The Journal of Arthroplasty Vol. 20 No. 7 2005

Does Body Mass Index Affect The Early Outcome of

Primary Total Hip Arthroplasty?

Matthew Moran, MRCSEd, P. Walmsley, MRCSEd,A. Gray, MRCS, and I.J. Brenkel, FRCS

From the DepartmHospital, Dunfermline

Submitted NovemNo benefits or fuReprint requests

Gardens, EH10 5JSn 2005 Elsevier0883-5403/05/19doi:10.1016/j.art

Abstract: There is little evidence describing the influence of body mass index on the

outcome of total hip arthroplasty (THA). Eight hundred patients undergoing primary

cemented THA were followed for a minimum of 18 months. The Harris Hip Score

(HHS) and Short Form 36 were recorded preoperatively and at 6 and 18 months

postoperatively. In addition, other significant events were noted, namely death,

dislocation, reoperation, superficial and deep infection, and blood loss. Multiple

regression analysis was performed to identify whether body mass index (BMI) was

an independently significant predictor of the outcome of THA. No relationship was

seen between the BMI of an individual and the development of any of the

complications noted. The HHS was seen to increase dramatically postoperatively in

all patients. Body mass index did predict for a lower HHS at 6 and 18 months. This

effect was small when compared with the overall improvements in these scores.

There was no influence on the Short Form 36 component scores. On the basis of this

study, we can find no justification for withholding THA solely on the grounds of BMI.

Key words: body mass index, total hip arthroplasty, Harris Hip Score.

n 2005 Elsevier Inc. All rights reserved.

Total hip arthroplasty (THA) provides long-lasting

improvement in quality of life and reduction in

pain for patients with disabling arthritis. However,

there are groups of patients that have been shown

to have outcomes that are poorer than the general

population. A poorer outcome may be affected by

the underlying diagnosis, for example, femoral

neck fracture [1]; by the choice of implant, for

example, the Capital THA [2]; or by the surgeon,

for example, infrequently performed THA [3].

866

ent of Orthopaedic Surgery, Queen Margaret, Fife, UK.ber 18, 2003; accepted February 3, 2005.

nds were received in support of the study.: Matthew Moran, MRCSEd, 19 PlewlandsEdinburgh, UK.Inc. All rights reserved.06-0004$30.00/0

h.2005.02.008

There are concerns that an increasing body mass

index (BMI) negatively impacts on the outcome

of THA, and surgeons may decline to operate

on the obese for fear of the complications that

may follow. Possible areas of increased complica-

tions include increased length of operative time

[4,5], venous thromboembolism [6], superficial

and deep wound infection [7], increased blood loss

[8,9], and aseptic loosening due to increased

loading through the THA.

Despite the theoretical increased rate of compli-

cations, there is evidence to show that the symp-

tomatic relief after THA is as effective in the obese as

in thinner patients [10,11]. In 2000, the UK National

Audit Office criticized orthopedic surgeons for the

use of bvarying criteria for weight above which they

may not operateQ [12]. It is important that decisions

about the suitability of patients for surgery are made

on good evidence. We set out to examine the early

complication rate in obese patients after THA.

0

50

100

150

200

250

300

350

<20 20-24.9 25-29.9 30-39.9 40+

Body Mass Index

Nu

mb

er

of

pati

en

ts

Fig. 1. Distribution of patients by obesity category.

Does Body Mass Index Affect The Early Outcome of Primary THA? ! Moran et al 867

Methods

Patients

Eight hundred consecutive patients undergoing

Charnley primary THA (De Puy International,

Leeds, UK) were investigated prospectively. The

patients were under the care of 6 consultant

orthopedic surgeons at a single hospital. The joint

arthroplasties were carried out between January

1998 and November 2000. A standard anterolateral

approach to the hip was used by all surgeons. Data

from the patients were collected by a specialist

nurse and stored on a local database.

The following events were recorded: length of

stay in the hospital, death, dislocation, reoperation,

superficial and deep wound infection, blood loss,

and transfusion requirement. Superficial wound

infection was diagnosed in the presence of dis-

charge from the surgical wound or spreading

cellulitis and a positive microbiological culture of

a microorganism known to be implicated in causing

wound infection. Deep infection was suspected on

clinical and radiological grounds but only diag-

nosed after the growth of putative microorganisms

from specimens taken at reoperation. Blood loss

was calculated from perioperative losses (suction

and swabs) plus postoperative drainage.

Concomitant medical problems were recorded

under the headings: smoker, cancer, atherosclero-

sis, cardiac, diabetes mellitus, osteoporosis, and

thromboembolism.

Outcome Measures

The Harris Hip Score (HHS) and Short Form 36

(SF-36) were the primary outcome measures used

[13,14]. The HHS combines scores for pain, func-

tion, activities, absence of deformity, and range of

motion to produce an overall score out of 100

(0, bad; 100, good). The score is mostly determined

by feedback from the patient and weighted strongly

toward pain and function. It has been shown to

be a reliable indicator of patient function and pain

before and after THA [15]. The SF-36 is a widely

used measure of patient health that is not specific

to one disease. It is a good measure of patient

symptoms after THA [16]. Both scores were

completed 7 days preoperatively as well as 6 and

18 months postoperatively.

Body mass index was taken as a marker of

obesity. It is calculated from the weight (kilograms)

divided by the square of the height (meters). The

BMI is widely recognized as a tool for the simple

calculation of obesity. It corrects the weight of the

patient for their height. A score is generated with

20 to 24.9 kg/m2 reflecting ideal weight, 25 to

29.9 kg/m2 reflecting overweight, 30 to 39.9 kg/m2

reflecting obesity, and 40 kg/m2 or greater reflect-

ing morbid obesity.

Statistics

The SPSSv9.0 (SPSS Inc, Chicago, Ill) computer

package was used to analyze results.

The paired Student t test was used to detect

changes in the HHS before and after surgery.

Univariate analysis was performed using v2 tests,

2-sample t tests, or Pearson’s correlation coefficient

to identify significant predictors of the measured

outcomes. The predictors were operating surgeon,

age, sex, side of surgery, length of stay in hospital,

concomitant medical problems (see above), blood

loss, transfused units of blood, and preoperative

HHS. The outcomes were reoperation, death,

dislocation, deep and superficial infection, and

HHS. Once possibly significant predictors of out-

come had been identified by this method, stepwise

multiple regression analysis was carried out to

identify any predictors that would independently

alter outcome. Multiple logistic regression was

performed for binary outcomes (eg, death) and

multiple regression linear analysis for continuous

variables (eg, HHS).

Results

Eight hundred total hip arthroplasties were

carried out in 759 patients. Sixty-one percent were

female and 39% male. Four hundred fifty-nine

THAs were left sided. The mean age was 68 years.

Of the 800 THA episodes, all completed a preoper-

ative HHS and SF-36. Seven hundred seventy-four

completed an HHS/SF-36 at 6 months and 687 com-

pleted the scores at 18 months. The mean BMI was

868 The Journal of Arthroplasty Vol. 20 No. 7 October 2005

27.8 kg/m2 (range, 17-49) with a standard devia-

tion of 5 (Fig. 1). There was no difference in the

BMI of patients defaulting to follow-up and those

who completed follow-up ( P = .32). On average,

0.7 units of blood were transfused and mean blood

loss was 537 mL (SD, 296 mL). The mean length of

stay was 10 days.

Thirty-three patients had died by the 18-month

follow-up (39 hips). There had been 13 dislocations

(at an average of 15 weeks). Fifteen patients

underwent a further operation (not including

reduction of a dislocated joint). Three revision

operations, 11 debridements, and 1 posterior lip

augment were carried out within the first 18 months.

Seven deep infections and 56 superficial wound

infections had occurred by 18 months.

The mean preoperative HHS was 42. This im-

proved to 81 at 6 months and 85 at 18 months

postoperatively. There was a significant improve-

ment in the HHS scores at 6 and 18 months when

compared with the preoperative score ( P b .0001).

Univariate analysis suggested that BMI might

predict for increased rates of superficial infection

and a lower HHS at 6 and 18 months postopera-

tively (all P b .05). However, once multiple logistic

regression was carried out, BMI was not found to

be a significant independent predictor of superficial

wound infection.

When multiple regression analysis was per-

formed for the HHS at 6 and 18 months, taking

into account other significant predictors, BMI was

still found to be significant ( P = .02 at 6 months and

P b .001 at 18 months). To calculate the individual

effect a change in BMI might have on HHS, the

0

5

10

15

20

25

30

Baseline

HHS

Length stay Comorbidity Drop Hb BMI

% v

ari

an

ce e

xp

lain

ed

HHS 6 HHS 18

Fig. 2. The relative contributions (percentage of variance

explained) of predictors to the HHS at 6 and 18 months.

Comorbidity indicates coronary and thromboembolism;

Hb, hemoglobin; HHS 6, Harris Hip Score at 6 months;

HHS 18, Harris Hip Score at 18 months.

multiple regression coefficient b was noted. At

6 months, b = �.25 (95% confidence intervals

[CIs], �.05 to �.45), and at 18 months, b = �.35

(95% CIs, �.15 to �.55). That is, for every 1 point

increase in BMI, the HHS dropped on average by

0.25 or 0.35. The other predictors with a significant

individual influence on the postoperative HHS were

length of stay, previous thromboembolism or coro-

nary heart disease, drop in hemoglobin at 24 hours,

and preoperative HHS. By far, the most significant

of these is the preoperative HHS (see Fig. 2).

Body mass index was not a significant predictor

for any of the SF-36 component scores.

Discussion

The HHS improved considerably after surgery.

The hip score is weighted toward the patient’s

assessment of pain, function, and activity (91 of

100 points), with lesser emphasis on surgeon-

determined measures such as range of motion

and absence of deformity (9 of 100). Ultimately,

the patients’ view on the outcome of surgery is

probably the most important, and the HHS is a good

measure of patient symptoms.

Body mass index independently predicted for

a lower HHS at 6 and 18 months. However, its

individual effect, whereas significant statistically,

was small. If we take a change in BMI of 20 points

(the difference between being underweight and

morbidly obese), we estimate that it will only

produce on average a lowering in the HHS of 5.0 at

6 months and 7.0 at 18 months. These changes are

small given that the mean improvement in the HHS

at 18 months is 43. None of the 9 component scores

of the SF-36 were predicted by BMI.

We saw no relationship between BMI and early

failure of THA. Although the obese may put

increased loads through their joint arthroplasty,

there is evidence that the more obese a patient, the

less active they are. Hence, the increased weight is

balanced by decreased cycles of loading [17].

In a paper comparing 41 obese and 125 nonobese

patients, Soballe et al [9] noted increased blood loss

in the obese group. In a series of 80 patients,

Bowditch and Vilar [8] also noted increased blood

loss in obese patients when compared to those

of ideal weight. Multiple regression analysis is a

sophisticated tool that allows for the correction of

other variables in assessing the individual influence

of BMI. Even with the large numbers in our study,

once other factors such as comorbidity are taken into

account, we did not find that BMI per se increased

measured blood loss or transfusion requirement.

Does Body Mass Index Affect The Early Outcome of Primary THA? ! Moran et al 869

It is likely that the comparison of obese and non-

obese without factoring in confounding data over-

simplifies the true state of affairs.

We did not divide the patients into groups

based on BMI, as these divisions of a continuous

variable would be arbitrary. One of the strengths

of this paper is the use of regression analysis to

identify independently significant predictors. For

example, in our initial univariate analysis, an

association between BMI and infection was sus-

pected. However, an increased BMI is associated

with an altered incidence of other conditions,

such as diabetes mellitus. It would have been erro-

neous to draw the conclusion that obesity is

responsible for increased rates of infection, with-

out allowing for the fact that BMI is also

associated with the incidence of diabetes. Regres-

sion analysis allows us to separate out diabetes

and BMI and test the effect of each on the

incidence of infection. There were only 9 patients

with a BMI more than 40 kg/m2. This is in

keeping with other studies that have investigated

the effect of BMI on the outcome of lower limb

arthroplasty [18]. We did not study enough

patients with a BMI more than 40 kg/m2 to make

any definite conclusions about this group.

Although BMI does predict for a slightly lower

HHS and SF-36, there is no association with other

outcome measures used. Even with a large cohort,

complications after primary THA are rare. For the

commonest complication, superficial wound infec-

tion, there was a 3-fold range for the CI for the odds

ratio for obese (BMI more than 30 kg/m2) relative

to nonobese patients. This may indicate that the

power of the study is too low to give a precise

estimate of the effect of BMI on the risk of rare

complications. The effect on the HHS is small and

we do not feel it is large enough to warrant

withholding THA from patients solely on the

grounds of body mass.

Our results reflect postoperative complications

and early outcome of THA. We continue to follow

these patients to ascertain the medium and long-

term effects of body mass.

Acknowledgments

We thank Dr Rob Elton for statistical support.

References

1. Furnes O, Lie SA, Espehaug B, et al. Hip disease and

the prognosis of total hip replacements. J Bone Joint

Surg Br 2001;83:579.

2. Massoud SN, Hunter JB, Holdsworth BJ, et al. Early

femoral loosening in one design of cemented hip

replacement. J Bone Joint Surg Br 1997;79:603.

3. Katz JN, Losina E, Barrett J, et al. Association

between hospital and surgeon procedure volume

and outcomes of total hip replacement in the United

Stated Medicare population. J Bone Joint Surg Am

2001;83:1622.

4. Perha C, Labs K, Muschik M, et al. The influence of

obesity on perioperative morbidity and mortality in

revision total hip arthroplasty. Arch Orthop Trauma

Surg 2000;120:267.

5. Jiganti JJ, Goldstein WM, Williams CS. A comparison

of the perioperative morbidity in total joint arthro-

plasty in the obese and nonobese patient. Clin Orthop

1993;289:175.

6. Kelsey JL, Wood PL, Charnley J. Prediction of

thromboembolism following total hip replacement.

Clin Orthop 1976;114:247.

7. Smith DM, Oliver CH, Ryder CT, et al. Complications

of Austin Moore arthroplasty. Their incidence and

relationship to potential predisposing factors. J Bone

Joint Surg Am 1975;57:31.

8. Bowditch MG, Vilar RN. Do obese patients bleed

more? A prospective study of blood loss at total hip

replacement. Ann R Coll Surg Engl 1999;81:198.

9. Soballe K, Christensen F, Luxhoj T. Hip replace-

ment in obese patients. Acta Orthop Scand 1987;

58:223.

10. Stickles B, Phillips L, Brox WT, et al. Defining the

relationship between obesity and total joint arthro-

plasty. Obes Res 2001;9:219.

11. Chan CL, Villar RN. Obesity and quality of life after

primary hip arthroplasty. J Bone Joint Surg Br

1996;78:855.

12. UK National Audit Office report—NHS executive.

Hip replacements: getting it right first time (HC417

99/00). London: The Stationery Office.

13. Harris WH. Traumatic arthritis of the hip after

dislocation and acetabular fractures. Treatment by

mold arthroplasty. J Bone Joint Surg Am 1969;

51:737.

14. Ware JE, Sherbourne CD. The MOS 36-item short-

form health survey (SF-36). 1 Conceptual framework

and item selection. Med Care 1992;30:473.

15. Soderman P, Malchau H. Is the Harris Hip Score

system useful to study the outcome of total hip

replacement. Clin Orthop 2001;384:189.

16. Nilsdooter AK, Roos AM, Westerlund JP, et al.

Comparative responsiveness of measures of pain

and function after total hip replacement. Arthritis

Rheum 2001;45:258.17. McClung CD, Zahiri CA, Higa JK, et al. Relation-

ship between body mass index and activity in hip

and knee arthroplasty patients. J Orthop Res 2000;

18:35.

18. Deshmukh RG, Hayes JH, Pinder IM, et al. Does

body weight influence outcome after total knee

arthroplasty? A 1-year analysis. J Arthroplasty 2002;

17:315.