ORIGINAL ARTICLE

A test of the Whitaker scoring system for estimating agefrom the bones of the foot

Catriona Davies & Lucina Hackman & Sue Black

Received: 18 August 2012 /Accepted: 26 September 2012 /Published online: 4 October 2012# Springer-Verlag Berlin Heidelberg 2012

Abstract Within the literature pertaining to skeletal ageestimation, there is a paucity of statistically validatedmethods of age estimation from the foot. Given theprevalence of recovery of pedal elements in isolation,it is critical that methods exist to facilitate the estima-tion of age from this anatomical region and that thosemethods be tested to ensure they are reliable, repeatableand statistically robust. A study was carried out todetermine the validity of using the Whitaker methodof age estimation from the bones of the foot as a toolin forensic age estimation within a modern Scottishpopulation. Two-hundred and sixty radiographs fromindividuals aged between birth and 18 years wereassessed according to the Whitaker method; the resultswere compared with chronological age. The results ofthis study suggest that the method of Whitaker et al. ishighly unlikely to estimate the age of females below16 years of age or males below 18 years of age cor-rectly. When the methodology was altered to correspondwith best practice principles of age estimation, the esti-mated age ranges were found to be too wide to be ofpractical value in forensic age estimation. The results ofthis study therefore suggest that the Whitaker methodfor estimating age from the bones of the foot should notbe used in forensic age assessment.

Keywords Forensic anthropology . Age estimation . Foot .

Skeletal age . Radiographs

Introduction

The estimation of age at death is one of the four keyparameters which can be established in the pursuit of thepositive identification of human remains. In conjunctionwith living stature, ancestral origin and biological sex, ageestimation allows investigators to narrow the scope of theirsearch of missing persons’ records and therefore may facil-itate the timely identification of the remains. Without cog-nisance of any one of these parameters, it becomesincreasingly difficult to assign a presumptive identity tothe recovered remains. As there is no control over thecondition of the remains presented for investigation, it isimperative that methods exist which are suitable for use incases of partial or fragmented remains.

The recovery of fragmented human remains necessitatesthe use of methods which allow the maximum amount ofinformation to be accessed in the establishment of identity[1]. The foot and ankle is a region of the skeleton that isfrequently recovered in isolation [2–5]. This may be broughtabout by the natural sequence by which remains disarticu-late in aquatic environments or in terrestrial biomes as aresult of human intervention, high-energy trauma or scav-enging [5]. In instances such as these, it is often due to theprotective and containing nature of footwear that skeletalelements of the lower limb, particularly those of the foot,remain associated and therefore available for analysis.

For age estimation to be of value in forensic assessments,the methods employed must be valid and based on statisti-cally robust methodologies [6]. This may require existingmethods to be reviewed regularly and, if necessary, alter-ations made to increase the scientific robusticity, reliabilityand validity of the method, thereby addressing the probativevalue of the resulting evidence [6].

As the remit of age assessment increasingly expands toincorporate living individuals, methods of forensic age

C. Davies : L. Hackman : S. Black (*)Centre for Anatomy and Human Identification,College of Life Sciences, University of Dundee,Dow Street,Dundee DD1 5EH, UKe-mail: s.m.black@dundee.ac.uk

Int J Legal Med (2013) 127:481–489DOI 10.1007/s00414-012-0777-4

estimation must be reviewed frequently, reassessed and test-ed on multiple and diverse populations to assure that stand-ards are maintained [1, 7, 8]. Due to the ease with which theanatomical region can be imaged and the high number ofdeveloping osseous centres from which skeletal maturation,and by extension chronological age, can be determined, themajority of recent research into age estimation methodolo-gies has focussed on the region of the hand and wrist [9–15].

In contrast to the hand and wrist, there is a paucity ofresearch relating to age estimation from the foot and ankle[16, 17]. Consequently, the methods used to assess age fromthis anatomical region have not necessarily been scientificallyvalidated. In the UK, recommendations made by the LawCommission of England and Wales in 2011 suggest that allresults proffered as evidence in criminal trials should bederived from methods that are based on statistically validand scientifically robust methodologies, enabling them tofulfil the requirements of evidentiary admissibility and proba-tive value [6]. It is therefore prudent to reevaluate methods ofscientific assessment which may be applied to forensic casesto ensure that only methods of confirmed validity are used.

As in other regions of the skeleton, estimation of age fromthe foot and ankle is primarily based on the maturation of theskeleton, including the appearance and fusion of ossificationcentres [16, 17]. In 2002, Whitaker et al. [16] produced ascoring system for use in the forensic assessment of age fromthe foot, the aim being to assess the chronological age of anindividual based on bone-specific maturity scores. The studysample included radiographs obtained from subadult patientsof podiatric clinics in San Francisco, California.

The scoring system devised by Whitaker et al. is complexand consists of three stages. The method incorporates onlythose bones which can be clearly visualised in both anterior–posterior and lateral radiographs of the foot, including thecalcaneus, all five metatarsals, and the proximal and distalphalanges [16]. The process is complex and therefore requiressome explanation. Three independent scores were assigned toeach bone according to (1) the degree of ossification of theprimary centre, (2) the degree of ossification of the secondarycentre and (3) the degree of fusion between the two centres.Tables 1 and 2 outline the criteria used by Whitaker et al. to

assign the scores to the primary and secondary centres ofossification and the degree of fusion between them, respectively[16]. The three independent scores were brought together toform a three-digit sequence which was used to attribute a finalmaturity score to each bone through the use of a pre-preparedtable (Table 3). This maturity score refers to a second table (notshown) which assigns an age range to each bone. For example,Fig. 1 illustrates the application of the method ofWhitaker et al.[16] to the proximal pedal phalanx. This bone would beassigned a score of 3 for the primary centre of ossification, a 2for the secondary centre of ossification and a 1 for the degree offusion between these centres. This would correspond to a bonematurity score of 4, which in turn relates to an estimated agerange of 48–84 months for this bone.

For each foot analysed, this process yields a series of 16 ageranges, one for each bone. Each of the age ranges provides ayounger and older estimated age. The final step in the system ofWhitaker et al. [16] involves using these age ranges to constructa composite age range from the highest younger estimated ageand the lowest older estimated age. It should be noted that dueto the sample size used in the development of the method byWhitaker et al. [16], there are some maturity scores which do

Table 1 Criteria for the assignment of ossification scores assigned byWhitaker et al. [16]

Score Criteria for primary and secondary centres

0 No bony material visible on X-ray due to artefact or film quality

1 Ossification centre absent

2 Ossification centre present, but rudimentary

3 Ossification centre present and mineralised bone resemblesadult shape

4 A fully adult bone (assumes complete fusion)

Table 2 Criteria for the assignment of fusion scores assigned byWhitaker et al. [16]

Score Criteria for fusion of ossification centres

0 Bones obscured or secondary centre unformed

1 Primary and secondary centres far apart

2 Primary and secondary centres approximate one another,but still separated by radiolucent band

3 Primary and secondary centres in partial fusion;radio-opaque connections present with radiolucent spaces

4 Primary and secondary centres in full contact;no radiolucent spaces but epiphyseal line still apparent

5 Fully adult fusion; no epiphyseal line apparent

Table 3 Maturity scoresderived from ossificationand fusion scores [16]

Primaryscore

Secondaryscore

Fusionscore

Maturityscore

0 0 0 X

1 1 0 X

2 1 0 1

2 2 1 2

3 1 0 3

3 2 1 4

3 2 2 5

3 3 1 6

3 3 2 7

3 3 3 8

3 3 4 9

4 4 5 10

482 Int J Legal Med (2013) 127:481–489

not have associated data. In addition, some maturity scores donot provide age ranges for all bones considered by the methodas their data are incomplete.

Despite almost a decade since its publication, the methoddevised by Whitaker et al. [16] does not appear to have beenretested, and consequently its accuracy and precision arelargely unchallenged. As so few methodologies for age esti-mation from the foot exist, it is imperative that those that havebeen published are retested for validity, reliability and accura-cy. The objectives of this study were therefore to test theaccuracy and reliability of the scoring system of Whitaker etal. [16] for estimating age from the bones of the foot in amodern Scottish sample and to assess the validity of its use asa method for forensic assessment of chronological age.

Materials and methods

Radiographs from 260 individuals (139 males, 121 females)were obtained from Ninewells Hospital, Dundee, with agesranging from birth to 18 years. The chronological age for

each individual was calculated by subtracting the date ofbirth from the date on which the image was taken. Thedistribution of the sample by sex and chronological age (inmonths) is presented in Fig. 2.

This analysis represents the application of the method inits original state to a modern Scottish population and istherefore constrained by the limitations of the source publi-cation and the sample upon which the method was devel-oped. As the method of Whitaker et al. [16] presents sex-specific age ranges, the radiographs were separated andanalysed according to sex. Only individuals for whom acomplete assessment could be undertaken were included inthe study sample. Each radiograph was assessed withoutknowledge of the individual’s chronological age.

The images used in the study were obtained from patientsadmitted to the Accident and Emergency or Clinical Radiologydepartments at Ninewells Hospital and are considered represen-tative of a cross-section of the subadult population of Tayside.The radiographs were taken for the purposes of clinical assess-ment; ethical approval was obtained fromNHSTayside for theirinclusion in this research as anonymous subjects.

Initially, composite age ranges were constructed accord-ing to the method devised by Whitaker et al. [16]. Thisrequired the composite age range for an individual to becomposed of the highest younger age and the lowest olderage from each range obtained from the assessment of theindividual bones of the foot. According to Whitaker et al.[16], this approach was taken to maintain consistency in themethodology with other forensic applications of bonegrowth studies. The authors of this study, however, areunaware of any other methods that follow such an approach.

A second composite age range was constructed for eachindividual from the youngest possible age and the oldestpossible age according to the ranges assigned to the 16bones included in the method. This resulted in the construc-tion of a composite age range which included all possibleages for each maturity score. The composite age rangesgenerated through this approach will be referred to as the‘adjusted composite age range’.

Fig. 1 An example of a proximal phalanx of a left first pedal raywhich would be assigned a three-digit code of 3, 2, 1 corresponding toa bone maturity score of 4

0

5

10

15

20

25

30

35

40

0-23 24-47 48-71 72-95 96-119 120-143144-167 168-191192-216

Nu

mb

er o

f In

div

idu

als

Chronological Age (months)

Females

Males

Fig. 2 Study sample bydistribution by sex and age (inmonths)

Int J Legal Med (2013) 127:481–489 483

Using both sets of composite age ranges (original andadjusted), the percentage of individuals whose chronologi-cal ages were included in the estimated age range wascalculated for the complete sample and in 24-month cohorts.The minimum and maximum age ranges generated by eachapproach were also calculated.

Intra-observer analysis

Intra-observer variation was assessed using a randomly se-lected subset comprising 33 radiographs (16 females and 17males). Each image was reassessed and the percentageagreement between the maturity score assigned to the 16bones for each individual calculated. The second round ofanalysis was carried out in excess of 4 months after the firstround and was completed without knowledge of the chro-nological ages of the individuals in the subset.

Inter-observer analysis

A subset of 28 individuals (14 males and 14 females) wasselected randomly by the first observer (CD) and was assessedby the second observer (LH) to determine the consistency ofthe method between observers. The percentage agreementbetween the maturity scores assigned to the 16 bones for eachindividual was calculated. The assessments were conductedby the second observer (LH) without knowledge of the matu-rity scores assigned by the first observer (CD) or the chrono-logical ages of the individuals within the sample.

Results

Intra-observer analysis

An assessment of the maturity scores assigned to a subsampleof female and male individuals was conducted to determine theconsistency of the method through the analysis of variation inresponses from a single observer. The percentage agreement

and the degree of variation in the scores assigned to eachindividual by the observer on the first and second occasionswere calculated. These results are presented in Fig. 3.

The results obtained from these analyses show that for61.72 % of the female individuals and 64.42 % of the maleindividuals, the scores assigned were identical.

Of the 98 bones from females where the scores differed,76.53 % of disagreements were within a single score, 21.43 %were within two scores, and 1.02 % within three scores. Onlyon one occasion was the difference between the scores assignedon the first and second rounds of assessment >3. Of the 74bones from males where variance in the assigned scores wasrecorded, 63.51 % of disagreements were within a single score,31.08 % varied by two scores, and 5.41 % varied by threescores. A one-way analysis of variance on ranks was performedto determine the statistical significance of the difference be-tween the maturity scores assigned at the first and secondrounds of assessment. The difference between groups was notfound to be statistically significant (H00.716, P00.397).

Inter-observer analysis

Analysis of the data obtained from the assessments carriedout by two observers was undertaken to determine theconsistency and, consequently, the reliability of the method.The degree of disagreement between the maturity scoresattributed by the two observers was recorded and the per-centage agreement was calculated. The results are presentedin Fig. 4.

The results yielded identical scores for 54.46 % of femalesand 45.98 % of males. Of the 102 bones from female subjectsfor whom the scores attributed by the observers varied, themajority differed by between one and three scores. The resultsshowed that 41.2 % of disagreements were discordant by asingle score, 33.3 % by two scores and 12.7 % by three scores.The remaining discordances (12.7 %) varied between four andfive stages, although these represent the minority of the sample.Within the male subject sample, the assigned maturity scoreswere found to disagree in 121 bones. Of these disagreements,

0.00

10.00

20.00

30.00

40.00

50.00

60.00

70.00

0 ±1 ±2 ±3 ±4 ±5

Per

cen

tag

e A

gre

emen

t

Difference between Scores

Females

Males

Fig. 3 Percentage agreementachieved during the intra-observer analysis

484 Int J Legal Med (2013) 127:481–489

25.6 % differed by a single score, 43 % by two scores and31.4 % by three scores.

A one-way analysis of variance on ranks was performedto determine the significance of the differences found be-tween the percentage agreements achieved for the femaleand male subject groups within the inter-observer assess-ments. These results showed that the differences betweenthe percentage agreements achieved for females and malesas part of the inter-observer analysis were not significantlydifferent (H00.0257, P00.937).

A one-way analysis of variance of ranks was performedto determine the significance of the differences found be-tween the percentage agreements within the intra- and inter-observer assessments. The results of this analysis showedthat there was no significant difference between the resultsobtained for percentage agreement found within the intra-observer and the inter-observer analyses (H00.336, P00.562).

Females

The nature of the method devised by Whitaker et al. [16]necessitates the use of an age range rather than a single-value assigned age±a standard deviation. Therefore, theanalysis of the results obtained from the evaluation of radio-graphs using this method involved the calculation of thepercentage of individuals whose chronological ages fellwithin the assigned age range and not a direct comparisonof an estimated age within a chronological age.

In their method, Whitaker et al. [16] directed that com-posite age ranges should be constructed from the highestyounger estimated age and the lowest older estimated age.Table 4 presents the analysis of the results when the com-posite age ranges were constructed according to the methodof Whitaker et al. [16]. The maximum span of the Whitakercomposite age ranges was 73 months, corresponding to6 years and 1 month, whilst the minimum range was2 months. Although the maximum age range would appear

to be of sufficient width to include the chronological age ofan individual, the percentage of individuals whose chrono-logical ages were included in the assigned age range acrossthe female sample was only 50.4 %.

Analysis of the composite age ranges suggested that themethod of Whitaker et al. [16] is more likely to assign acomposite age range which is inclusive of the chronologicalage to individuals over the age of 144 months than toindividuals between birth and 144 months. Only two cohortswere shown to exhibit complete inclusion of chronologicalage within the estimated age range; however, one of thesewas represented by a single individual and therefore shouldnot be considered a reliable predictor of application of themethod to that age group (48–71 months).

To adapt the method of Whitaker et al. [16] to conform toforensic principles as outlined by Ritz-Timme et al. [1], asecond set of composite age ranges was constructed usingthe youngest and oldest ages from the ranges calculated foreach of the skeletal elements considered by the sourcemethod. Table 5 presents the percentage of individuals

0.00

10.00

20.00

30.00

40.00

50.00

60.00

0 ±1 ±2 ±3 ±4 ±5

Per

cen

tag

e A

gre

emem

tDifference between Scores

Females

Males

Fig. 4 Percentage agreementachieved during the inter-observer analysis

Table 4 Percentage inclusion of chronological age within the com-posite age ranges of Whitaker et al. [16] for females

Age cohort(months)

Percentageinclusion ofchronologicalage

Max rangespan (months)

Min rangespan (months)

0–23 (n01) 0 8 8

24–47 (n07) 0 46 12

48–71 (n01) 100 21 21

72–95 (n08) 12.5 26 0

96–119 (n016) 40 51 2

120–143 (n025) 30.8 53 2

144–167 (n026) 57.7 73 2

168–191 (n016) 56.25 70 11

192–216 (n021) 100 70 70

Total (n0121) 50.4 73 2

Int J Legal Med (2013) 127:481–489 485

whose chronological ages were included within the adjustedcomposite age ranges in addition to the minimum and max-imum number of months represented by the adjusted com-posite age ranges.

When the results were considered across the entire sam-ple, the chronological ages of 98.4 % of individuals wereincluded in the estimated age ranges. This increase in theinclusion rate of the chronological age, compared to theunadjusted composite age ranges, can be explained by theincrease in the span of the adjusted composite age rangescompared with those constructed according to the sourcemethod.

The maximum span of the adjusted composite age rangesfor the female sample was 182 months and the minimumwas 88 months, which correspond to 15 years and 2 months,and 7 years and 4 months, respectively. These results indi-cate that although the manner in which the adjusted com-posite age ranges were constructed is in accordance withgood practice for forensic age estimation, the maximum andminimum spans of the age ranges within each cohort areunacceptably wide, resulting in age ranges of no practicalvalue.

Males

The construction of composite age ranges was carried outaccording to the method prescribed by Whitaker et al. [16].Table 6 presents the percentage of male individuals whosechronological ages were included in the assigned compositeage range and the maximum and minimum age rangesassigned to individuals within each 24-month cohort. Al-though these ranges suggest that a high degree of concor-dance is present within the assigned age ranges, the meanpercentage of individuals whose chronological ages fellwithin the assigned range was only 21.6 %.

The results also indicate that individuals between birthand 119 months of age were likely to be represented bywider age ranges. A maximum age range of 40 months(3 years and 4 months) was found in the 96- to 119-monthcohort, corresponding to individuals aged 8 or 9 years. Theminimum range of 1 month was recorded in the 168- to 191-month and 192- to 216-month cohorts, which correspond toindividuals between 14 and 18 years of age.

The currently accepted principles of age estimation re-quire an age range to be of sufficient inclusivity to accom-modate all possible ages for which there is osteologicalevidence whilst maintaining exclusivity in order to isolateany erroneous data which cannot be supported by scientificdata [1]. Table 7 presents the percentage inclusion of thechronological age within the adjusted composite age ranges.

These results suggest that the adjusted age ranges includethe chronological age of 98.6 % of the sample. The only

Table 5 Percentage inclusion of chronological age within the adjustedcomposite age ranges for females

Age cohort(months)

Percentageinclusion ofchronologicalage

Max rangespan (months)

Min rangespan (months)

0–23 (n01) 100 116 116

24–47 (n07) 71.4 126 88

48–71 (n01) 100 147 147

72–95 (n08) 100 182 117

96–119 (n016) 100 182 126

120–143 (n025) 100 182 126

144–167 (n026) 100 182 126

168–191 (n016) 100 180 157

192–216 (n021) 100 157 157

Total (n0121) 98.4 182 88

Table 6 Percentage inclusion of chronological age within the com-posite age ranges of Whitaker et al. [16] for males

Age cohort(months)

Percentageinclusion ofchronologicalage

Max rangespan (months)

Min rangespan (months)

0–23 (n00) 0 0 0

24–47 (n012) 20 37 6

48–71 (n07) 20 37 6

72–95 (n07) 14.3 20 9

96–119 (n012) 16.7 40 2

120–143 (n012) 8.3 21 2

144–167 (n035) 36.1 24 2

168–191 (n034) 5.9 23 1

192–216 (n020) 35 10 1

Total (n0139) 21.6 40 1

Table 7 Percentage inclusion of chronological age within the adjustedcomposite age ranges for males

Age cohort(months)

Percentageinclusion ofchronologicalage

Max rangespan (months)

Min rangespan (months)

0–23 (n00) 0 0 0

24–47 (n012) 18.2 147 78

48–71 (n07) 100 147 78

72–95 (n07) 100 168 141

96–119 (n012) 100 170 92

120–143 (n012) 100 170 127

144–167 (n035) 100 170 127

168–191 (n034) 100 189 119

192–216 (n020) 100 159 119

Total (n0139) 98.6 189 78

486 Int J Legal Med (2013) 127:481–489

cohort in which the chronological age was not included in theestimated age range in every individual was that includingindividuals between 24 and 47 months (2–4 years). Althoughthese results suggest that the adjusted composite age rangesare accurate, the mean number of months included in theranges suggests that they are of no practical value. The min-imum age range was 78 months (6 years, 6 months) and themaximum represented 189 months (15 years, 9 months).

Although the chronological age of a high percentage of maleindividuals fell within their assigned age range, the adjustedcomposite age ranges are prohibitively wide and therefore areof no value within the context of forensic age estimation.

Discussion

The assignment of a biological profile to unidentified hu-man remains necessitates the analysis of biological param-eters including that of chronological age. The methods usedto determine these characteristics are dependent on the skel-etal elements present and the degree to which the bone hasundergone taphonomic change. Regardless of the quantityof the remains recovered, the estimation of chronologicalage is a critical component of all anthropologicalexaminations.

The results obtained from this study suggest that themethod devised by Whitaker et al. [16] should not beapplied to individuals of either sex who are believed to beyounger than 12 years. In addition, due to the age at whichfemales attain full skeletal maturity in the foot, the methodof Whitaker et al. [16] cannot be applied to female individ-uals over the age of 16 years. At no point do the inclusionrates of chronological age for male individuals reach 50 %.The method of Whitaker et al. [16] is therefore unsuitablefor use in male individuals.

Through the establishment of a biological profile and theanalysis of chronological age, missing persons’ databasesmay be searched and a putative identity can be confirmed orrefuted. For the estimated age to be considered viable, itmust assume the form of a range of values which encompassall possible chronological ages based on the maturationalstage of the bone whilst maintaining a degree of exclusivitywhich restricts the range, thereby increasing its precisionand, hence, practical value. The age ranges developed fromthe scores assigned through the application of the method ofWhitaker et al. [16] were constructed from the highestyounger age and the lowest older age assigned to the indi-vidual bones of the foot, thereby creating the minimumcomposite age range. The results of this study, outlined inTables 4 and 6 for females and males, respectively, showthat the age ranges composed using the method outlined byWhitaker et al. [16] are of insufficient accuracy to be appliedin a forensic context. The only cohort in which a significant

level of consistency was exceeded was that containing fe-male 16- to 18-year-olds. This, however, should be treatedwith caution as skeletal maturation of the foot is largelycomplete in female individuals by 16 years of age. It istherefore not surprising that 100 % inclusion was found inthis cohort.

Similarly, the highest percentage inclusion within themale cohorts was observed in the 16- to 18-year group.Unlike females, the pedal skeleton of males between theseages is likely to be in the final stages of maturation. Thiscould explain why the percentage inclusion within thiscohort is lower than in the equivalent female group. It doesnot, however, explain why only 35 % of the chronologicalages were included in the estimated age ranges generated bythe method. With the exception of the 16- to 18-year cohort,the estimated age ranges only included the chronologicalages of individuals within two other groups (8- to 10-yearand 12- to 14-year cohorts).

The evaluation of the chronological age of skeletalremains requires the application of a method of assessmentwhich is equally applicable to all stages of maturation. Thepresence of age bias within a method of assessment placeslimitations on its suitability of application to forensic case-work. Any method proffered for use in forensic assessmentshould therefore result in a high percentage of correct clas-sification of chronological age when applied to a sample ofmixed age groups. This study found that when compositeage ranges were constructed according to the method pre-scribed by Whitaker et al. [16], the percentage of individualswhose chronological ages were included in the estimatedage range was insufficient to meet the requirements ofreliability and accuracy. As a consequence, the probativevalue of this method must be brought into question.

The data, presented in Tables 4 and 6 for females andmales, respectively, suggest that the low percentage inclu-sion of the chronological age within the estimated age rangeis a reflection of the narrow age ranges attributed by themethod of Whitaker et al. [16] to both sex groups. As theapproach taken in the source study was to minimise the ageranges attributed to each bone through the use of the oldestyounger age and the youngest older age, the potential forerror was increased. This is therefore an area in which theapproach taken by Whitaker et al. [16] can be improved toconcur with current forensic age estimation principles ofinclusivity and exclusivity as outlined by Ritz-Timme etal. [1].

Whilst the use of the minimum composite age range inplace of the maximum range appears to provide narrow ageranges, this approach runs the risk of creating age rangeswhich exclude the chronological age of the individual. Todetermine whether this occurred in this study, a second setof composite age ranges was constructed using the lowestand highest estimated ages from each bone, i.e. the

Int J Legal Med (2013) 127:481–489 487

maximum composite age range. The percentage inclusion ofthe chronological ages within these ranges was then calcu-lated. Tables 5 and 7 present the results for the female andmale samples, respectively.

It is immediately apparent that for both the female andmale samples, the percentage of individuals whose chrono-logical ages were included within the adjusted compositeage ranges (Tables 5 and 7) increased with respect to boththe cohort and total groups compared with the unadjustedcomposite age ranges of Whitaker et al. [16] (Tables 4 and6). This increase in inclusion can be attributed to the in-crease in the mean span of the adjusted composite ageranges compared with those constructed according to themethod detailed in the source publication. It is important tonote, however, that the number of months included withinthe adjusted composite age ranges renders them impracticalfor use in forensic age estimation.

For the results of forensic assessments to be consideredof probative value, it is crucial that the method used isreliable and repeatable. As the method of Whitaker et al.[16] generates age ranges for 16 bones, the repeatability ofthe method was determined by calculating the percentageagreement between the maturity scores assigned to all 16bones included in the method by the two observers.

These results have potential implications for the role ofthis method within forensic anthropology as it pertains tolegal admissibility. The absence of a significant differencebetween the intra-observer and inter-observer assessmentssuggests that the experience of the observer has little or noeffect on the application of the method. Although this wouldenable forensic practitioners to apply the method regardlessof their level of experience in its application, the low per-centage inclusion of the chronological age within the ageranges and the lack of a significant difference betweenobservers or observations suggest that even experiencedforensic practitioners or individuals with significant experi-ence in the application of the method are unlikely to assignan age range which is inclusive of the chronological age.There is therefore insufficient consistency and reliabilitywithin the method for it to be accepted as a method offorensic age estimation.

Conclusion

Age estimation is one of the four key parameters used to aidin the identification of unknown human remains. Unlikeother regions of the skeleton such as the hands and wrists,there is a paucity of methods based on the assessment of thefoot and ankle [16–19]. One such method is the ‘ScoringSystem for Estimating Age from the Foot Skeleton’ byWhitaker et al. [16]. As there are few methods availablefor estimating age from the foot, it is imperative that those

which exist are tested and validated whether they are to beused to assist forensic investigators.

According to current forensic principles, the methods ofage estimation should generate age ranges which are wideenough to include all possible ages based on the level ofskeletal development whilst being sufficiently narrow toexclude chronological ages for which there is no osteolog-ical evidence [1]. The results of this study suggest thatwhilst the composite age ranges generated from the originalmethodology of Whitaker et al. [16] fulfilled the require-ment of exclusivity, they were not sufficiently inclusive ofthe chronological age to warrant their use in cases of foren-sic identification. In contrast, the adjusted age ranges weretoo wide to yield any precise information regarding thechronological age of the individual.

There was no significant difference between the percent-age agreement found in the intra-observer and inter-observerassessments, and neither of the intra-observer nor the inter-observer agreement was sufficient to support the use of thismethod. The results of this study suggest that the method byWhitaker et al. [16] does not generate age ranges which areof sufficient accuracy or reliability to be applied to casesunder investigation. Based on the results of this study,therefore, it is suggested that the scoring system for estimat-ing age from the foot by Whitaker et al. [16] is not appro-priate for use in forensic age estimation and that otheravenues should be pursued in cases which require an eval-uation of skeletal age from the pedal skeleton.

References

1. Ritz-Timme S, Cattaneo C, Collins MJ, Waite ER, Schütz HW,Kaatsch HJ, Borrman HIM (2000) Age estimation: the state of theart in relation to the specific demands of forensic practise. Int JLegal Med 113(3):129–136

2. British Broadcasting Corporation (2008) ‘Foot’ discovered onCanada shore. http://news.bbc.co.uk/1/hi/world/americas/7726165.stm. Accessed 10 August 2012

3. British Broadcasting Corporation (2008) Fifth human foot found inCanada. http://news.bbc.co.uk/1/hi/world/americas/7458468.stm.Accessed 10 August 2012

4. Gunn I (2008) Foot find confuses Canadian police. http://news.bbc.co.uk/1/hi/world/americas/7418239.stm. Accessed 10August 2012

5. Haglund W, Sorg M (2002) Human remains in water environ-ments. In: Haglund WD, Sorg MH (eds) Advances in forensictaphonomy: method, theory, and archaeological perspectives.CRC, London

6. Commission TL (2011) Expert evidence in criminal proceedings inEngland and Wales. United Kingdom Government, London

7. Schmeling A, Olze A, Reisinger W, Rösing FW, Geserick G(2003) Forensic age diagnostics of living individuals in criminalproceedings. HOMO—J Comp Hum Biol 54(2):162–169

8. Cunha E, Baccino E, Martrille L, Ramsthaler F, Prieto J,Schuliar Y, Lynnerup N, Cattaneo C (2009) The problem ofaging human remains and living individuals: a review. Foren-sic Sci Int 193(1–3):1–13

488 Int J Legal Med (2013) 127:481–489

9. Büken B, Erzengin B, Büken E, Safak AA, YazIcI B, Erkol Z (2009)Comparison of the three age estimationmethods: which ismore reliablefor Turkish children? Forensic Sci Int 183:103.e101–103.e107

10. Cameriere R, Ferrante L, Mirtella D, Cingolani M (2006) Carpals andepiphyses of radius and ulna as age indicators. Int J Legal Med 120(3):143–146

11. Mentzel H-J, Vilser C, Eulenstein M, Schwartz T, Vogt S, BöttcherJ, Yaniv I, Tsoref L, Kauf E, Kaiser W (2005) Assessment ofskeletal age at the wrist in children with a new ultrasound device.Pediatr Radiol 35(4):429–433

12. Schmeling A, Schulz R, Danner B, Rösing F (2006) The impact ofeconomic progress and modernization in medicine on the ossifica-tion of hand and wrist. Int J Legal Med 120(2):121–126

13. Tanner JM, Whitehouse RH, Marubini E, Resele LF (1976) Theadolescent growth spurt of boys and girls of the HarpendenGrowth Study. Ann Hum Biol 3:109–126

14. Tanner JM, Whitehouse RH, Healy MJR (1962) A new system forestimating skeletal maturity from the hand and wrist with standards

derived from a study of 2600 healthy British children. Part II. Thescoring system. International Child Centre, Paris

15. Tanner JM, Healy MJR, Goldstein H, Cameron N (2001) Assess-ment of skeletal maturity and prediction of adult height (TW3method). Saunders, London

16. Whitaker JM, Rousseau L, Williams T, Rowan RA, Hartwig WC(2002) Scoring system for estimating age in the foot skeleton. AmJ Phys Anthropol 118(4):385–392

17. Hoerr NL, Pyle SI, Francis CC (1962) Radiographic atlas ofskeletal development of the foot and ankle: a standard of reference.Charles C. Thomas, Springfield

18. Greulich WW, Pyle SI (1959) Radiographic atlas of skeletaldevelopment of hand and wrist. Stanford University Press,California

19. Groell R, Lindbichler F, Riepl T, Gherra L, Roposch A, Fotter R(1999) The reliability of bone age determination in Central Euro-pean children using the Greulich and Pyle method. Br J Radiol 72(857):461–464

Int J Legal Med (2013) 127:481–489 489