Osteoarthritis and Cartilage (2001) 9, 137–146

© 2001 OsteoArthritis Research Society International 1063–4584/010/020137+10 $35.00/0

doi:10.1053/joca.2000.0369, available online at http://www.idealibrary.com on

A comparison of outcomes in osteoarthritis patients undergoing total

hip and knee replacement surgery
C. J. M. Bachmeier*, L. M. March†, M. J. Cross*, H. M. Lapsley‡, K. L. Tribe*, B. G. Courtenay*
and P. M. Brooks§ for the Arthritis Cost and Outcome Project Group**
*Departments of Medicine and Orthopaedics, St Vincent’s Hospital, University of New South Wales, Australia
†Departments of Rheumatology and Orthopaedics, Royal North Shore Hospital, University of Sydney,
Australia
‡School of Health Services Management, University of New South Wales, Australia
§Office of the Executive Dean (Health Sciences), University of Queensland, Australia
\Australian Institute of Musculoskeletal Research, Sydney, Australia
¶North Sydney Orthopaedic & Sports Medicine Centre, Australia

Summary
Objective: The aims of this study were to assess changes in physical function and quality of life with the Western Ontario and McMaster
Universities Osteoarthritis Index (WOMAC) and the instrument of the Medical Outcomes Study SF-36 Health Survey (MOS SF-36),
respectively, in patients undergoing hip and knee joint replacement surgery and to compare the responsiveness of these two outcome
measures 1 year after surgery.
Design: One hundred and ninety-four patients with osteoarthritis (OA knee 108, OA hip 86) admitted to four hospitals in Sydney were
followed over a period of 1 year at 3 monthly intervals.
Results: WOMAC measures improved significantly after 1 year for OA hip and OA knee: there was reduction in pain of 71% and 53%,
reduction in stiffness of 55% and 43% and improvement in physical function of 68% and 43%, respectively. MOS SF-36 measures in those
having hip surgery improved significantly for pain (222%), physical function (247%), physical role functioning (402%), general health (110%),
vitality (143%), social functioning (169%) and mental health (114%). For those in the knee surgery group, significant improvement was seen
for pain (175%), physical function (197%), physical role functioning (275%), vitality (125%) and social functioning (119%). The WOMAC was
a more responsive measure than the MOS SF-36.
Conclusion: WOMAC and MOS SF-36 detect significant and clinically meaningful changes in outcome after hip and knee replacement.
WOMAC requires a smaller sample size and is more responsive in the short term. For a follow-up longer than 6 months MOS SF-36 provides
additional information. The improvement in outcomes following hip joint surgery were significantly greater than those following knee surgery.
© 2001 OsteoArthritis Research Society International
Key words: Hip/knee surgery, Osteoarthritis, WOMAC, MOS SF-36.

Introduction than 400 000 primary hip and knee arthroplasties are
performed each year and the annual costs of these proce-
Joint replacement has had a major impact on the manage- dures were estimated to exceed US$ 10 billion1. In
ment of osteoarthritis (OA). Economically it requires an Australia AUS$ 13.5 million Commonwealth benefits were
important and increasing part of the health budget in every paid in the financial year 1997/98 for 17 000 hip and knee
country where it has been investigated. In the U.S.A. more replacements (customized tables, Health Insurance
Commission), compared to AUS$ 8.5 million in 1991/92 for
Received 13 May 1999; accepted 10 June 2000. 14 000 hip and knee replacements2. Expenditures on
**Including: C. A. Cass*; M. R. J. Coolican†; M. J. Cross\¶; M. J. primary joint arthroplasty surgery will increase in the future
Neil*; L. A. Pinczewski\¶; J. S. Quain*; F. W. Robertson*; S. J. with our aging population.
Ruff†. In clinical research, outcome instruments are of major
Sources of support: Australian National Health and Medical importance. They have to be valid, reliable and responsive
Research Council, Ciba-Geigy-Jubiläumsstiftung Basel Switzer- to change. The Western Ontario and McMaster Universities
land, Karger Fonds Basel Switzerland, Schweizerische Gesells- Osteoarthritis Index (WOMAC) is a multidimensional, self-
chaft der Wissenschaften Basel Switzerland, Schweizerische administered outcome measure, which has been devel-
Akademie der Wissenschaften in Basel, Switzerland and
Schweizerische Gesellschaft für Rheumatologie in Switzerland.
oped by Bellamy3 for clinical trials in patients with hip or
Address correspondence to: Marita Cross, Level 9, Garvan knee osteoarthritis. It probes for the dimensions of pain (5
Institute of Medical Research, 384 Victoria Street, Darlinghurst items), stiffness (2 items) and physical function (17 items).
NSW 2010, Australia. Tel: 0061 2 9295 8419; Fax: 0061 2 9295 It has been intensively validated4 and shown to be a valid,
8401. reliable and responsive instrument. The Medical Outcomes

137
138 C. J. M. Bachmeier et al.: Outcome comparison after joint replacement surgery

Study SF-36 Health Survey (MOS SF-36) is a self- ANALYSIS

administered, generic health-related quality of life instru-
ment. It was developed in the Rand Health Insurance Data were entered using Paradox software and then
experiment from the Rand Health Insurance long form. analysed using Statistical Package for Social Sciences
Quality of life is assessed by 36 items across eight dimen- (SPSS) for Windows version 6.0 program9. Both question-
sions (physical functioning, role/physical functioning, bodily naires were scored according to guidelines of their
pain, general health, vitality, social functioning, role/ authors10,11. P-values ≤0.05 were considered significant.
emotional functioning, mental health). It has been tested for Parametric Students t-tests (independent and paired tests)
its psychometric properties5,6. It has not been specially were used to assess the significance of the mean differ-
designed for patients with OA but seems to be applicable to ences before and after the operation, when scores were
a broad spectrum of diseases. Compared to the WOMAC it approximately normally distributed. The non-parametric
is a more general instrument with the advantage of the test (Mann–Whitney U-Wilcoxon sum W test) was used for
ability to assess pain, physical function and quality of life, the samples that were not normally distributed. Normality of
which are the most important outcomes in clinical trials with distribution was examined by standard diagnostic tests
osteoarthritic patients. The WOMAC and the MOS SF-36 (Lillefors test, box plot, normal probability plot and histo-
have been recommended as valid outcome measures in gram). The distribution of age, gender and duration of
OA research at the OMERACT III conference7 and in the disease in the groups of hip and knee joint replacement
guidelines of the Osteoarthritis Research Society8 to allow were assessed by chi-squared tests (for the dichotomous
standardization and comparability between studies. In this variable, gender) and independent t-tests (for continuous
study we wanted to quantify outcome after hip and knee variables). The change in the different domains was calcu-
replacement using these two recommended measures, the lated as the difference of the pre-operative score to the
WOMAC and the MOS SF-36, and compare them prospec- follow-up score. The improvement in the hip replacement
tively. We hypothesized that the WOMAC would be the group compared to the knee replacement group was the
more efficient instrument, being disease-specific. comparison of the mean score difference of the pre-
operative score and the follow-up score between the hip
and the knee replacement groups. These two comparative
Method analyses used paired tests limiting the calculations to those
people who completed both questionnaires at each time
STUDY POPULATION point. These calculations required multiple comparisons.
Adaptation of the significance level was not performed. The
This study is part of an ongoing large prospective trial comparison of the improvement in the hip and knee
assessing costs of arthritis. Nine orthopedic surgeons in replacement groups was further assessed using multivari-
four hospitals in Sydney, Australia (St Vincent’s Public and ate analysis to allow for age, gender and OA duration
Private Hospitals, the Centre for Bone and Joint Disease as potential confounders. The models were created by
and Mater Misericordiae) provided operation waiting lists backward stepwise selection.
for total primary hip or knee joint replacement. Both pri- The estimates of the comparative responsiveness of the
vately insured and public (non-insured) patients were two instruments were calculated with the relative efficiency
included. All patients with a diagnosis of OA and rheuma- (RE) using the following formula)12:
toid arthritis were eligible to enter the study. Waiting lists
were checked fortnightly to monitor the continuing recruit-
ment. Patients were asked by telephone to take part in the
trial. In this paper only patients with OA were analysed.
Most patients were visited at home by the research assist-
ant for an interview after having signed a consent form, and This comparison is related to pain and physical function,
were also seen on admission for surgery. Follow-up was by the only two dimensions measured in both questionnaires.
postal questionnaires or home visit if necessary. Follow-up The MOS SF-36 scores on a 0–100 worst to best scale.
home visits were made to four hip replacement and eight The WOMAC scores on a 1–5 best to worst scale, which
knee replacement patients. For two of the hip patients and was transformed accordingly to the MOS SF-36 0–100
two of the knee patients reading English was a problem, worst to best scale.
so the questions were read out to these people by the
interviewer. For the remainder, the questionnaires were
self-administered and the research assistant provided
assistance where necessary. Results
From 24 February 1994 to 1 July 1996, 376 patients with
ASSESSMENTS OA on the waiting lists for hip or knee joint replacement
were eligible to take part in the study. Of these, 124 patients
Over a 12-month period the patients were mailed did not take part in the study for the following reasons: 49
WOMAC and MOS SF-36 questionnaires at 3-monthly patients could not be contacted in time, 27 were not
intervals. This analysis includes the first 12-month period of interested, 18 felt too old or unwell, the English of 12
follow-up. Shortly before the questionnaires were due back patients was too poor, seven patients could not be con-
patients were reminded by telephone calls. In case of tacted at the address provided by the surgeon, five patients
failure to respond the patients were again reminded by did not have a telephone to be contacted by, four patients
telephone. Returned questionnaires were checked for found the study too confusing and the operation was
missing answers and clarified by telephone if necessary. To cancelled in two patients. There was no difference in age,
minimize the drop-out rate patients were informed about sex or type of operation between those who provided pre-
the study progress with quarterly newsletters and personal and post-operative questionnaires and those who did not
contacts such as birthday cards. take part in the study.
Osteoarthritis and Cartilage Vol. 9, No. 2 139

Table I
Baseline characteristics of patients*
Characteristic Hip replacement Knee replacement Entire surgical
group (N=86) group (N=108) group (N=194)

Age (years)† 65 (11.5) 72 (7.0) 69 (9.9)

Duration of disease (years)‡ 8.5 (9.4) 12.0 (11.9) 10 (11.1)
Female proportion (%) 39 61 52

*Providing at least one pre-operative and one post-operative WOMAC questionnaire.

†Mean (standard deviation).
‡Median (standard deviation).

Two hundred and fifty-two patients started the study, vitality P=0.54, social function z= −1.86, role emotional
corresponding to a response rate of 67.3%. Thirty-six function P=0.68, mental health P=0.98).
patients provided pre-operative questionnaires and with- Four in the hip group had their other hip replaced and
drew before providing post-operative questionnaires for the three in the knee group had their other knee replaced in the
following reasons: two patients died, in seven patients the 12-month follow-up period. Those who had another joint
operation was cancelled, two patients moved from their replaced did not differ significantly when WOMAC and
address and left no forwarding details and 25 patients SF-36 scores were compared with those who had no
denied further participation in the study. These 36 patients further joint replacement at the 12-month follow-up. There
(14.3%) did not differ regarding sex and type of joint was a low number of hospitalizations for adverse events
replacement operation but were older compared to the 194 most probably related to the primary arthroplasty (3% hip
patients in the analysis (t= −2.506, P=0.013, mean age of and 13% knee). These included 10 knee patients requiring
patients lost to follow-up and participants: 70±9.9 years manipulation under anesthesia, three with an infection in
and 74±10.4 years, respectively). Twenty-two patients the operated joint and one knee patient and three hip
started to provide pre-operative data but their first post- patients with thromboembolic complications. The 12-month
operative questionnaire was answered after 1 July 1996. WOMAC and SF-36 measures showed a significant differ-
For 194 patients (86 with OA of the hip and 108 with OA ence in WOMAC physical function only with those requiring
of the knee) at least one pre-operative and one post- readmission to hospital within 12 months of knee arthro-
operative WOMAC questionnaire was available. One hun- plasty reporting worse function at 12-month follow-up. No
dred and seventy-three patients (73 with OA of the hip and significant differences were seen in the hip replacement
100 with OA of the knee) provided at least one pre- group in WOMAC or SF-36 scores at 12 months between
operative and one post-operative MOS SF-36 question- those requiring readmission and those who did not.
naire in addition to the WOMAC questionnaires. Ninety-one In addition, four knee replacement patients reported they
per cent of hip questionnaires were completed within 4 required hospitalization for arthritis-related conditions not
weeks prior to surgery, another 1.2% up to 5 weeks and related to the primary replacement within the 12-month
7.8% were longer than 5 weeks. For the knee patients, 88% follow-up period (such as neck pain due to arthritis, removal
were completed within 4 weeks pre-surgery, another 4.5% of cartilage from other knee). These patients reported
within 5 weeks and 7.5% in excess of 5 weeks. Where significantly worse WOMAC pain, stiffness and function
questionnaires were completed in excess of 5 weeks prior at 12-month follow-up than all other knee replacement
to surgery, these people were sent questionnaires, but did patients (which includes those who were admitted for
not complete them immediately pre-operatively. The mean another arthroplasty or for conditions likely to be related to
time to surgery from pre-operative evaluation was 14.9 their primary arthroplasty). No difference was seen in any
days for knee replacement patients and 14.2 days for hip scale of the SF-36.
replacement patients.
The overall cohort was followed for a total of 2031 IMPROVEMENT OVER TIME IN THE HIP REPLACEMENT GROUP
person-months, with an average of 11 months. For the hip
ACCORDING TO WOMAC
group, the total follow-up was 891 person-months (average
11 months) and for the knee group 1140 person months in The improvement over time in the patient group under-
total (average 10 months). going hip replacement surgery is shown in Fig. 1. In the hip
group there was a reduction in the mean scores of pain,
stiffness and physical function as assessed by the WOMAC
STUDY POPULATION CHARACTERISTICS
of about 50% in the first 3 follow-up months after the
The study population characteristics of age, duration of operation. This improvement continued up to 1 year, when
disease and gender distribution are listed in Table I. In the improvement in physical function and pain from base-
general, the patients in the hip replacement group were line was more than 60%. The changes were statistically
younger (t-value= −5.27, 130 df, P<0.001) and had a significant at 3, 6, 9 and 12 months for each of the WOMAC
shorter duration of their disease (z-value= −2.68, P=0.007) dimensions.
than those undergoing knee replacement surgery. Pre-
operatively all the domains in the WOMAC and the MOS IMPROVEMENT OVER TIME IN THE KNEE REPLACEMENT GROUP
SF-36 were comparable between the two surgical cohorts ACCORDING TO WOMAC
(WOMAC: pain P=0.19, stiffness P=0.99, physical function
P=0.52; SF-36: physical function P=0.78, role physical The patients undergoing knee replacement surgery
function z= −1.69, pain P=0.46, general health P=0.16, showed similar results (Fig. 2). The pre-operative scores
140 C. J. M. Bachmeier et al.: Outcome comparison after joint replacement surgery

Fig. 1. WOMAC scores (means and standard deviations) in the patient group undergoing hip joint replacement surgery. *P<0.05. Lower
scores indicate a better health state. Baseline N=86; 3 months N=82; 6 months N=64; 9 months N=61; 12 months N=43.

Fig. 2. WOMAC scores (means and standard deviations) in the patient group undergoing knee joint replacement surgery. *P<0.05. Lower
scores indicate a better health state. Baseline N=108; 3 months N=101; 6 months N=84; 9 months N=67; 12 months N=48.
Osteoarthritis and Cartilage Vol. 9, No. 2 141

Table II
Improvement in the hip replacement group compared to the knee replacement group according to WOMAC
WOMAC domain Mean differences (95% CI) between hip and knee change from baseline scores to post-operative follow-up
3 months 6 months 9 months 12 months
N Hips: 79 Hips: 62 Hips: 67 Hips: 42
Knees: 101 Knees: 84 Knees: 59 Knees: 48
Pain −1.3 (−2.7,0.04) −0.9 (−2.4,0.6) −2.0 (−3.6,−0.5)* −2.6 (−4.6,−0.6)*
Stiffness −0.9 (−1.5,−0.2)* −0.8 (−1.4,−0.2)* −1.1 (−1.7,−0.5)* −0.8 (−1.6,0.1)
Physical function −4.3 (−8.7,0.0)* −5.2 (−9.4,−0.9)* −7.9 (−12.7,−3.0)* −10.8 (−17.1,−4.5)*

*P<0.05.
A negative score means a greater improvement in the hip replacement group.
CI=Confidence intervals.

for pain, stiffness and physical function as assessed by the for pain, vitality, social functioning and mental health
WOMAC improved after operation. The improvement com- was statistically significant only for the first quarter
pared with baseline was statistically significant at 3, 6, 9 postoperatively.
and 12 months. After 1 year the reduction in pain and
stiffness and the improvement in physical function was
each about 50%, respectively. IMPROVEMENT OVER TIME IN THE KNEE REPLACEMENT GROUP
ACCORDING TO MOS SF-36

IMPROVEMENT IN THE HIP REPLACEMENT GROUP COMPARED TO Figure 4(a) (physical function, physical role function,
THE KNEE REPLACEMENT GROUP ACCORDING TO WOMAC bodily pain and general health) and (b) (vitality, social
function, emotional role function and mental health) present
The improvement in the hip joint replacement group the improvement over time in those undergoing knee joint
compared to the knee joint replacement group is shown in replacement surgery. In the knee replacement group all
Table II. The mean improvement in pain, stiffness and domains of the questionnaire improved post-operatively
physical function between baseline and the follow-ups were except general health, which was not influenced by the
more pronounced in patients who had undergone a hip surgical intervention. The improvement in physical function-
replacement compared with patients who had knee joint ing and pain was statistically significant at every follow-up,
replacement. Using univariate analysis the improvement whereas the improvement in physical role functioning,
was statistically significantly greater at all follow-ups vitality and social functioning was significant from 6 months
regarding physical function, at follow-ups 9 and 12 months on. As in the hip group the relative improvement at one year
regarding pain and at follow-ups 3 and 6 months regarding was most pronounced for the physical functioning (197%)
stiffness. Multivariate analysis adjusting for age, sex and and the physical role functioning (275%).
duration of disease confirmed a greater improvement in
outcome for hip surgery patients at 9 months follow-up
(pain P=0.008, stiffness P=0.0001 and physical function IMPROVEMENT IN THE HIP REPLACEMENT GROUP COMPARED TO
P=0.001, respectively) and at 12 months follow-up (pain THE KNEE REPLACEMENT GROUP ACCORDING TO MOS SF-36
P=0.02, stiffness P=0.02 and physical function P=0.001,
respectively). Table III shows the change in the outcome of patients
undergoing hip joint replacement surgery compared to
those undergoing knee joint replacement surgery. Com-
IMPROVEMENT OVER TIME IN THE HIP REPLACEMENT GROUP pared with patients after knee joint replacement, patients in
ACCORDING TO MOS SF-36 the hip replacement group had a greater improvement for
all domains at all follow-ups except regarding emotional
The improvement over time in patients undergoing hip role functioning at 12 months as assessed by the MOS
joint replacement is presented in Fig. 3(a) and (b) (the SF-36. In univariate and multivariate analysis the greater
domains of physical function, physical role function, bodily improvement was significant at most of the quarterly follow-
pain and general health and the domains of vitality, social ups regarding physical function, pain and vitality. Signifi-
function, emotional role function and mental health, cantly greater improvement in the hip group regarding
respectively). In the hip replacement group all eight physical role functioning using univariate analysis was
domains of the MOS SF-36 improved after the joint opera- confirmed by multivariate analysis only at three months
tion. The improvement in physical functioning, physical role post-operatively, whereas the greater improvement in men-
functioning, pain, vitality, social functioning and mental tal health in the hip group was significant at 9 and 12
health compared to baseline was statistically significant at months after adjusting for age, sex and OA duration.
3, 6, 9 and 12 months. The improvement in general health
was statistically significant at 9 and 12 months, whereas
the improvement in emotional role functioning was statisti- RELATIVE EFFICIENCY OF THE WOMAC COMPARED TO MOS SF-36
cally significant only at 9 months. The relative improvement
after 1 year was most pronounced for physical functioning The WOMAC had a greater relative efficiency for pain
(247%) and physical role functioning (402%). These two and physical function at all follow-ups except for physical
latter dimensions showed statistically significant improve- function in the knee replacement group at 12 months,
ment also from 3–6 months. The quarterly improvement where both questionnaires are similarly efficient (Table IV).
142 C. J. M. Bachmeier et al.: Outcome comparison after joint replacement surgery

Fig. 3. (a) The MOS SF-36 scores of the domains physical function, physical role function, bodily pain and general health (means and
standard deviations) in the patient group undergoing hip joint replacement surgery. *P<0.05. Higher scores indicate a better health state.
Baseline N=73; 3 months N=71; 6 months N=57; 9 months N=54; 12 months N=53. (b) The MOS SF-36 scores of the domains vitality,
social function, emotional role function and mental health (means and standard deviations) in the patient group undergoing hip joint
replacement surgery. *P<0.05. Higher scores indicate a better health state. Baseline N=73; 3 months N=71; 6 months N=57; 9 months
N=54; 12 months N=53.
Osteoarthritis and Cartilage Vol. 9, No. 2 143

Fig. 4. (a) The MOS SF-36 scores of the domains physical function, physical role function, bodily pain and general health (means and
standard deviations) in the patient group undergoing knee joint replacement surgery. *P<0.05. Higher scores indicate a better health state.
Baseline N=100; 3 months N=89; 6 months N=77; 9 months N=66; 12 months N=51. (b) The MOS SF-36 scores of the domains vitality,
social function, emotional role function and mental health (means and standard deviations) in the patient group undergoing knee joint
replacement surgery. *P<0.05. Higher scores indicate a better health state. Baseline N=100; 3 months N=89; 6 months N=77; 9 months
N=66; 12 months N=51.
144 C. J. M. Bachmeier et al.: Outcome comparison after joint replacement surgery

Table III
Improvement in the hip replacement group compared to the knee replacement group according to SF-36
SF-36 domain Mean difference (95% CI) between hip and knee change from baseline score to post-operative follow up
3 months 6 months 9 months 12 months
N Hips: 66 Hips: 52 Hips: 49 Hips: 42
Knees: 88 Knees: 76 Knees: 63 Knees: 45
Physical function 10.7 (3.6,17.8)* 4.3 (−3.6,12.2) 8.4 (−0.2,17.0) 11.9 (1.6,22.3)*
Physical role function 17.3 (3.7,30.8)* 16.1 (0.5,31.7)* 17.0 (0.8,33.3)* 9.2 (−10.2,28.7)
Bodily pain 12.3 (4.5,20.2)* 8.2 (−1.2,17.6) 18.4 (8.4,28.4)* 18.5 (7.1,30.0)*
General health 6.2 (−0.5,12.9) 5.5 (−1.2,12.2) 10.9 (3.1,18.7)* 5.3 (−2.8,13.4)
Vitality 7.8 (−0.5,16.1) 8.0 (0.2,15.9)* 16.6 (8.3,24.9)* 11.2 (2.2,20.2)*
Social function 3.7 (−10.3,17.8) 3.9 (−8.5,16.3) 7.7 (−6.4,22.0) 9.5 (−5.2,24.3)
Emotional role function 6.2 (−12.2,24.6) 0.5 (−18.5,19.6) 5.2 (−16.2,26.5) −1.2 (−22.7,20.3)
Mental health 3.8 (−2.3,9.9) 1.7 (−4.5,7.8) 8.2 (0.7,15.5)* 7.4 (−0.2,15.0)

*P<0.05.
A positive score means a greater improvement in the hip replacement group.
CI=Confidence intervals.

The greater relative efficiency of the WOMAC decreased replacement although significant changes in health status
over time for both surgical groups regarding pain and could be demonstrated. An unchanged general health
physical function. The relative efficiency of the WOMAC status was seen previously in other 2-year follow-up
was most pronounced for pain assessment in the knee joint studies looking at total hip and knee arthroplasties
replacement group. assessed by the MOS SF-3614,15. This may be explained
by the fact that even after a very successful operation pain
and restrictions in daily life remain post-operatively, which
Discussion are perceived by patients strongly enough to rate their
Eighty-six patients undergoing hip joint replacement and general health as insufficient. Perhaps this paradox reflects
108 patients undergoing knee joint replacement were some limitations of a generic health questionnaire which
assessed pre-operatively and at 3-monthly follow-up does not include other important aspects of quality of life,
intervals post-operatively with WOMAC and MOS SF-36 such as sleep and sexual function. Correlations between
questionnaires. pain reduction and functional independence and psycho-
As ascertained by the WOMAC both patient groups logical well-being have been shown by others to be
experienced significant improvement in pain, stiffness and inconsistent16.
physical function. After 1 year, pain was reduced to 29% Although patients in the hip and the knee replacement
and 47% of the pre-operative measures in the hip and knee group experienced improvement in virtually all domains
group, respectively, stiffness was reduced to 45% and 57%, after their operation, the improvement was earlier and more
respectively and physical function increased by 68% and pronounced in patients with hip replacement for all domains
43%, respectively. The greatest improvement was seen in of the WOMAC and for physical function, pain, general
both groups within the first 3 months post-operatively. After health, vitality and mental health assessed by the MOS
3 months all measures continued to improve with the SF-36. The reason for these differences is not clear, as
absolute values levelling off. A dramatic improvement by there were no statistically significant differences in the
three months and smaller changes thereafter in these outcome measures at baseline and using multivariate
patients has been described previously13. analysis potential confounders such as age, sex and dura-
As assessed by the MOS SF-36 both patient groups tion of disease were adjusted for. Within this 1 year of
improved at all domains except that the general health of follow-up no patient had to undergo a revision of the study
patients after knee replacement remained unaltered. Again, knee or hip arthroplasty or a previously implanted hip or
the greatest improvement in patients undergoing hip or knee replacement. It is possible that these differences
knee replacement was seen within the first 3 months between hip and knee replacement groups could be
post-operatively. It seems a paradox that patients’ percep- explained by a different rate of post-operative complica-
tion of their general health was not influenced after knee tions, which were slightly higher in the knee group. Comor-
bidities may also contribute; however, these were not
assessed. A greater relative improvement in osteoarthritic
Table IV hip vs knee replacement for pain and disability has been
The relative efficiency for WOMAC vs MOS SF-36 described elsewhere17,18, including a 10-fold difference in
Follow-up Pain Physical function Quality Adjusted Life Years (QALYs) gained19 when hip
(months) replacement (4 QALYs gained) were compared to knee
Hip Knee Hip Knee replacement (0.42 QALYs gained)20.
0–3 1.43 1.94 1.73 1.73 As a generic health status measure, the SF-36 has been
0–6 1.54 1.92 1.40 1.17 found to be more relevant and more responsive than the
0–9 1.42 1.84 1.48 1.24 Sickness Impact Profile and is the preferable generic health
0–12 1.13 1.74 0.99 1.30 status measure in patients undergoing hip replacement21.
In this study the disease-specific WOMAC was a more
Relative efficiency >1 if WOMAC is the more responsive responsive measure than the MOS SF-36. This was most
measure. pronounced in the assessment of pain in patients with knee
Osteoarthritis and Cartilage Vol. 9, No. 2 145

replacement. After 1 year the greater relative efficiency of supported by the Ciba-Geigy-Jubiläumsstiftung, Karger
the WOMAC regarding ascertainment of pain and physical Fonds, Schweizerische Gesellschaft für Rheumatologie
function in both patient groups levelled somewhat com- and Schweizerische Akademie der Wissenschaften in
pared to the first months of observations. This may be Basel, Switzerland.
important for short-term studies with a duration of less than
1 year, where using the WOMAC questionnaire requires a
smaller sample size. In agreement with the results of this
study the WOMAC has been found to be the more sensitive References
measure in detecting knee disabilities compared to the
MOS SF-3622. This is not surprising, as the WOMAC has 1. Lavernia CJ, Drakeford MK, Tsao AK, Gittelsohn A,
been developed as a specific outcome measure for Krackow KA, Hungerford DS. Revision and primary
patients with OA of the lower limb. hip and knee arthroplasty. A cost analysis. Clin
Applying the MOS SF-36 questionnaire provided Orthop 1995;311:136–41.
additional information on the patients’ quality of life as 2. Brooks P, Lapsley H, Oh ES, Cross M, Courtenay B,
expressed by physical role functioning, general health, Schwarz J. et al. The economic impact of joint
vitality, social functioning, emotional role functioning and replacement surgery: a preliminary analysis.
mental health. This information is clinically important and (Abstracts) Proceedings of APLAR Congress, April
patient-relevant, as well as the more traditional outcomes 21–26 1996. Melbourne: Australia.
of pain, stiffness and physical function. Quality of life is a 3. Bellamy N. Osteoarthritis—an evaluative index for
major determinant of the patients’ global assessment of clinical trials. M.Sc. Thesis. 1982. McMaster Univer-
satisfaction with their health status. Interestingly, quality of sity, Hamilton, Ontario, Canada.
life was slower to improve after the operation and the 4. Bellamy N, Buchanan WW, Goldsmith CH, Campbell J,
relative changes were more subtle compared to the relative Stitt LW. Validation study of WOMAC: a health status
improvement in physical outcome measures. Similarly, a instrument for measuring clinically important patient
less dramatic improvement of psychological function com- relevant outcomes following total hip and knee
pared to pain was noted in a cohort study of patients arthroplasty in osteoarthritis. J Orthop Rheumatol
undergoing hip replacement surgery with a 6–8 week 1988;1:95–108.
follow-up23 which may reflect a slower adaptation of psy- 5. McHorney CA, Ware JE Jr, Rogers W, Raczec AE, Lu
chological parameters compared to physical issues in a JFR. The validity and relative precision of MOS short-
new life situation. and long-form health status scales and Dartmouth
The strengths of the study are its prospective design, COOP charts: results from the medical outcome
high participation and follow-up rate reducing well known study. Med Care 1992;30(Suppl. 5):MS253–65.
sources of biases of retrospective and cross-sectional 6. McHorney CA, Ware JE Jr, Raczec AE. The MOS
studies. The characteristics of the study population from 36-item Short-Form Survey (SF-36). II. Psychometric
multiple orthopedic surgeons and its source both in private and clinical tests of validity in measuring physical
and public hospitals should allow the generalizability of and mental health constructs. Med Care 1993;31:
the results. Comparable results have been reported in 247–63.
American and British populations, but this is the first 7. Bellamy N, Kirwan J, Boers M, Brooks P, Strand V,
Australian study published17,18,22. A further strength of this Tugwell P, et al. Recommendations for a core set of
study is the high power of the analyses: the comparison of outcome measures for future phase III clinical trials
the WOMAC domains in both surgical groups had a power in knee, hip, and hand osteoarthritis. Consensus
well over 80%. A similar power was calculated for physical development at OMERACT III. J Rheumatol 1997;24:
function and bodily pain in the SF-36 questionnaire. Adjust- 799–802.
ment for age, sex and disease duration was performed in 8. Hochberg MC, Altman RD, Brandt KD, Moskowitz RW.
multiple regression analysis. Design and conduct of clinical trials in osteoarthritis:
In summary, WOMAC and MOS SF-36 detect significant preliminary recommendations from a task force of
and clinically meaningful changes in outcome after hip and the Osteoarthritis Research Society. J Rheumatol
knee replacement. We conclude that the WOMAC is a 1997;24:792–4.
more responsive measure than the MOS SF-36, thus 9. SPSS. Release 6.0 (computer program). Chicago:
requiring a smaller sample size. This is important when SPSS Inc. 1993.
designing short-term studies saving resources in time and 10. Bellamy N. WOMAC Osteoarthritis Index. A User’s
money. In studies with a follow-up longer than 6 months Guide. London, Ontario: N. Bellamy, 1995.
MOS SF-36 provides useful additional information com- 11. Ware JE, Snow KK, Kosinski M, Sandek B. SF-36
pared with using the WOMAC alone. For long-term studies Health Survey Manual and Interpretation Guide,
it is useful to apply both the WOMAC and the MOS SF-36 1993. Boston, MA: The Health Institute, New
questionnaires. After adjusting for age, gender and duration England Medical Centre.
of OA the WOMAC outcomes of pain, stiffness and function 12. Bellamy N. Musculoskeletal clinical metrology. 1st ed.
and the SF-36 outcomes of pain, physical function, vitality Dordrecht, Boston, London: Kluwer Academic
and mental health following hip joint replacement surgery Publishers, 1993.
were greater than the improvement following knee joint 13. Liang MH, Fossel AH, Larson MG. Comparison of five
replacement. health status instruments for orthopedic evaluation.
Med Care 1990;28:632–42.
14. Ritter MA, Albohm MJ, Keating EM, Farris PM, Meding
Acknowledgments JB. Comparative outcomes of total joint arthroplasty.
J Arthroplasty 1995;10:737–41.
The study was funded by the Australian National Health 15. McGuignan FX, Hozack WJ, Moriarty L, Eng K,
and Medical Research Council. Dr C. Bachmeier was Rothman RH. Predicting quality of life outcomes
146 C. J. M. Bachmeier et al.: Outcome comparison after joint replacement surgery

following total joint arthroplasty. J Arthoplasty Ann Royal College of Surgeons England,
1995;10:742–7. 1992;74:286–90.
16. Wilcock GK. Benefits of total hip replacement to older 20. Williams A. Economics of coronary artery bypassing.
patients and the community. BMJ 1978;2:37–9. BMJ 1985;291:326–9.
17. Kirwan JR, Currey HLF, Freeman MAR, Snow S, 21. Stucki G, Liang MH, Phillips C, Katz JN. The Short
Form 36 is preferable to the SIP as a generic health
Young PJ. Overall long-term impact of total hip and
status measure in patients undergoing elective total
knee joint replacement surgery on patients with oste- hip arthroplasty. Arthritis Care and Research
oarthritis and rheumatoid arthritis. Br J Rheumatol 1995;8:174–81.
1994;33:357–60. 22. Hawker G, Melfi C, Paul J, Green R, Bombardier C.
18. Kiebzak GM, Vain PA, Gregory AM, Mokris JG, Comparison of a generic (SF-36) and a disease
Mauerhan DR. SF-36 general health status survey to specific (WOMAC) instrument in the measurement of
determine patient satisfaction at short-term follow-up outcomes after knee replacement surgery. J Rheu-
after total hip and knee arthroplasty. J South Orthop matol 1995;22:1193–6.
Assoc 1997;6:169–72. 23. Petrie K, Chamberlain K, Azariah R. The psychological
19. Drewett RF, Minns RJ, Sibly TF. Measuring outcome of impact of hip arthroplasty. Aust NZ J Surg
total knee replacement using quality of life indices. 1994;64:115–17.