as per the trial conditions and prevalence in all ‗outdoor finished‘ pigs was the same as per the trial conditions.

Hazard

Identified

Total number of cases potentially detected/missed by inspection method per year

Traditional Visual Potentially Missed

Endocarditis (1PAH) 1400

350 - 2310

0 1400

350 - 2310


2.54* 0 2.54*


20^ 10^ 10^

*based on 29 per 8M ^based on figures from Hill et al., 2011 for both number of cases and visual detection

The number of granulomatous lesions to include porcine tuberculosis lesions cases

potentially missed is not of an order that would change the risk profile for either public or

animal health. The number of endocarditis cases potentially missed is also not of an order

that would significantly change the risk profile for public health. The public health risk arises

from the systemic bacteraemia for which endocarditic lesions are an indicator. Systemic

involvement should be detected by visual inspection and endocarditis may be observed in

carcases in combination with other lesions that may be detected on visual inspection, thus it

is only a proportion of the endocarditis lesions potentially missed that would lead to a

carcass that would have been heat treated entering the food chain (i.e. a proportion of

1400). For example in FAO (2010), if there is evidence from the carcass that the pig had a

fever it should be condemned. Carcases with ulcerative and verrucose endocarditis with no

systemic changes may be conditionally approved and then heat treated but the heart is

condemned. If the endocarditis lesions involve fibrous tissue infiltration then the carcass can

be approved and the heart only is condemned. We do not have any data on what this

proportion might be in the UK with the current traditional inspection. We have, therefore,

taken a cautious approach, assumed that it would be 100% and designated the associated


risk as non-negligible, despite the numbers above being a worst case scenario (Tables

A4:10 & A4:11).

The number of endocarditis cases potentially missed is also not considered to be of an order

that would significantly change the risk profile for animal health. While it would have some

impact an impact on the availability of relevant pathological data to feed back to the

producers there is no evidence that currently this is either an issue of importance or that they

take action to mitigate or control the causes of endocarditis in their herds after receiving

such information. This may be due to the apparent low level of occurrence and a

corresponding lack of industry concern. In addition, both S. suis and E. rhusiopathiae cause

other clinical problems in pigs. We would not miss an increase in their prevalence if we don‘t

have endocarditis as an indicator, because an increase in the prevalence of other conditions

would act as indicators. In the historical analysis of data from the study abattoir only 3.5%

(95% C.I. 2.6 – 4.4%) of batches of free range pigs were affected with endocarditis and the

mean prevalence in the batches in which the condition was present was low at 0.8% (95%

C.I. 0.6 – 0.9%) However, we have taken a cautious approach, despite the numbers above

being a worst case scenario (Tables A4:10 & A4:12). In our field study the apparent

prevalence of endocarditis by traditional inspection at a batch level (i.e. percentage of

batches affected with the condition) was significantly higher (19.3% 95% C.I. 10.8 – 31.7)

than in the historical data analysis (3.5%, 95% C.I. 2.6 – 4.4%), although the within batch

mean prevalence was similar (1.1, 95% C.I. 0.4 – 4.2 compared to (0.8% (95% C.I. 0.6 –

0.9%). The wider 95% C.I. are due to the smaller sample size of the field study.


Table A4:11: Summary of the Risk Characterisation for Public Health for the five hazards for which detection could be or was affected by the inspection method

Identified Hazard Hazard Characterisation

[Occurrence – Severity – Fatality - Treatment]

Exposure Assessment

Amount attributable

Risk characterisation

(Baseline – current – traditional inspection)

Change in risk profile if inspection method changed to visual

Revised risk

(for visual-only inspection)

Human Human cases Public health Change Relevance Public health

Endocarditis

Streptococcus spp.

Extremely Rare – Severe - Extremely rare – Possible

NEGLIGIBLE NEGLIGIBLE Yes Unlikely VERY LOW

Endocarditis

Erysipelothrix rhusiopathiae.

Rare (occupational) – Generally mild – Extremely rare – Possible

NEGLIGIBLE NEGLIGIBLE Yes Unlikely VERY LOW


Extremely rare (opportunistic) – Severe – Likely – a challenge

NEGLIGIBLE NEGLIGIBLE Possible Highly unlikely

NEGLIGIBLE


Extremely rare (opportunistic) – Severe – Likely – Possible, can be a challenge

NEGLIGIBLE NEGLIGIBLE Possible Highly unlikely

NEGLIGIBLE

Salmonella spp. Not uncommon - Mild to moderate – Unusual – Possible

LOW VERY LOW No N/A VERY LOW

Y.enterocolitica Uncommon – Mild to Moderate – Highly Unlikely – Self-limiting/possible

MEDIUM LOW No N/A LOW

Total aerobic and Enterobacteriaceae counts

Indicator of hygiene process rather than direct clinical relevance

N/A N/A Yes Possible Potential for reduced contamination of carcases


Table A4:12: Summary of the Risk Characterisation for Animal Health (AH) for the two major hazards for which detection could be or was affected by the inspection method

Identified Hazard Hazard Characterisation

[Occurrence – Severity – Fatality - Treatment]

Exposure Assessment

Amount attributable

Risk characterisation

(Baseline – current – traditional inspection)

Change in risk profile if inspection method changed to visual

Revised risk

(for visual-only inspection)

Pigs Animal cases Animal health

Change Relevance AH

Endocarditis

Streptococcus spp.

Mostly asymptomatic carriage – Varies – Possible –Possible

NEGLIGIBLE NEGLIGIBLE Yes Highly unlikely

VERY LOW

Endocarditis

Erysipelothrix rhusiopathiae.

Mostly asymptomatic carriage – Mild to severe – Can be high - Possible

NEGLIGIBLE NEGLIGIBLE Yes Highly unlikely

VERY LOW


Asymptomatic – N/A NEGLIGIBLE NEGLIGIBLE Possible Extremely unlikely

NEGLIGIBLE


Asymptomatic – N/A NEGLIGIBLE NEGLIGIBLE Possible Extremely unlikely

NEGLIGIBLE


Risk assessment discussion

We have used a modified CAC risk assessment approach to assess the potential change in

risks to human (public health via a food-borne route), animal health and animal welfare of a

change in the meat inspection method, from the traditional method currently employed to a

visual-only (‗hands-off‘) methodology, for fattening pigs from non-controlled housing

management systems i.e. raised outdoors from weaning to slaughter. We have used data

from previous work, scientific literature, publically available information and our own field

study to inform the risk assessment. Based on this information we have taken a cautious

approach and considered a worse case scenario. Of the five public health hazards we have

assessed (endocarditis, granulomatous lesions, Salmonella spp., Yersinia spp., total aerobic

and Enterobacteriacae counts), only two have a revised risk on a change in inspection

method. One, endocarditis, changes from negligible to non-negligible i.e. very low, while for

the other (total aerobic and Enterobacteriacae count) it is possible that the risk of cross-

contamination between carcases is reduced.

Hill et al., (2011) determined that the risk of transferring to a visual-only system was

negligible for all pigs. They rated the risk from endocarditis to be negligible, as in their

estimate around 300-400 hearts with endocarditis would be missed per year from all pigs

both indoor and outdoor (controlled and non-controlled housing conditions) over and above

those already missed by traditional meat inspection methods. Our estimate just for outdoor

pigs is of the order of three to four times that. This may be a reflection of the uncertainty in

the estimates in both pieces of work. There is also a finding from our study data that cannot

definitively be explained: endocarditis was observed by the traditional method in our study in

almost six times as many batches when compared to the free-range pigs in the historical

data analysis from the study abattoir. This did not have a statistically significant farm

component but due to the small number of farms (12) that contributed to our study some

bias due to a ‗farm‘ influence may still exist. We cannot completely rule out the possibility of

an inspection/study bias as it was not possible to blind the inspection teams. It is however

unlikely because to make such a marked effect above the norm when there was the usual

frequent rotation of the duty on study days this would have had to apply to all those doing

the traditional inspection. There is no known seasonality for endocarditis lesions. We,

therefore, have no clear explanation why this finding was of this magnitude but it

subsequently inflates our estimate of endocarditis lesions missed per year compared to that

of Hill et al., (2011).

Despite the ‗very low‘ revised risk classification for public health attributable to endocarditis

and the slight increase in risk for visual ‗hands-off‘ inspection compared to traditional

inspection for outdoor pigs, the fact still remains that outdoor pigs from non-controlled

housing conditions present at least the same, if not less (Hill et al., 2011), of a risk than

indoor pigs from controlled housing conditions. This is supported by the findings from the

historical data analysis from the study abattoir. Here, although the differences were not

statistically significant, we found that the mean prevalence in the batches in which the

condition was present for indoor pigs was at the higher end of the values for free range pigs

(1.0%, 95% C.I. 0.9-1.1% compared to 0.8%, 5% C.I. 0.6 – 0.9%), and the percentage of the

batches affected with the condition was higher (5.2%, 95% C.I. 4.6 – 5.8% compared with

3.5%, 2.6 – 4.4%). Visual inspection is acceptable for pigs from the former type of

management systems (Anon., 2004b); therefore there is no reason relevant to the public

health risk presented to exclude the latter purely on grounds of the management system


from which they originate. This is also the case for animal health risk. Action by producers is

unlikely to be taken on the basis of information received about endocarditis lesions from

post-mortem data feedback. For these causal agents, action would be taken in response to

clinical signs in live pigs with associated production losses and their economic impact.

One of the arguments for a move from a traditional palpation and incision inspection system

to a visual-only based one is that it could reduce cross contamination of carcases that would

occur via the hands and knives of meat inspectors. This, however, is an area where

evidence is sparse and it is difficult to draw robust conclusions. Mousing et al., (1997)

suggested that the increase in the number of carcases not recorded as ‗faecal

contamination‘ in a visual-only system would lead to an increase in the number of carcases

not detected that carried Yersinia spp. or Salmonella spp. (based on extrapolation from

microbiology of gut contents) although this would be outweighed by the reduction in cross-

contamination from hazardous bacteria, particularly from the pharyngeal and pluck region.

Hamilton et al., (2002) found that on microbiological examination unseen contamination on

pre-chill carcases was at similar levels with both inspection methods. The outcomes from

our field study demonstrated that although there was a statistical difference in the number of

faecal contaminated carcases recorded by each inspection method, the mean of the

differences was very small (1.62% (95% C.I. 0.8 – 2.44%) and it was highly unlikely that this

would be of clinical significance. This was confirmed by the microbiological investigations.

No difference was found in the isolation of Yersinia spp. or Salmonella spp. total aerobic

count or the presence/absence of Enterobacteriacae; however, when present the

Enterobacteriacae count was lower on carcases that had been visually inspected than

traditionally inspected, implying less contamination. The abattoir used for the field study had

a particularly good hygiene process with considerable attention being paid to procedures on

the line to ensure this. This was the norm and not due to the presence of the study team. It is

possible that a change in the inspection method from traditional to visual would lead to a

similar result in any abattoir with a level of contamination as low as or higher than the study

premises. If the level of contamination is lower, then it could be hypothesised that the

potential for cross-contamination would be lower; however, we cannot draw that as a

conclusion from our study.

In this risk assessment we have not considered the component of public health risk that is

due to occupational exposure. This would be particularly relevant for those organisms that

are considered to be occupational hazards, i.e. S. suis and E. rhusiopathiae both of which

could be contracted by those handling and incising the hearts to examine them for

endocarditis lesions. It is theoretically possible that by moving to visual-only inspection it

would eliminate the exposure of meat inspectors to these risks; however, this is likely to

have a negligible effect on public health given the current extremely small number of clinical

cases in humans in the UK.

It is beyond the remit of this study to assess if the risk assigned to Salmonella spp. and

Yersinia spp. within this risk assessment is either acceptable or comparable with that from

fattening pigs from controlled housing conditions. References are cited in (EFSA, 2011c) that

indicate that for Yersinia, organic farming systems may have a lower within-herd prevalence

despite herd level prevalence being similar to that of conventional production units. We have

not determined the pathogenicity of the Yersinia spp. found in our field study.


It has been considered that finishers from integrated production systems that are kept

indoors from weaning will have less variation in disease pattern than finished pigs from other

types of production (Alban et al., 2008). So our study pigs were fattening pigs raised entirely

outdoors from weaning to slaughter. As far as the authors are aware, all such commercial

fattening pigs in GB are also born in non-controlled housing conditions. There are, however,

fattening pigs in the UK that are raised outdoors (non-controlled) and enter controlled

housing conditions at some point, to be finished before they are sent to the abattoir. We

have assumed in the design of the study, that fattening pigs raised entirely outdoors from

weaning to slaughter would be more likely to be different to pigs raised totally indoors than

fattening pigs raised partially outdoors and partially indoors; therefore they would potentially

present the greatest risk if inspection systems were to be changed. It could be argued that

this is not the case and that fattening pigs raised partially outdoors and partially indoors

experience the worst, rather than the best of both worlds in terms of disease status; however

the authors are not aware of any data with this level of information on origin that are

available with which to make such an assessment. This became apparent in the course of

sourcing appropriate premises in which to undertake the study. However, given the

outcomes of this risk assessment and the one by Hill et al., (2011) fattening pigs from such

outdoor-indoor/combined management systems would have to be significantly different in

terms of disease status order to present an appreciable risk.

Risk assessment conclusions

1. Five hazards that are related to public health and might be affected by the inspection

method were identified - endocarditis, granulomatous lesions, Salmonella spp. and

Yersinia spp. and the hygiene process indicators - total aerobic and

Enterobacteriaceae count.

2. The risk to public health associated with one of these hazards, endocarditis, alters

with a change in inspection method.

3. The risk to public health from endocarditis in outdoor pigs from non-controlled housing

conditions inspected by a visual-only method is higher relative to the same pigs

inspected by the traditional method (very low: negligible). The absolute risk to public

health from endocarditis in outdoor pigs from non-controlled housing conditions

inspected by a visual-only method is negligible.

4. There is some evidence from the hygiene process indicators - total aerobic and

Enterobacteriaceae count – that the visual-only method results in a lower level of

contamination of carcases than inspection by the traditional method.

5. Two hazards that are related to animal heath and subsequently their welfare that

might be affected by the inspection method were identified – endocarditis and

granulomatous lesions.

6. The risk to animal health and welfare associated with neither of these two hazards

alters with a change in inspection method.


Risk mitigation measures

From this risk assessment, based on current evidence, we do not consider that there is any



that which currently exists with traditional inspection. However, we have considered potential

mitigation measures that could be considered, if it is perceived that further additional

measures are necessary to manage any residual risk.

If endocarditis lesions are perceived to represent a residual risk then arrangements could be

made to open hearts for inspection, where it is thought to be necessary.

If granulomatous lesions are perceived to represent a residual risk then arrangements could

also be made to incise relevant lymph nodes for inspection, where it is thought to be

necessary.

The difference in the frequencies of faecal contamination recorded between the visual and

traditional inspection methods found in our field study may be explained by the positioning of

the visual-only inspection point on the line and the inability to turn the carcass by the total

―hands off‖ method used. This hypothesis is supported as there were many occurrences of

―faeces contaminations in any part except tail‖ that were recorded by the MHI at the visual

inspection point, whereas in the traditional inspection recordings they were mainly of ―faeces

contamination in tail‖. The visual inspection point was placed earlier in the line than the

traditional inspection point, which was at its usual position. Between the two there was the

opportunity for plant staff to trim the carcass. Furthermore, it is unlikely that a total hands-off

visual-only inspection, even in the usual inspection position on the line, will allow all surfaces

of the carcases to be visible. Measures such as strategically placed mirrors or systems that

rotate carcases would be needed to facilitate this.

Harmonised epidemiological indicators have been identified and have been proposed for use

for food-borne hazards to public health related to pigs and pork such as Salmonella and

Y. enterocolitica (EFSA 2011c). These should form the basis for any further risk mitigation

measures, if it is deemed necessary for these micro-organisms.


Impacts

What would be the impact of the introduction of visual-only inspection of fattening pigs from

non-controlled housing conditions in the UK on…?

Impact on public health

On the basis of the evidence available to inform our risk assessment and the previous risk

assessment by Hill et al., (2011) the impact on public health of the introduction of visual-only

inspection of fattening pigs from non-controlled housing conditions would not be detrimental.

In addition, it would be similar, for public health risks associated with endocarditis and

granulomatous lesions, to the impact if visual-only inspection of fattening pigs from

controlled housing conditions were to be introduced. For those associated with Salmonella

spp. and Yersinia spp. it would be equivalent to the current impact of traditional inspection of

fattening pigs from non-controlled housing conditions.

Impact on occupational exposure

On the basis of the evidence available to inform our risk assessment and the previous risk

assessment by Hill et al., (2011) it is theoretically possible that a move to visual-only

inspection would eliminate the exposure of meat inspectors, or others involved on the

slaughter line, to occupational hazards. In the case of the bacterial diseases, S. suis and

E. rhusiopathiae, this is likely to have a negligible impact on public health. We have not

considered the impacts on other occupational hazards such as those associated with the

use of knives, which could potentially be reduced by moving to a visual-only inspection

system.

Impact on animal health

We do not consider that, on the basis of the evidence available to inform our risk

assessment and the previous risk assessment by Hill et al., (2011) there would be any

significant impact on the animal health status of the population related to the introduction of

visual-only inspection of fattening pigs from non-controlled housing conditions. What would

be extremely important to be aware of is that it would lead to a significant change in the

baseline of the recorded MHS/FSA data. It would be necessary to ensure that anyone who is

involved in any data analysis and interpretation is aware of this change i.e. one could not

assess trends in the recorded frequency of conditions from before and after (or during) any

change-over period, as one would not be comparing like with like. For example, there is

increasing interest in using available data for surveillance purposes, or if information is

currently being fed back to producers based on the MHS/FSA data, they and their advisors

would need to be made aware that this was the case. This would also be the case if visual-

only inspection of fattening pigs from controlled housing conditions is introduced.

It is not possible to assess with any degree of certainty what impact such a change in

inspection methods would have on new disease threats. It is likely to be at least as low as

any of the identified known animal health hazards; however, it remains an unknown,

unknown. In this case (and the case of any known or re-emerging disease) it would depend

on the likelihood of missing the condition i.e. the probability that the disease causes

significant pathology in an inaccessible and/or uninspected part of the carcass that would

have been identified as being ‗different‘ or unusual enough from any currently known

pathology if it had been inspected, at a level that prompts further investigation, without


having caused a significant impact in the live animal and so escaping detection by clinical

signs, either on the production unit, during transport or at ante-mortem inspection.

Impact on animal welfare

We do not consider that, on the basis of the evidence available to inform our risk

assessment and the previous risk assessment by Hill et al., (2011) there would be any

significant impact on the animal welfare status of the population related to the introduction of

visual-only inspection of fattening pigs from non-controlled housing conditions.

Impacts on resources

There is not the scope here to complete a full resource impact assessment. On the basis of

the evidence available to inform our risk assessment, the outcomes of this risk assessment,

the previous risk assessment by Hill et al., (2011) and the experience and opinions of the

authors, the impacts on the resources required related to the introduction of visual-only

inspection of fattening pigs from non-controlled housing conditions would be no different to

those related to the introduction of visual-only inspection of fattening pigs from controlled

housing conditions. Some considerations are as follows (Table A4:13):

Table A4:13: Potential impacts on resources of the introduction of visual-only inspection for fattening pigs

Resource Potential impact of introduction of visual-only inspection

MHI time on the line inspecting The time that a MHI takes on the line to inspect a carcass is unlikely to be significantly reduced i.e. inspection is not usually the current limiting factor for line speed (ROA pers. opinion). Rotation of inspectors, as is currently the practice, will need to be maintained.

Equipment There will be a reduced resource requirement in terms of knives and the means to ensure that they are appropriate for use (clean, sharp etc.). The resources requirement for the recording of observations will not change.

Training Additional resources will be required to train existing MHIs to be aware of, competent in, change to and implement a new system. There may be a reduction in resources required to train new MHIs, as knife skills will no longer be required.

Operations Any operational/systems requirements for implementation of a change will be an initial additional resource e.g. the revision of operations manuals, SOPs, etc.

Other See Other impacts

If any additional risk mitigation measures are perceived to be required then these will have

resource implications. These might include capital expenditure within the abattoirs to install

mirrors or rotating systems; changes to slaughter house procedures such as logistics,

ordering of submissions etc, and staff deployment and/or numbers if carcases are required

to be presented to the MHIs in a different way; for example, if hearts are required to be

opened by a member of staff and then presented to the MHI for a visual inspection of the

opened heart. Any such resource implications may have economic implications for individual

premises. The resource implications will need to be carefully considered by risk managers

with respect to whether they are deemed proportionate to the apparent risk.


If endocarditis lesions are perceived to represent a residual risk then arrangements could be

made to open hearts for inspection, either for all carcases or where it is thought to be

necessary. However, given the low percentage of batches affected with the condition, the

low mean prevalence in batches in which the condition is present and the lack of use of the

current information for feedback on animal health status it is difficult to envisage what criteria

would be used, or set, on which to make a decision to identify any relevant subset. To make

a proactive decision a priori to the slaughter and inspection of a batch of pigs, the criteria

would need to be based on appropriate and relevant information on the disease

pattern/status from the production unit and/or from previous post-mortem inspection. What

would constitute such information for endocarditis remains to be determined. To make a

reactive decision during slaughter of a batch in response to increased visual evidence of

systemic bacteraemia is unlikely to be possible without an extremely flexible and adaptive

inspection system. The impacts on resources arise from the work needed to set up,

implement and maintain such systems. An increase in resources will be required, the scale

and duration will depend on the scope of the system proposed. The current system of Food

Chain Information submission would need to be adapted and optimised to achieve the

required aims.

If granulomatous lesions are perceived to represent a residual risk then arrangements could

be made to incise relevant lymph nodes for inspection, where it is thought to be necessary.

Again, given the extremely low percentage of batches affected with the condition (none in

our field study), and the lack of an appreciable animal health problem in the pig population, it

is difficult to envisage what criteria would be used, or set, on which to make such a decision.

To make a proactive decision a priori to the slaughter and inspection of a batch of pigs, the

criteria would need to be based on appropriate and relevant information: one possibility

could be an assessment based on the disease status in cattle in the area of the production

unit. The impacts on resources would be similar to those required for addressing risk

mitigation measures for endocarditis lesions.

There will also be resource implications if the use of harmonised epidemiological indicators

is required to mitigate the perceived risk due to Salmonella spp. and Yersinia spp.

Other impacts

Although the present meat inspection system facilitates the implementation of the sampling

for carcass contaminants such as chemicals, pharmaceuticals and their by-products, their

presence or absence is not determined by the current inspection procedures but by

laboratory analysis. A change from the traditional to a visual inspection system will not,

therefore, lead directly to any change in the risk profile; however, consideration must be

given to whether it would have any impact on the implementation of the sampling protocols

for carcass contaminants. Similar consideration should be given to any potential impact on

any other surveillance or survey activity that might currently fall within the possible remit of

MHIs; for example – if their skills set is changed (no knife skills required) will it preclude them

from other activities?


Benefits

The primary benefit of a visual-only system of inspection that encompassed pigs from non-

controlled housing conditions would be, in the UK, the ability to implement such an

inspection system. At present although it is theoretically possible to do so for pigs from

controlled housing conditions, in terms of the regulatory process, such systems have not

been implemented because slaughterhouses process fattening pigs from different

management systems. If some require traditional inspection and others can be inspected on

a visual-only basis, it is simpler to keep to the common process of traditional inspection.

Mousing et al., (1997) concluded that the main benefit of a visual-only inspection system for

pigs would ―...probably be a reduced level of cross-contamination with hazardous bacteria.‖

This is still a debatable point, although our field study provides some evidence for reduced

carcass contamination with Enterobacteriacae on carcases where it was present. Mousing et

al., (1997) also concluded that an indirect benefit would be a reduction in labour that could

be a resource that could be utilised elsewhere. It is not clear on what evidence this

conclusion was based.

If a visual-only inspection system could be implemented for all fattening pigs in the UK, then

the benefit would be the initiation of a methodology that could change the emphasis of

inspection and form the basis of a system that maximises aspects related to consumer

safety, in addition to providing reassurance of the use of sound hygiene processes in

production and, ultimately, food safety.


Relative risk

Figure A4:2 Public and animal health hazards that arise from animal health conditions and their relative risks comparing inspection methods

Animal health conditions identified as hazards in outdoor pig

carcasses - 1PAH

Endocarditis

Visual

Traditional

Granulomatous lesions

Visual

Traditional

Very Low

Negligible

Negligible

Negligible


Relative risk

Figure A4:3: Public health (foodborne) hazards from carcass microbial contaminants and their relative risks comparing inspection methods

Microbiological agents identified as public health (food-borne) hazards on

outdoor pig carcasses - 2A-PHM

Salmonella spp.

Visual

Traditional

Y. Enterocolitica*

Visual

Traditional

Very Low

Very Low

Low

Low


REFERENCES

All cited websites in this report were accessed in the period March – July 2012

Anon., (2000) SCVMRPH Opinion of the scientific committee on veterinary measures

relating to public health on revision of meat inspection procedures 24 February 2000,

SCVPH.

Anon., (2004b) Regulation (EC) No 854/2004 of the European Parliament and of the Council

of 29 April 2004 laying down specific rules for the organisation of official controls on products

of animal origin intended for human consumption. http://eur-

lex.europa.eu/LexUriServ/LexUriServ.do?uri=CELEX:32004R0854:en:NOT

Anon., (2005) Regulation (EC) 2073/2005 on microbiological criteria for Foodstuffs published

in the Official Journal of the European Communities on 22 December 2005 pdf available

from http://europa.eu.int/eur-lex/lex/en/index.htm

Alban L. et al., (2008) Assessment of risk for humans associated with Supply Chain Meat

Inspection – the Danish Way

http://www.lf.dk/Aktuelt/Publikationer/~/media/lf/Aktuelt/Publikationer/Svinekod/Modernisatio

n%20of%20Meat%20Inspection_DK.ashx

Alban L. et al., (2011) Overview on current practices of meat inspection in the EU - A

scientific report submitted to EFSA from the Danish Agriculture and Food Council

BPEX, 2008. British Pig Health Scheme 2005-2008. BPHS Report. www.bpex.org.uk

CAC (1999) Principles and guidelines for the conduct of microbiological risk assessment,

CAC/GL-30 http://www.who.int/foodsafety/publications/micro/cac1999/en/

COMISERV (2011) External scientific report submitted to EFSA on the Contribution of meat

inspection to animal health surveillance in Swine.

http://www.efsa.europa.eu/fr/supporting/pub/191e.htm

de la Rua-Domenech, R., (2006) Human Mycobacterium bovis infection in the United

Kingdom: Incidence, risks, control measures and review of the zoonotic aspects of bovine

tuberculosis. Tuberculosis (Edinb) 86, 77-109.

EFSA (2008) Report of the Task Force on Zoonoses Data Collection on the analysis of the

baseline survey on the prevalence of Salmonella in slaughter pigs, in the EU, 2006-20071

Part A: Salmonella prevalence estimates. EFSA Journal 135, 1-111

EFSA (2009a) Technical specifications for harmonized national surveys on Yersinia

enterocolitica in slaughter pigs. EFSA journal 7(11):1374

EFSA (2009b) Analysis of the baseline survey on Salmonella in breeding pigs in the EU,

2008. EFSA Journal; 7(12)1377

EFSA (2010a)The Community Summary Report on Trends and Sources of Zoonoses,

Zoonotic Agents and food-borne outbreaks in the European Union in 2008, EFSA Journal;

8(1):1496

http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=CELEX:32004R0854:en:NOT

http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=CELEX:32004R0854:en:NOT

http://europa.eu.int/eur-lex/lex/en/index.htm

http://www.lf.dk/Aktuelt/Publikationer/~/media/lf/Aktuelt/Publikationer/Svinekod/Modernisation%20of%20Meat%20Inspection_DK.ashx

http://www.lf.dk/Aktuelt/Publikationer/~/media/lf/Aktuelt/Publikationer/Svinekod/Modernisation%20of%20Meat%20Inspection_DK.ashx

http://www.bpex.org.uk/

http://www.who.int/foodsafety/publications/micro/cac1999/en/

http://www.efsa.europa.eu/fr/supporting/pub/191e.htm


EFSA (2010b). Scientific Opinion of the Panel on Biological Hazards on a Quantitative

Microbiological Risk Assessment of Salmonella in slaughter and breeder pigs. EFSA

Journal, 8(4):1547

EFSA (2011a) EFSA and ECDC: The European Union Summary Report on Trends and

Sources of Zoonoses, Zoonotic Agents and Food-borne Outbreaks in 2009, EFSA Journal,

9(3):2090

EFSA (2011b) Scientific Opinion on the public health hazards to be covered by inspection of

meat (swine) EFSA Panel on Biological Hazards (BIOHAZ) EFSA Panel on Contaminants in

the Food Chain (CONTAM) EFSA Panel on Animal Health and Welfare (AHAW) European

Food Safety Authority (EFSA), Parma, Italy. EFSA Journal 2011; 9(10):2351

EFSA (2011c) Technical specifications on harmonised epidemiological indicators for public

health hazards to be covered by meat inspection of swine. EFSA Journal 2011; 9(10):2371

EFSA (2012) European Food Safety Authority and European Centre for Disease Prevention

and Control (ECDC): The European Union Summary Report on Trends and Sources of

Zoonoses, Zoonotic Agents and Food-borne Outbreaks in 2010; EFSA Journal 2012;

10(3):2597

All EFSA documents are available online from: www.efsa.europa.eu/efsajournal

FAO (2010) Manual on meat inspection for developing countries : Specific diseases of pigs.

Available at: http://www.fao.org/docrep/003/t0756e/t0756e05.htm,

Fosse et al., (2008) Zoonoses transmitted by pork consumption. Vet. Res. 39 :01 DOI:

10.1051/vetres:2007039

Gürtler et al., (2005): Prevalence of Yersinia enterocolitica in fattening pigs. J. Food. Prot.

68(4) 850-4

Hamilton D.R., et al., (2002) Risk-based evaluation of post-mortem inspection procedures for

pigs in Australia. Veterinary Record 151, 110-116

Hill et al., (2003) A ‗farm-to-consumption‘ risk assessment for the adverse effects to human

health of Salmonella Typhimurium in pigs. Proceedings of the 10th International Symposium

on Veterinary Epidemiology and Economics. Available at www.sciquest.org.nz

Hill A. et al., (2011) A Risk and Benefit Assessment for visual-only meat inspection of UK

indoor and outdoor pigs. Project FS245009 final report.

http://www.foodbase.org.uk//admintools/reportdocuments/695-1-

1179_MC1002_FINAL_REPORT.pdf

HPA (2007) Uncommon pathogens involved in bacteraemia, England, Wales and Northern

Ireland: 2001 – 2006.pdf

HPA (2011a) Uncommon pathogens involved in bacteraemia, England, Wales and Northern

Ireland: 2006 – 2010.pdf

Leps, J., Fries, R., (2009) Incision of the heart during meat inspection of fattening pigs - A

risk-profile approach. Meat Science 81, 22-27.

http://www.efsa.europa.eu/efsajournal

http://www.fao.org/docrep/003/t0756e/t0756e05.htm

http://www.sciquest.org.nz/

http://www.foodbase.org.uk/admintools/reportdocuments/695-1-1179_MC1002_FINAL_REPORT.pdf

http://www.foodbase.org.uk/admintools/reportdocuments/695-1-1179_MC1002_FINAL_REPORT.pdf


Lewis M., (2010) Farm Business Survey 2009/2010 Pig production in England. Rural

Business Research, Askham Bryan College. http://www.askham-

bryan.ac.uk/media/398849/Pig%20Report%202010-11%20_final_%20(2).pdf

Makrai, L. et al., (2008): Isolation and characterisation of Rhodococcus equi from

submaxillary lymph nodes of wild boars (Sus scrofa). In: Vet Microbiol. PMID 18499361

doi:10.1016/j.vetmic.2008.04.009

Mousing, J., et al., (1997) Meat safety consequences of implementing visual post-mortem

meat inspection procedures in Danish slaughter pigs. Veterinary Record 140, 472-477.

NADIS (2012) http://www.nadis.org.uk/bulletins/whole-carcass-condemnation.aspx

OIE (2004) Handbook on Import Risk Analysis for Animals and Animal Products. OIE, Paris,

France.

Ortiz Martinez P. et al,. (2010) Wide variety of bioserotypes of enteropathogenic Yersinia in

tonsils of English pigs at slaughter. Int. J. Food. Microbiol. 139 (1-2): 64-9

Ortiz Martinez P. PhD thesis (2010): Prevalence of enteropathagenic Yersinia in pigs from

different European countries and contamination in the pork production chain.

http://urn.fi/URN@ISBN:978-952-10-6393-0

Ortiz Martinez P. et al., (2011): Variation in the prevalence of enteropathogenic Yersinia in

slaughter pigs from Belgium, Italy and Spain. Foodborne Pathog. Dis. 8 (3) 445-50

Reboli, A.C., Farrar, W.E., 1989, Erysipelothrix-rhusiopathiae - an occupational pathogen.

Clinical Microbiology Reviews 2, 354-359.

Wang et al., (2010) Erysipelothrix rhusiopathiae. Vet Microbiol. 2010 Jan 27;140(3-4):405-

17.

http://www.askham-bryan.ac.uk/media/398849/Pig%20Report%202010-11%20_final_%20(2).pdf

http://www.askham-bryan.ac.uk/media/398849/Pig%20Report%202010-11%20_final_%20(2).pdf

http://www.nadis.org.uk/bulletins/whole-carcass-condemnation.aspx

FSA145003 Final report ANNEX 5 v6 Page 1 of 25

ANNEX 5


OBSTACLES TO IMPLEMENTATION - OBJECTIVE 5

EXECUTIVE SUMMARY The aim of this work was to identify any possible obstacles that may hamper the

implementation of a risk-based visual only inspection system and to suggest appropriate

adaptations to overcome those obstacles.

We have identified ten main areas where potential obstacles to implementation may arise.

These include:

the development of (and agreement on) how to classify batches of pigs according to

risk;

how to ensure that appropriate arrangements are in place to process batches of pigs

according to their risk classification;

the resource implications in terms of alterations to plant layout, staff availability and

amendments to plant operating procedures to ensure meat quality;

maximising the visibility of the carcass and offal while minimising microbial

contamination;

the potential loss of data for animal health and surveillance purposes;

and, issues associated with change in the job, methods and responsibilities for FSA

staff.

There are adaptations that could be made to overcome most of these obstacles; some

however, will require further work and some will need to involve change management and

behavioural change; areas that are outwith the scope of this project.

We conclude that the majority of the obstacles identified are the same as those that are

expected if visual-only inspection were to be implemented for fattening pigs from controlled

housing conditions and that any adaptations required would be similar.


Table of contents

Executive summary ............................................................................................................... 1

INTRODUCTION .................................................................................................................. 4

Objectives ............................................................................................................................. 5

Particulars of inspection during the trial ................................................................................. 5

Obstacle 1: ............................................................................................................................................. 6

Adaptation 1: ......................................................................................................................................... 6

MATERIALS AND METHODS .............................................................................................. 7

RESULTS AND DISCUSSION .............................................................................................. 7

Decision tree (flow chart) for traditional and visual-only inspection ........................................................ 7

Identification of obstacles and possible solutions (including the ones identified in the decision tree).. 12

Obstacle 2: ........................................................................................................................................... 12

Adaptation 2: ....................................................................................................................................... 12

Obstacle 3: ........................................................................................................................................... 14

Adaptation 3: ....................................................................................................................................... 14

Obstacle 4: ........................................................................................................................................... 14

Adaptation 4: ....................................................................................................................................... 14

Obstacle 5: ........................................................................................................................................... 15

Obstacle 6: ........................................................................................................................................... 15

Adaptations for 5 & 6: ......................................................................................................................... 15

Obstacle 7: ........................................................................................................................................... 16

Adaptation 7: ....................................................................................................................................... 16

Obstacle 8: ........................................................................................................................................... 17

Obstacle 9: ........................................................................................................................................... 17

Adaptations for 8 & 9: ......................................................................................................................... 17

Obstacle 10: ......................................................................................................................................... 18

Adaptation 10: ..................................................................................................................................... 18

Limitations of trial ................................................................................................................ 18

CONCLUSIONS AND RECOMMENDATIONS .................................................................... 21

Conclusions ........................................................................................................................ 21

Recommendations .............................................................................................................. 22

REFERENCES ................................................................................................................... 25


Tables and figures

Figure A5:1: Decision tree for traditional inspection of fattening pigs in the UK ..................... 9

Figure A5:2: Decision tree for visual-only inspection of fattening pigs in the UK .................. 10

Table A5:1: Differences in visual-only inspection compared to traditional inspection in

outdoor pigs during FSA trial ............................................................................................... 11

Figure A5:3: Figurative layout of inspection, detection and recording for both inspection

methods at study abattoir .................................................................................................... 19

Figure A5:4: Bird‟s-eye schematic of the layout of inspection, detection and recording for

both inspection methods at study abattoir ........................................................................... 19

Table A5:2: The ten potential obstacles to implementation of a risk-based visual meat

inspection system for fattening pig carcases and adaptations that may be required to

overcome them. .................................................................................................................. 23


INTRODUCTION Traditionally meat inspection performed at abattoirs consists of visual inspection, palpation

and incision of specific parts of the carcase and offal from slaughtered pigs as specified by

the Regulation (EC) 854/2004 (Anonymous, 2004b) in order to declare the meat fit for

human consumption. Such inspection is now not totally appropriate for the control of food-

borne risks to public health. Historically, meat inspection detects gross pathological lesions

in the carcase and offal from slaughtered pigs, thus enabling the detection and

condemnation of meat unfit for human consumption. As traditional inspection is restricted to

the identification of conditions and/or diseases with macroscopic signs, it excludes

microbiological food-borne hazards, such as Salmonella, Campylobacter, and Yersinia.

These potential food-borne risks for public health are not controlled by the traditional

inspection methods.

Information from farms of origin was not historically available to the official veterinarian prior

to the slaughter of the animals. Knowledge about management and other factors on the farm

of origin can give indications of the likely health status of pigs sent to the abattoir and it can

help to identify any disease or conditions in carcase and offal that might endanger public

health. Post-mortem inspection is currently (Anonymous, 2004b) part of a whole system

which includes ante-mortem inspection and the latter must include information from the farm

of origin.

The European Commission reviewed the current Meat Inspection legislation (Anonymous,

2000). They considered some alternative methods of inspection for fattening pigs that were

included in Regulation (EC) No 1244/2007 (Anonymous, 2007b). The Scientific Committee

on Veterinary Measures relating to Public Health was asked to review the current inspection

at slaughterhouses for pigs. They concluded that traditional inspection is limited in its goal to

prevent zoonotic infections in humans from fattening pigs processed in abattoirs and they

explored alternative post-mortem inspection methods. They added that, not only did

traditional inspection not prevent current food-borne risks for humans, but it could increase

the microbial risk by cross contamination of the carcases.

Abattoir meat inspection is implemented in the United Kingdom (UK) according to the

European legislation, Regulation (EC) 854/2004. Under certain conditions, fattening pigs

from controlled housing conditions in integrated production systems since weaning are only

required to be visually inspected at abattoirs (Anonymous, 2007b). This is not the case for

fattening pigs from non-controlled housing conditions; these still have to be inspected at

abattoirs by traditional methods.

The present use of farm information delivered to the Official Veterinarian (OV) in the abattoir

in the form of Food Chain Information (FCI), prior to the slaughtering of pigs, should help to

identify hazards before the carcases are on the line. Procedures to improve the detection of

specific conditions that might be at a higher risk of occurring in a particular lot coming from a

farm with a history of a condition, e.g. with a high prevalence of Salmonella, can then be

implemented for that lot. This use of FCI is a process introduced in the UK, by the Food

Standards Agency (FSA), for pigs in 1st January 2008 as required by Regulation (EC) No.

853/2004 (Anonymous, 2004a), in order to improve the quality of the ante-mortem inspection

in animals prior to slaughter. Information must be sent by producers at least 24 hours before

the arrival of the animals to the abattoir. Information regarding the farm of origin is included


to help the OV to decide if any further inspection or tests are needed during the inspection of

animals/carcases to declare them fit for human consumption.

It is believed that there is better control of the health of the animals at farm level, if pigs are

housed in controlled conditions (Anonymous, 2011). The prevalence of certain diseases

affecting public health (such as tuberculosis (TB) or Trichinella) is also believed to be lower

in pigs from controlled housing conditions, than in outdoor pigs, as they are not exposed to

sources of disease. If this is not the case and the risk from batches of pigs raised under non-

controlled housing conditions is not higher than that of batches of pigs coming from

controlled housing conditions, then outdoor pigs can also be included in the visual-only

inspection process.

The aim of Project FS145003 was to investigate the implications of changing the inspection

method, from the traditional method to a visual-only method, for fattening pigs from non-

controlled housing conditions in the UK.

OBJECTIVES The objectives of this part of the study were:

To identify any possible obstacle, at any level, that might hamper the implementation

of visual-only inspection system for fattening pigs from non-controlled housing

systems in abattoirs.

To suggest any appropriate adaptation of the visual-only inspection that will help to

overcome the obstacles identified.

For those conditions where visual-only inspection appears to be insufficient

compared to traditional meat inspection a risk analysis approach will be carried out to

identify strategies to improve the detection level of visual-only inspection.

Particulars of inspection during the trial

The visual-only post-mortem inspection carried out during the trial is different to the

proposed visual-only inspection that would be implemented in abattoirs, if a change of

inspection method was to be applied. Due to the nature of the study, no handling of the

carcase and/or offal was permitted by Meat Hygiene Inspectors (MHIs)1 involved in the

visual-only inspection during the trial. Microbiological swabs were collected after both

inspections as part of the trial (Annex 3) so the visual-only inspection was restricted to only

visual inspection, with no handling of carcases i.e. a „hands-off‟ inspection.

The visual-only inspection proposed for future implementation (Huey, 2012) would be a form

of risk-based inspection, where:

Ante-mortem inspection would include not only an inspection of the live animals but

also the consideration of the post-mortem inspection results of previous batches of

pigs from the same source and the FCI information provided.

1 Meat Hygiene Inspectors is the UK term for Official Auxiliaries


If the batch of pigs was classified during the ante-mortem inspection as low risk and

there was no reason to suspect that there would be conditions during the inspection

that require any further inspection i.e. Officials would not need to palpate and/or

incise any part of the carcase and/or offal during post-mortem inspection to carry out

their inspection: then the batch will undergo ONLY VISUAL INSPECTION.

If the batch of pigs was classified during ante-mortem inspection as high risk and/or

there was reason to suspect that there would be conditions that require any further

inspection i.e. Official would need to palpate and/or incise any part of the

carcase/offal during post-mortem inspection to carry out their inspection: then the

batch will undergo VISUAL, PALPATION AND INCISION INSPECTION (referred to

from here on as „further inspection‟).

For example: animals coming from batches with a high risk (e.g. a previous history of the

conditions in the abattoir) of tumours might undergo further incisions of lumps found in the

carcase by Officials to identify them as potential tumours.

One potential method of classification of batches, to enable the official inspectors to specify

the risk and therefore ensure the appropriate method of post-mortem inspection for each

batch, could be to consider two levels of risk. This is an example for illustrative purposes

only, not a recommendation.

i. Low risk: No currently recorded presence in the herd of origin of any conditions such

as abnormal high mortality, Salmonella, or any disease diagnosed or suspect in the

herd and a previously low presence/absence of post-mortem conditions identified in

the abattoir in recent slaughter batches of animals from the herd.

ii. High risk: Recorded presence in the herd of origin of any conditions such as

abnormal high mortality, Salmonella, or any disease diagnosed or suspect in the herd

and a previously high frequency of post-mortem conditions identified in the abattoir in

recent slaughter batches of animals from the herd.

For the purpose of the study, no herd categorisation according to risk was carried out during

the trial. This is an area that needs more work (Annex 4).

Obstacle 1: The development of appropriate criteria, on which relevant parties agree, in

order to classify the risk status of batches of pigs is a potential obstacle to implementation of

a risk-based inspection system.

Adaptation 1: Further work is required

We include in this report any potential obstacle to the implementation of risk-based visual-

only inspection as it would be implemented were the system to be changed. This will include

some obstacles not found during the trial as a consequence of the alterations made, but

expected to be present if the whole method was implemented i.e. including the classification

of the risk level of the batch at ante-mortem inspection.


MATERIALS AND METHODS 1. Data sources

i. Observational data collected during the trial carried out in December 2011,

January 2012, February 2012, and March 2012 at the study abattoir. This

information includes comments given by MHIs working on the plant during the

trial.

ii. Observational experiences of the primary author (ROA) from different

abattoirs in the UK. Use of the knowledge of the layout of abattoirs visited by

the primary author (ROA) in UK and FSA inspection procedures in those

premises is used to extrapolate the observations to other premises within the

UK.

iii. Analyses of the outcomes from objectives 1 to 4 to identify those conditions

where visual-only inspection is insufficient compared to traditional inspection.

iv. Literature research on the subject to identify obstacles found in similar

situations, such as trials carried out in other countries.

2. Methods

i. Decision tree (flow-chart) of visual-only inspection compared to flow-chart of

traditional inspection for post-mortem inspection at slaughterhouse for

fattening pigs from non-controlled housing conditions.

ii. Analysis of the differences between these two procedures and the logistics

needed to put in place the alternative method (visual-only inspection).

iii. Identification of logistical issues which may jeopardise the aim of post-mortem

inspection of carcases in the slaughterhouse.

iv. Identification of potential issues affecting the likelihood of detecting conditions

in carcases of fattening pigs from non-controlled housing conditions under the

visual-only inspection.

v. Analysis of the impact for public health, animal health and welfare of those

conditions under-detected by visual-only inspection compared to traditional

inspection.

vi. Discussion of possible and practical strategies to improve the detection of the

above conditions where public health, animal health and welfare are reduced.

RESULTS AND DISCUSSION

Decision tree (flow chart) for traditional and visual-only inspection

We have drawn a decision tree for traditional inspection (see Figure A5:1), as it is currently

implemented in abattoirs in the UK for fattening pigs, and one for the alternative inspection

method as it would be if a risk-based visual inspection system (Figure A5:2) was

implemented. In the visual inspection decision tree we have identified three steps that were

not present in traditional inspection:


- Analysis of the risk status of a batch of pigs prior to slaughter

- Further inspection could be required for certain batches (high risk pigs)

- Further need for handling of carcasses and/or offal by abattoir personnel

In order to ensure that visual inspection is successfully implemented these new procedures

would have to be addressed at each abattoir. These new procedures are the same as those

that would be expected to be required if visual-only inspection were implemented in fattening

pigs from controlled housing conditions (which is deemed acceptable in the UK but not

implemented, for logistical reasons).

We analysed the needs for these new procedures together with other issues detected during

the trial.


Decision tree in traditional inspection

Ante mortem

Disposal

Post mortem inspection: Traditional inspection

Chillers

Disposal

Fit for slaughter

Unfit for slaughter

Unfit for human consumption

Fit for human consumption

FCI

Figure A5:1: Decision tree for traditional inspection of fattening pigs in the UK


Decision tree in visual-only inspection

Ante mortem

Low risk

High risk

Post mortem inspection: Visual-

only inspection

Post mortem inspection:

Further inspection (V-P-I)



Chillers

Disposal

Chillers

Disposal

Unfit for slaughter

Fit for slaughter

Risk status analysis based on

FCI



Chillers

Chillers

Disposal

Is there any need for further handling?

Is there any need for further handling?

Yes

Yes

No

No

Plant staff corrections

Plant staff corrections

New

pro

ced

ure

Fit for slaughter

Ante mortem

Unfit for slaughterDisposal

Figure A5:2: Decision tree for visual-only inspection of fattening pigs in the UK


Table A5:1: Differences in visual-only inspection compared to traditional inspection in outdoor pigs during FSA

trial

Based on the legal requirements contained in Regulation (EC) 854/2004.

¹When for human consumption ²When necessary ³Unless penis discarded 4Sows

Traditional Inspection

Visual-only Inspection

Head visual visual

Tongue visual visual

Submaxillary lymph nodes visual, incise visual

Mouth visual visual

Fauces visual visual

Throat visual visual

Lungs visual, palpate, incision¹ visual

Trachea visual, incision¹ visual

Main bronchi branches visual, incision visual

Oesophagus visual visual

Bronchial and mediastinal lymph nodes


Pericardium visual visual

Heart visual, incision visual

Diaphragm visual visual

Liver visual, palpate visual

Hepatic and pancreatic lymph nodes


GIT and mesenteric visual visual

Gastric and mesenteric lymph nodes

visual, palpate, incision² visual

Spleen visual, palpate² visual

Kidneys visual, incision² visual

Renal lymph nodes incision² visual

Pleura and peritoneum visual visual

Genital organs visual³ visual

Udder visual visual

Supramammary lymph nodes visual, incision²-4 visual

Umbilical region (young) visual, palpate, incision² visual

Joints (young) visual, palpate, incision² visual

It should be noted that, based on current evidence, we do not consider that there is any



that which currently exists with traditional inspection (Annex 4).


Identification of obstacles and possible solutions (including the ones identified in the decision tree)

1. Changes to FCI (included as part of the current ante-mortem inspection)

As discussed earlier, the analysis of FCI as part of the ante-mortem inspection is essential if

a new method of post-mortem inspection is to be based on risk. Decisions would be made

prior to animals‟ arrival at the abattoir in order to adapt the inspection to the perceived risks

from different batches and/or farms. The analysis of FCI from different batches and/or farms,

at least 24 hours before arrival for slaughter, may lead to the conclusion that further

inspection of carcases is required, i.e. of those with a perceived higher risk of conditions

which might affect the fitness for human consumption.

FCI plays an important role in the detection of conditions/diseases during the post-mortem

inspection. Healthier animals are normally taken to the abattoir, leaving the sick ones on the

farm of origin (there is a welfare pre-requisite of „fitness to travel‟ and health, in order to be

transported and to be eligible for slaughter, respectively). Animals sent for slaughter should

appear healthy; however they may still have an underlying problem. It may be more difficult

to identify them as unfit for human consumption during post-mortem inspection, if the latter is

visual-only. When the information from the rest of the herd is available as FCI to the Official

at the abattoir, procedures might be put in place to increase the likelihood of detecting

conditions/diseases (risk-based inspection). These procedures might vary depending on the

conditions expected to be present, e.g. if the presence of clinical signs indicative of infection

with organisms known to lead to endocarditis are recorded and so endocarditis is expected

to be present or occur at a high frequency in the batches, then opening of hearts prior to the

Officials‟ post-mortem inspection may be required; or if due to prior knowledge of the history

of the herd, a high prevalence of milk spot is expected, further handling of offal by Officials

will be required to increase the likelihood of detection of the condition.

Information recorded on FCI must be complete and comprehensive to allow the Official to

take the appropriate measures. FCI should include information such as history of recent

Salmonella in the herd, any disease suffered in the herd recently, mortality rate, vaccination

status of the herd, etc. to help the Official to define the health status of the herd of origin of

the batch sent to slaughter. It might also be important to make the FCI available to the

Official with more time than the current 24 hours prior to arrival of animals to the abattoir.

This will allow the arrangement of any specific post-mortem inspection, such as slaughter at

the end of the day to avoid cross contamination of carcases when slaughtering animals

coming from a farm with history of Salmonella.

Obstacle 2: the selection of the post-mortem method has to be made prior to animals

arriving at the abattoir based on the risk of each batch.

Adaptation 2: the improvement of the availability of Food Chain Information (FCI) to allow

the Officials at the slaughterhouse to gather enough information with sufficient time to define

the risk of each batch, according to the agreed criteria, and to implement the appropriate

post-mortem method and arrangements.


2. Changes to plant layouts and slaughtering procedures

During traditional inspection in the study abattoir (which is similar to most abattoirs in the

UK): all carcases with any conditions that need trimming and/or other process that need to

be rectified (e.g. after MHIs have identified an issue to be addressed before the carcase can

be considered as fit for human consumption) are sent to the “detention”, or “detained”. Here

plant personnel will take action and then present the carcase to an additional MHI who will

confirm that the carcase is fit and stamp it with the health mark. It will then be released to the

chillers. These detained rails have a limited capacity. The number of carcases they can take

is different in each abattoir; it was around 20 carcases in the study abattoir. Once they are

full the main slaughter line has to be stopped to allow time to clear space on the detention

rail before a new carcase can be placed on it.

How to deal with aesthetic conditions (such as hair in carcases, bruising, bite wounds, or

blood splash) should be included as part of Hazard Analysis Critical Control Point (HACCP)

plan for any plant. Different strategies may be used: trimming in the detection rail detailed

above is only one possibility. Plant operators may favour the rectification of those conditions

on the slaughter line; training expert staff in trimming them without increasing the risk of

contamination to other carcases or delaying the speed of the line, or they might manage

them during the further processing of the carcases, such as preparing meat in the adjacent

cutting plant.

It has been suggested (Anonymous, 2011) that, during the visual post-mortem inspection,

carcases suspected visually to be affected with any conditions that may classify the carcase

as unfit for human consumption should be placed on a “detention rail”. Here, further

inspection (palpation and/or incision) to confirm fitness for human consumption would be

carried out. The intention is to place them in the “detention rail” and carry out the further

inspection without causing any cross contamination to the adjacent carcasses in the

slaughter line. These methods might be safe but, in the opinion of the primary author (ROA)

not practical, or feasible, in the modern UK slaughterhouse, as the capacity of current

“detention rails” will not be sufficient for the amount of potential carcases that need further

inspection. In the author‟s opinion, those further inspections will have to occur on the

slaughter line (right after the visual inspection). Training will need to be put in place to

ensure that the risk of cross contamination to other carcases is minimised when these

inspections are done.

Aesthetic conditions will be no longer inspected by the Officials, as there is no food safety

implication. This raises a potential additional problem that may occur with the logistics within

abattoirs. Officials will stamp all carcases fit for human consumption and send them to the

chillers, but as we have explained earlier, plant staff will need to handle them to resolve the

aesthetic issues, for example they may need to trim bruising. This could be done before or

after the official post-mortem inspection. If it is done after inspection, it is possible that an

increase will be required in the capacity of chillers to keep carcases and offal in need of

further handling by plant staff. This might affect the layout of the abattoir. Affected carcases

and offal would need to be stored in chillers where practices such as opening hearts, lungs

or removing parts of carcases for aesthetic purposes (when no risk for public health has

been identified) would take place. These modifications of the layout of the abattoir would

need to be part of the plant‟s HACCP plan and the complexity will depend on each particular

abattoir.


The layout of the slaughter line might not permit visual access to the whole carcase for a

complete visual-only inspection. This problem was observed in the study abattoir. The back

of the carcase is an area particularly difficult to access with a “hands off” system. Handling of

carcases is not specifically included in the process of traditional inspection but it is

necessary to an overall inspection; for example, opening the carcase to achieve better

inspection of the inside or turning the carcase around to inspect the back of the carcase.

During the trial in the study abattoir, visual-only inspection was located in an area of the

slaughter line where this “turning” was logistically not possible; the visual-only MHI was

unable to achieve an overall inspection, especially of the tail area. This might not be the

case for all abattoirs and would not have been the case if the visual-only inspection had

occurred at the traditional inspection point in the study abattoir. The use of tools, such as

mirrors or turning devices, has been suggested to help the visual access to the whole

carcase without handling. Further research would be necessary to study the practicability

and usefulness of different tools to improve the visual-only inspection in abattoirs in UK.

The effect of the layout of the study abattoir might have biased the results observed during

the trial. The frequency of detection of conditions identified during the visual-only inspection

might have been higher if the poor access has not impeded the visibility to the whole of the

carcase.

Obstacle 3: A potential increase in the number of carcases in need of further handling

before they are released to the chillers and the consequent resource implications, whether

that be in terms of capital investment and alterations in the layout of abattoirs, or plant staff

numbers and training.

Adaptation 3: Alterations to plant layout and staff resources to handle additional carcass

dressing will need to be plant specific.

Obstacle 4: The implementation of mechanisms to ensure that detection of conditions by

visual inspection is maximised (e.g. some parts of carcases/offal may not be accessible to

the visual inspection) and the consequent resource implications.

Adaptation 4: Additional mechanisms to ensure access and visibility to carcases and offal

will need to be plant specific. Risk managers will need to consider if the resource

implications are proportionate to the perceived risk and the potential gain. Examples include

changes to the layout of the abattoir, or plant staff tasks, such as introducing rotating tools or

mirrors to allow access to all parts of carcases/offal or the preparation of offal (opening of

hearts) prior to inspection.

3. Changes to plant HACCP and procedures

Historically, meat inspection has enabled alterations to carcases that have little impact on

the risk to public health e.g. conditions that affect the aesthetics of the carcases but are not

considered a risk for humans. If this is no longer part of the duty of the Officials from the

Competent Authority (FSA in UK), then it will become part of the food business operators‟

(FBO) responsibilities at the abattoir.

If visual-only post-mortem inspection were to be implemented at the abattoir, these

conditions might no longer be detected and resolved by Officials. In order to deal with these

aesthetic issues there would need to be further handling of carcases and offal by the


operators. These additional procedures will need to be included in the HACCP protocol in

each abattoir. There is a potential risk of microbial cross contamination of carcases and/or

offal during these additional handling procedures by the Food Business Operator (FBO)

personnel. This would probably occur after the official inspection (Fig A5:2) and this, must be

included as an important control point for the plant HACCP plan.

Training on how to handle carcases, detect and rectify these conditions would need to be

organised and delivered by the FBO to their staff. The risks that arise from potential

occupational hazards, which currently apply to Officials, would transfer to the plant personnel

staff trained to handle the carcases and/or offal not rejected at visual-only inspection by

Officials.

Obstacle 5: The need to transfer additional meat quality responsibilities to the food business

operators to deal with all the aesthetic conditions in carcases (such as hair contamination)

that no longer fall within the Competent Authority‟s remit.

Obstacle 6: The potential for microbial cross contamination of carcases and/or offal during

the plant procedures to handle and deal with the above aesthetic conditions (after official

post-mortem inspection).

Adaptations for 5 & 6: There are no immediate solutions to these obstacles; they will be

plant-specific. FBOs will need revise their HACCP plans to address these issues.

4. Changes to data generated in the abattoir

Information collected in the abattoirs has a role in the surveillance of animal health and

welfare in UK and there is increasing interest in its use and utility. Farmers receive (or, at

least, in theory could receive) the Collection and Communication of Inspection Results

(CCIR) for all batches of pigs processed in the abattoir. This could help the farmers to

improve the health status of their herd by identifying conditions and/or diseases missed at

the farm level but identified during the ante-mortem and post-mortem inspection of the

animals.

If visual-only inspection is implemented and some conditions (regarded as low risk for public

health, such as certain parasites) are not identified and/or recorded, there is a potential for

the loss of information that currently benefits farmers and national animal health and welfare

status. This includes aesthetic conditions discussed in point 2, such as bruising, bite wounds

and blood splash. Where these are considered to be useful welfare indicators additional

effort will be required to ensure that they are still recorded.

These conditions could be identified later on by plant staff (when further handling of carcase

and offal takes place). To ensure that all post-mortem conditions (regardless of who

identified them) are recorded and information is fed back to the appropriate people, CCIR

could collect data not exclusively from the FSA records but also from the “further inspection

by plant”. These conditions might not be essential for public health but they are still

information that should be fed back to farmers/veterinarians as part of the Knowledge

Transfer, so they can be aware of the status of their herds.


Obstacle 7: a perceived, or actual, loss of data on animal health and welfare from animals

slaughtered in the abattoir.

Adaptation 7: Further work to identify the utility and benefits of the use of alternative data

sources, such as the plant‟s own records of conditions found in the abattoir (e.g. HACCP

records), followed by the implementation of appropriate systems where relevant.

5. Lack of competence, skills and understanding of the new method of post-mortem

inspection: visual-only inspection

5.1. Meat Hygiene Inspectors (MHIs)

Meat hygiene inspectors might perceive the new system as a potential way to reduce their

work loads and subsequently lead to a loss of their jobs. Efforts must be made to ensure that

this is not the case. The new system will need to be explained; the science behind it and the

role of meat inspection should be adequately communicated to MHIs for the alternative

method.

There will be a need to retrain MHIs for the new visual-only inspection of meat. MHIs have

been carrying out traditional inspection for a long period of time and their routine (and habits)

may be difficult to modify.

Lack of observation during the visual-only inspection due to the reduction of activities and

monotony of the new system (Mousing et al., 1997) might lead to a reduction of conditions

detected by MHIs. This could be minimised by reducing the inspection period in the

slaughter line for MHIs in the rota. These rotas are individual to each abattoir, and will need

to be adjusted in each one.

There is a possible need for better communication between the inspection points for

carcases and offal during the post-mortem inspection. In visual-only inspection the detection

of several conditions in different parts of carcase and/or offal will be required to identify

carcases and offal unfit for human consumption; for example, enlargement of lymph nodes in

different parts of the carcase to identify tuberculosis (TB). In the study abattoir there was

some information missing as a consequence of the layout of visual-only inspection during

the trial. MHIs at the carcase inspection point and at the offal inspection point were located

in different positions due to spatial limitations and logistics. There was no possible

interchange of information, therefore the information was not correlated and the MHIs did not

have an overall picture of conditions found in the pig (carcase and offal). If, in the study

abattoir, only a visual-only inspection system was in place, then both MHIs would be in the

same (or close) location so information about a carcase and from the corresponding offal

could be more easily exchanged. This would help to make decisions about the outcome of

the post-mortem inspection.

To minimise this risk of missing conditions the layout of post-mortem inspection must be

adapted to the new method. This must be analysed for each abattoir (because the design of

the slaughter line and inspection points might be different to the study abattoir) to ensure

practicality and efficiency.

.


Obstacle 8: Official inspection team - perception of possible loss of their jobs as a

consequence of new system.

Obstacle 9: Reduced awareness due to lack of experience with the new post-mortem

inspection method, insufficient communication between carcase and offal inspection points

and/or the potential for a reduction in the attention paid to the task, during post-mortem

inspection, due to the monotony of the process.

Adaptations for 8 & 9: It is outwith the scope of Project FS145003 to suggest methods of

change management and how to achieve behavioural change; however, training (and

subsequent auditing) of all of the Official inspection team involved with the new system will

be needed to ensure an adequate understanding of it and to maintain a high standard of

communication and service.

5.2. Official Veterinarians

With traditional inspection, ante-mortem inspection was primarily used to assess if animals

that arrived at the abattoir were fit for slaughter; to search for any clinical signs of any

disease and to check that there were no welfare issues, either due to transport or arising

from the farm of origin.

If a new or revised risk-based inspection is to be applied, the OVs must be trained in any

new requirements that are needed to carry out risk-based categorisation of pig batches,

using FCI and previous data (conditions found in previous batches from the same farm). This

is required to ensure that the decisions, on which inspection should be applied, are made in

a standardised, consistent manner.

This means that the FCI must be complete and available for assessment at least 24 hours

before arrival, but preferable earlier than that. This will allow Officials to ensure that the

logistical issues are addressed and they are ready to undertake the appropriate inspection.

The role of FCI and ante-mortem inspection is crucial. Animals sent to the abattoir might

appear healthy, so FCI from the farm of origin may help to assess the real health status of

the farm and detect the hidden presence of some subclinical conditions, which might make

the meat from those animals unfit for human consumption, regardless of the result of ante-

mortem inspection at the abattoir. The quality of FCI must be assessed to ensure that the

data provided by producers are comprehensive and suitable for the purpose of ante-mortem

inspection. There may need to be training for producers on what information is needed for

FCI. The FCI provided may need to be adapted and/or revised. FCI needs to include further

information from the herd of origin, such as recent diseases observed in the herd

(Salmonella, Erysipelas...), any conditions observed in the herd (such as lameness),

abnormal high mortality, vaccination strategy or any management details that might help

Officials to determine the health status of the herd of origin. This would be necessary to

predict the likelihood of certain conditions (or take extra precautions to avoid cross

contamination, e.g. in case of potential Salmonella presence in pigs sent to slaughter when

animals come from a farm with recent history of Salmonella) during the post-mortem

inspection, such as higher prevalence of pleurisy, Erysipelas-like lesions, or arthritis.


If the FCI was available to Officials with more time than is currently required, it would be

helpful. The appropriate time period required would need to be investigated and optimised; it

has been suggested that, ideally having it a week prior to the arrival of the batch, would be of

use. Information from the actual batch to be sent to the abattoir might be unknown (as

management in the farm might be unsure of the specific batch to be sent) but information

regarding the herd of origin is known and of great importance for the Officials. Information

from the FCI and post-mortem data from previous batches received in the same abattoir,

collected and kept by the FSA will need to be used for decision-making. FSA could then

make any logistical arrangements in advance; for example, the availability of additional staff,

re-arrangement of slaughter strategies (e.g. pigs coming from farms with a history of

Salmonella to be killed at the end of the day or in a specific day of the week) or any other

decision to enhance the efficiency of the post-mortem inspection (to maintain the safety of

the meat produced in the abattoir) and to minimise the cross-contamination of carcases

during the process.

This refined approach to ante-mortem inspection may also be important to identify any

animal health or welfare problem at the farm, which, with traditional inspection, would have

been identified later on during the post-mortem inspection. One of the concerns when

changing to visual-only inspection is that diseases such as TB might be missed. If suitable

predictive risk factors can be determined through appropriate studies, then animals at a

higher risk of being infected, e.g. with TB, might potentially be identified during the ante-

mortem inspection of FCI.

Obstacle 10: A perceived, or actual, decrease in the quality of the Official Veterinarian

work, such as identification of high risk batches in need of a post-mortem inspection with

inclusion of further incision of palpation, due to the lack of experience with the new system.

Adaptation 10: It is outwith the scope of Project FS145003 to suggest methods of change

management and how to achieve behavioural change; however, training (and subsequent

auditing) of all FSA personnel involved with the new system will be needed to ensure an

adequate understanding of the concepts and responsibilities that are involved and so to

maintain a high standard of service.

Limitations of trial

The recording of the conditions was not exactly the same in the traditional system (touch

screen) and visual-only system (spreadsheet) during the first pilot week of the trial, therefore

the visual-only spreadsheet was amended to reflect the same conditions as in the traditional

inspection. During the first week these missing conditions in the spreadsheet were recorded

by MHIs as “other” (as requested) until amended spreadsheets were created and circulated

for the following weeks.

At the study abattoir, during the traditional inspection, conditions were detected at two points

on the slaughter line (Figure A5:3 & A5:4) and identified using plastic markers attached to

the carcase. These conditions were not recorded. In order to speed up processing on the

line, an FBO operative between the first traditional inspection and final carcase

inspection/health marking would remove conditions such as hair contamination. Carcases

found to be free of abnormalities at the second traditional inspection point, including some

where conditions had been rectified (i.e. some hair removal, stripping of pleura, peritoneum)


went along the main rail for further processing (no further inspection). Any carcass not

stamped as fit for human consumption at the second inspection point would be sent to the

„detain‟ rail, where further rectification and trimming could take place. Any carcase rejection

conditions were recorded at the end of the detained line. These conditions are, therefore,

recorded in the system. There were two points where rectification could have been

performed before this inspection and recording process.

For logistical reasons, during the trial at the study abattoir, the visual-only inspection point for

the carcases and the recording of the abnormalities seen had to be positioned at the point of

the 1st traditional MHI inspection. This was before any rectification or trimming process.

There will, therefore, be a higher frequency of recording of certain conditions, such as hair

contamination, by visual-only inspection compared to traditional inspection. It might seem,

that visual-only inspection is able to detect a greater number of these conditions, but this

would be an incorrect assumption; usually those conditions were detected, resolved but

never recorded.

Figure A5:3: Figurative layout of inspection, detection and recording for both inspection methods at study

abattoir

Visual Only Inspection(Recorded)

2. Traditional Inspection(no recording)

1. Traditional Inspection(Recorded)

Detention Rail

Offal inspection(recorded)

Some rectification

Rectification

PigFlow

3. Traditional InspectionNo recording

Figure A5:4: Bird‟s-eye schematic of the layout of inspection, detection and recording for both inspection

methods at study abattoir

Visual Inspection of Carcase including recording of any conditions seen

Traditional First Inspection of Carcase (no recording here)

Removal of conditions such as hair and pluerisy by FBO operative (no

recording here)

Final Inspection and health marking of carcase or send

carcase for further rectification

Recification of carcase (including recording of conditions that have

gone down the detained line)


The role of the plant personnel positioned between the two points of inspections may need

some clarification: this operator had a role prior to the start of the trial in the study abattoir.

This operator removed any aesthetic conditions (mainly hair left on the carcases) to improve

the presentation of the carcase. They did not remove any conditions that affect the fitness of

the carcases for human consumption, so the final MHI carrying out the traditional inspection

was presented with all conditions affecting the safety of the meat.

The effect of this positioning can be seen in the outcomes from the field trial; for example, in

the visual-only inspection many occurrences of “faeces contaminations in any part except

tail” were recorded. This was due to no visual access to the back of the carcass; whereas in

the traditional inspection recordings are mainly of “faeces contamination in tail”. Any other

contamination (apart from tail) had been identified by the first MHI (at the visual-only and 1st

traditional inspection point) and dressed by the plant staff before the final MHI at the

traditional inspection point; therefore they were not recorded in the system. Contamination in

the tail area has not been detected by first MHI (and similarly at the visual-only point) and, as

a result it was not addressed before the final MHI‟s traditional inspection; hence, it is

recorded in the system.

We also noticed that there could be another reason as to why there was a higher prevalence

of hair recorded on carcases at visual inspection than at the traditional inspection. When

conducting traditional inspection, an MHI will usually have a routine e.g. incise the sub

maxillary lymph nodes first and then visually work their way up the carcase. It was noticeable

that, when conducting the visual inspection, the MHI tended to focus more at eye level. This

could possibly lead to more incidences of hair contamination being recorded.

In addition, we noticed during the field trial that there was a difference in how different MHIs

record certain conditions, or more specifically, the location of the conditions. For example,

conditions that were located in a small part of the carcase could be recorded either under the

specific area (such as „bile in belly‟) or under a general area (such as „bile in trim‟). During

the sampling and recording of the trial it was observed that it depended on the person. Some

of the MHIs were locating these conditions mostly in trim, and other MHIs were more precise

in the location. Both recording systems are fine but may interfere with any comparison of the

similarity or difference in the identification of conditions between visual-only and traditional

inspection methods. This problem was considered during the condition analysis (Objectives

1, 2, and 3) and resolved. Conditions were regrouped accordingly to the type of conditions

and not to the location of conditions. Once this was done, the inspection bias was controlled.

It will be essential to train, audit, assess, validate and evaluate the new inspection method

when implemented to ensure that it is applied consistently by all official food safety

personnel.

During the trial MHIs were not allowed to palpate or incise any carcass during the process of

the inspection. In theory, if visual-only inspection were implemented for outdoors pig

carcasses in UK abattoirs, palpation and incision of any part of the carcass will be

acceptable whenever the Officials suspect a condition that will need further investigation.

There were a few occasions during the field trial where the MHIs mentioned the fact that

he/she would progress to palpate and/or incise a specific area of the carcass after

suspecting a condition; for example, arthritis of the joints or to differentiate an abscess from

a tumour. As explained, no such further inspection took place during the trial. The outcomes


of the trial may not, therefore, reflect the reality of a visual & risk based system. It is probable

that the visual-only system as implemented during the field study is a “worst case scenario”

for the detection and observation of conditions in pig carcasses i.e. the potential of a visual &

risk based inspection system to detect conditions in pig carcases will be greater. However, in

the trial, one of the objectives was to compare the microbiological contamination due to

handling (palpation and incision) of carcasses during the traditional inspection and during the

visual-only inspection, in order to investigate whether contamination is lower when visual-

only inspection was implemented. The outcomes of this investigation may potentially reflect

a better situation than would exist in a visual & risk based inspection system, especially if

non-food safety trimming is carried out later by plant staff (ANNEX 3 & 4).

CONCLUSIONS AND RECOMMENDATIONS Conclusions

We have identified ten main areas where potential obstacles to the implementation of a risk-

based visual inspection system for fattening pigs may occur (Table A5:4). Adaptations

should be possible to overcome all of them; however further work may be required. Specific

areas include:

1.1. Certain lack of competence, skills and understanding of the new method of

post-mortem inspection (visual-only inspection) is expected. Training of all FSA

personnel for the new visual-only inspection is required and should suffice to

overcome this obstacle; to include subsequent audit, evaluation, inter and intra-

operator validation and revision of training and methods.

1.2. Use of tools might be needed to help Officials to carry out a whole carcase

visual inspection. These tools might be a rotated system or mirrors (so the back

of the carcase will be accessible to the visual inspection).

1.3. It is advised (EC, Anon. 2011) that all carcase handling and trimming by plant

staff, such as removal of tumours, should be done in a detained area to avoid

any cross contamination to other carcasses: however, the difficulty of such a

procedure is understood. It is mainly due to the busy nature and high speed of

the slaughter line combined with the number of carcases that are expected to

be detained, which is high. It could lead to delays and associated costs. This

could be partly solved if abattoir personnel are deployed in those detain rails

permanently and routinely perform these tasks. Alternatively they could be

sited on the line near the visual inspection point with adequate training to

minimise cross contamination between carcases.

1.4. Animal health and welfare associated information from carcases could be

collected with different methods that are already in place in some abattoirs,

such as the British Pig Health Scheme (BPHS) inspections; or new methods,

such as information from further inspection done by the plant. The return of this

information to producers is relevant for animal health and indirectly relevant for

public health as it can be used by farmers to improve their herd health status.


1.5. The result from the microbial study (objectives 1, 2, and 3 of Project FS145003)

could be used as the baseline recommended by the Scientific Committee

(Anonymous, 2000) for the study abattoir before introducing the new system.

Each abattoir will require a baseline to be established via a microbiological

survey of an appropriate design.

1.6. Revision of current HACCP plans will be needed in abattoirs to ensure that all

requirements necessary for the implementation of visual-only inspection are in

place and implemented to the relevant standards: auditing of all HACCPs in all

abattoirs slaughtering fattening pigs will be required. Again, these HACCP

plans must be individual for each abattoir.

The goal of modern meat inspection is to minimise the food-borne risk for humans

(Anonymous, 2011) (i.e. Food Safety). It is commonly accepted that it is not possible to

reduce the risk to zero; there will always be certain conditions missed by any method of

inspection. The impact of such omissions will be considered in the report for Objective 8.

Recommendations

A list of recommendations for practices that could be included in the abattoirs once the

visual-only inspection is applied is as follows:

i. Availability of the FCI in sufficient time to allow risk management decisions to be

made and the logistics to be organised, prior to the arrival of pigs for slaughter, is

essential. The FCI might contain information from the whole herd, as specific batches

to be sent to the abattoir might be unknown at the point of information submission.

ii. Information in the FCI should include the recent history of the herd. This should

include diseases and conditions of interest, such as Salmonella, or Erysipelas,

vaccination strategies, and welfare indicators e.g. lameness, tail biting) mortality rate

etc.

iii. Information of post-mortem inspection from previous batches killed at the same

abattoir from the same farm should be available and used by Officials.

iv. Officials at the abattoir should use the information from both the FCI and previous

post-mortem inspection results from each herd to categorise the batches of pigs prior

to arrival at the abattoir according to risk. This will enable the allocation of

appropriate resources to improve the quality and efficiency of the post-mortem

inspection and maintain the safety of the meat produced.

v. Re-arrangement of slaughtering strategies could include the slaughtering of high risk

batches (such as pigs coming from farm with a recent history of Salmonella) at the

end of the day, with a gap between previous batches (to minimise potential cross

contamination to low risk batches) or even in a particular day of the week; it may be a

pre-requisite that hearts are opened prior to post-mortem inspection in the line (e.g.

in cases of recent history of high prevalence of endocarditis from batches coming

from the same farm); extra plant staff in the line to trim conditions such as pleurisy

without disturbing the speed of the line and the work of the FSA inspectors if

likelihood of presence of pleurisy is expected to be high.


vi. Use of any tool available to allow access to the whole carcase (included the back of

the carcase) for the visual inspection, such as mirrors, rotating devises or inspectors

at both sides of the line. Alternatively, allow handling of carcases/offal when

necessary, e.g. handling of offal in batches with recent history of high prevalence of

milk spot, to improve the detection of milk spot.

vii. Communication with the FBO if high risk batches are expected to be slaughtered at a

particular day or time, allowing the abattoir management team to allocate staff and

resources to manage the potential increase of procedures such as preparation of

carcases/offal prior to official post-mortem inspection, trimming, detection of carcases

or further handling of carcases during the further preparing of the meat before

dispatch.

viii. Officials should be allowed minimal handling of carcases if the visual access to the

whole carcase/offal is impeded and not resolved with any alternative method. The

visual inspection of carcases/offal must be comprehensive and complete to ensure

the efficiency of the visual-only inspection. This additional handling must be kept to

the minimum and only when visual access is not possible by any other means.

ix. MHIs should be authorised to carry out minimal incision of carcases/offal when any

condition/pathology is suspect and require the incision for confirmation, such as

abscess. The process must be kept to the minimal and hygiene must be ensured to

avoid any cross contamination to other parts of the carcase/offal or adjacent

carcases/offal.

x. Any of the above should be based on appropriate evidence and agreed a priori by

risk managers and the industry to ensure that any mitigation measures and the

consequent resource implications are proportionate to the perceived/apparent risk.

Table A5:2: The ten potential obstacles to implementation of a risk-based visual meat inspection system for

fattening pig carcases and adaptations that may be required to overcome them.

Potential Obstacles Potential Adaptations

1. The development of appropriate criteria, on which relevant parties agree, in order to classify the risk status of batches of pigs is a potential obstacle to implementation of a risk-based inspection system

Further work is required to develop agreed protocols for the classification of batches of pigs according to the perceived or apparent risk to public health

2. The selection of the post-mortem method has to be made prior to animals arriving to the abattoir based on the risk of each batch

The improvement of the availability of Food Chain Information (FCI) to allow the official veterinarian and inspectors to gather enough information with sufficient time to define the risk of each batch, according to the agreed criteria, and to implement the appropriate post-mortem method and arrangements

3. A potential increase in the number of carcases in need of further handling before they are released to the chillers and the consequent resource implications, whether that be in terms of capital

Alterations to plant layout and staff resources to handle additional carcass dressing will need to be plant specific.


Potential Obstacles Potential Adaptations

investment and alterations in the layout of abattoirs, or plant staff numbers and training.

4. The implementation of mechanisms to ensure that detection of conditions by visual inspection is maximised (e.g. some parts of carcases/offal may not be accessible to the visual inspection) and the consequent resource implications.

Additional mechanisms to ensure access and visibility to carcases and offal will need to be plant specific. Risk managers will need to consider if the resource implications are proportionate to the perceived risk and the potential gain. Examples include changes to the layout of the abattoir, or plant staff tasks, such as introducing rotating tools or mirrors to allow access to all parts of carcases/offal or the preparation of offal (opening of hearts) prior to inspection

5. The need to transfer additional meat quality responsibilities to the FBO to deal with all the aesthetic conditions in carcases (such as hair contamination) that no longer fall within the food safety authorities remit

There are no immediate solutions to these obstacles; they will be plant-specific. Food business operators will need revise their HACCP plans to address these issues

6. The potential for an increase in microbial contamination of carcases during the plant procedures to handle and deal with the above aesthetic conditions (after official post-mortem inspection).

7. A perceived, or actual, loss of data on animal health and welfare from animals slaughtered in the abattoir

Further work to identify the utility and benefits of the use of alternative data sources, such as plant own records of detection found in the abattoir (e.g. HACCP records), followed by the implementation of appropriate systems where relevant.

8. FSA personnel - perception of possible loss of their jobs as a consequence of new system

It is outwith the scope of Project FS145003 to suggest methods of change management and how to achieve behavioural change; however, training (and subsequent auditing) of all FSA personnel involved with the new system will be needed to ensure an adequate understanding of the concepts and responsibilities that are involved and so to maintain a high standard of service.

9. Decrease of the FSA MHI service quality due to different factors, such as the lack of experience with the new post-mortem inspection method, or the decrease of attention during post-mortem inspection due to the monotony of the task.

10. A perceived, or actual, decrease in the quality of the Official Veterinarian work, such as identification of high risk batches in need of a post-mortem inspection with inclusion of further incision of palpation, due to the lack of experience with the new system.


REFERENCES

Anonymous, 2000. Opinion of the Scientific Committee on Veterinary Measures Relating to

Public Health on Revision of Meat Inspection Procedures. European Commission. Health &

Consumer Protection directorate-General. 24th February 2000

Anonymous, 2004a. Regulation (EC) No 853/2004 of the European Parliament and of the

Council of 29 April 2004 laying down specific hygiene rules for on the hygiene of foodstuffs.

Off. J. Eur. Comm., L139/55

Anonymous, 2004b. Regulation (EC) No 854/2004 of the European Parliament and of the

Council of 29 April 2004 laying down specific rules for the organisation of official controls on

products of animal origin intended for human consumption. Off. J. Eur. Comm., L155/206

Anonymous, 2007b. Commission Regulation (EC) No 1244/2007 of 24 October 2007

amending Regulation (EC) No 2074/2005 as regards implementing measures for certain

products of animal origin intended for human consumption and laying down specific rules on

official controls for the inspection of meat. Off. J. Eur. Comm., L281/12-18

Anonymous, 2011. Scientific Opinion on the public health hazards to be covered by

inspection of meat (swine) EFSA Panel on Biological Hazards (BIOHAZ) EFSA Panel on

Contaminants in the Food Chain (CONTAM) EFSA Panel on Animal Health and Welfare

(AHAW) European Food Safety Authority (EFSA), Parma, Italy. EFSA Journal 2011;

9(10):2351

Huey, R. 2012.Thoroughly modern meat inspection. Veterinary Record 2012 170: 68-70

Mousing et al., 1997 Meat safety consequences of implementing visual post mortem meat

inspection procedures in Danish slaughter pigs. Veterinary Records 140, 472-477. 1997

ANNEX 1 FINAL REPORT FOR PROJECT FS145003 HISTORICAL DATA …

Documents

Transcript of ANNEX 1 FINAL REPORT FOR PROJECT FS145003 HISTORICAL DATA …