OsteoArthritis and Cartilage (2006) 14, 777e783

ª 2006 OsteoArthritis Research Society International. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.joca.2006.02.004

International
Cartilage
Repair
Society

Brace treatment for osteoarthritis of the knee: a prospective

randomized multi-centre trial
R. W. Brouwer M.D.y*, T. M. van Raaij M.D.y, J. A. N. Verhaar Ph.D.y,
L. N. J. E. M. Coene Ph.D.z and S. M. A. Bierma-Zeinstra Ph.D.x
y Department of Orthopedics, Erasmus Medical Centre, Rotterdam, The Netherlands
z Department of Orthopedics, Leyenburg Hospital, The Hague, The Netherlands
x Department of General Practice, Erasmus Medical Centre, Rotterdam, The Netherlands

Summary
Objective: To evaluate the effect of a brace intended to reduce load in patients with medial or lateral compartmental osteoarthritis (OA) and
concurrent varus or valgus alignment, respectively.
Design: This multi-centre randomized controlled trial (performed 2001e2003) studies the additive effect of a brace intended to reduce load in
conservative treatment of unicompartmental OA of the knee. Setting: Orthopedic department of a university medical centre and of one general
hospital. The follow-up was 12 months. Patients: 117 patients with unicompartmental OA of the knee. Intervention group (n ¼ 60) comprising
conservative treatment with additional brace treatment and a control group (n ¼ 57) comprising conservative treatment alone. Primary outcome
measures: Pain severity and knee function score. Secondary outcome measures: Walking distance and quality of life. Analysis: Multiple linear
regression models according to the intention-to-treat-principle were used to assess outcome differences for the entire group of patients. In ad-
dition, we performed explorative subgroup analyses on primary overall outcomes stratified for alignment, degree of OA, origin of OA, and age.
Results: Although the primary outcome measures were improved in the intervention group in comparison with the controls at each assessment
point, the differences reached only borderline significance. The reported walking distances at 3 months, 12 months and overall were signif-
icantly longer in the brace group (P ¼ 0.03, P ¼ 0.04 and P ¼ 0.02, respectively). Subgroup analysis showed a better effect in the varus group,
in patients with severe OA, in patients with secondary OA and in patients younger then 60 years. In total 25 patients in the brace group and 14
in the control group changed their initial treatment, mostly (74%) because of a lack of beneficial effect.
Conclusions: The results indicate that a brace intended to reduce load shows small effects in patients with unicompartmental OA. However,
many patients do not adhere in the long run to this kind of conservative treatment.
ª 2006 OsteoArthritis Research Society International. Published by Elsevier Ltd. All rights reserved.
Key words: Randomized controlled trial, Knee, Osteoarthritis, Brace, Malalignment.

Introduction Malalignment increases the risk for progression of knee

OA and predicts decline in physical function5. Overall,
Osteoarthritis (OA) of the knee is a common medical condi- more patients with OA have varus alignment (76e93%)
tion that is often seen in general practice and causes con- than valgus alignment6.
siderable pain and immobility. In the United States The initial treatment for OA of the knee is conservative,
approximately 6% of the population aged 30 years and older consisting of patient education (adaptation of activities
and 12% of the population aged 65 years and older suffer and/or weight loss), and if needed physical therapy and
from knee OA1. In addition to the consequences for the pa- medication7e13. The general purpose of a brace is to de-
tient, OA forms a considerable burden for society because crease pain and improve function; valgisation and varisation
of its chronic course and the high costs of interventions2. braces are available for unloading the medial and lateral
OA of the entire knee is distinguished from that of one compartment, respectively14e18.
compartment, which is generally caused by a mechanical A recently published Cochrane review concluded that
problem3,4. Patients with OA of the medial compartment of- there is very limited evidence for the effectiveness of brace
ten have a varus alignment and the mechanical axis and treatment for knee OA, mainly because of lack of studies on
load bearing pass through the medial compartment. this issue19. Therefore, the present study investigated the
Patients with OA of the lateral compartment generally additive effect of a brace intended to reduce load in conser-
have a valgus alignment and the mechanical axis and vative treatment of unicompartmental OA with varus align-
load bearing pass through the lateral compartment. ment or valgus alignment.

*Address correspondence and reprint requests to: R. W. Material and methods

Brouwer, M.D., Department of Orthopedics, Martini Hospital, PO
Box 30033, 9700 RM Groningen, The Netherlands. Tel: 31-50- STUDY DESIGN
5245245; Fax: 31-50-5246969; E-mail: [email protected]
Received 24 September 2005; revision accepted 4 February A multi-centre randomized controlled trial was designed
2006. to study the additive effect of a brace intended to reduce

777
778 R. W. Brouwer et al.: Brace treatment for OA of the knee

load in the conservative treatment of unicompartmental TREATMENT GROUPS

knee OA (Fig. 1).
Patients were randomly assigned to either an intervention
The study was conducted at the orthopedic outpatient de-
group comprising conservative treatment with additional
partments of a university medical centre and of a general
brace treatment, or to a control group comprising conserva-
hospital. The medico-ethical committees of both hospitals
tive treatment alone.
approved the study protocol.
The conservative treatment was identical in both groups
and consisted of standard care: i.e., patient education (ad-
INCLUSION/EXCLUSION CRITERIA aptation of activities and/or weight loss), and (if needed)
The inclusion criteria were symptomatic unicompartmen- physical therapy and analgesics.
tal knee OA and a malalignment in patients aged 18 years In the intervention group patients were fitted with a knee
and over. We diagnosed the OA as unicompartmental when brace (OAsys brace, Innovation Sports, Irvine, CA, USA);
the symptoms (pain and tenderness of the joint margins) this brace is commercially available for right/left leg in four
were located over the medial or the lateral tibiofemoral com- sizes (Fig. 2). The brace is accepted and refunded by all
partment of the knee in combination with osteoarthritic signs Dutch health insurance companies. The brace consists of
according to the Ahlbäck score (Ahlbäck > 0) in the same a thigh shell and a calf shell (both of carbon fiber) con-
medial or lateral tibiofemoral compartment of the knee as nected by titanium hinges on the medial and lateral sides.
well as in combination with varus alignment (in combination The adjustable slide bar on the medial side of the brace pro-
with medial compartment OA) or valgus alignment (in com- vides valgisation (1e12.5 degrees) with medial unloading,
bination with lateral compartment OA), respectively20. The or varisation (1e10 degrees) with lateral unloading. The de-
degree of malalignment and mechanical axis was mea- gree of varisation or valgisation depends on the degree of
sured on a whole leg radiograph in standing position and malalignment and the acceptance of the patient (extensive
determined according to one line (mechanical axis of the fe- correction will cause pressure ulcers). A specialized ortho-
mur) from the centre of the femur head using Mose circles pedic technician applied the brace and gave instructions
to the middle of the distance between the tibial spines, to the patients. During the follow-up this specialized ortho-
and a second line (mechanical axis of the tibia) from the pedic technician was present at the orthopedic outpatient
centre of the ankle to the centre of the tibial spines. department. If necessary the brace was adjusted during
Patients with concurrent symptomatic OA of medial and the follow-up visits.
lateral compartments, symptomatic patellofemoral OA
(scored on the lateral radiograph of the knee), no malalign-
ment, rheumatoid arthritis, previous high tibial osteotomy, BASELINE EVALUATION
symptomatic hip or ankle pathology, and an insufficient Age, gender, Body Mass Index (BMI), duration of com-
command of the Dutch language were excluded. plaints, severity of OA, varus alignment, pain severity, hos-
pital for special surgery score (HSS score), walking
PROCEDURES distance, quality of life, and analgesic use were scored at
baseline.
After reading the patient information form informed con-
sent was given and baseline measurements were made,
patients were randomized according to a computer-gener- OUTCOME ASSESSMENTS AT 3, 6 AND 12 MONTHS
ated procedure in blocks of 24; the allocation of treatment
was concealed until after the patient was included and Primary outcome measures were pain severity measured
baseline measurements were executed; sealed envelopes with a visual analogue scale (VAS; range 0e10), and
contained the group assignment. a knee function score using the HSS (range 0e100). The
The follow-up assessments that took place at 3, 6 and 12 HSS is divided into six categories (pain, function, range of
months included standardized questionnaires and physical motion, muscle strength, flexion deformity, and instability),
examination by one investigator. is often used in orthopedic interventions in knee OA, and

Included for the trial

N = 118

Randomization 1 denied further treatment

and cooperation
Brace group Control group
1 High Tibial Osteotomy N = 60 N = 57
1 High Tibial Osteotomy
1 Unicompartment Knee Prosthesis 3 Total Knee Prosthesis
2 Total Knee Prosthesis 2 Lost to follow-up
3 months follow-up 3 months follow up
1 High Tibial Osteotomy
2 Total Knee Prosthesis
1 Unicompartment Knee Prosthesis
2 Lost to follow-up
1 Total Knee Prosthesis
6 months follow-up 6 months follow up

3 High Tibial Osteotomy 2 High Tibial Osteotomy

1 Unicompartment Knee Prosthesis 5 Total Knee Prosthesis
12 months follow-up 12 months follow-up

Fig. 1. Flowchart showing the patients on the waiting list for surgical treatment, or who were lost to follow-up during the trial.
Osteoarthritis and Cartilage Vol. 14, No. 8 779

placed on the waiting list for surgical intervention (e.g.,

high tibial osteotomy, hemi/total knee prosthesis) during
follow-up, the last available measurement or the last preop-
erative measurement was entered.
The overall outcomes during the 12 months were
analyzed using linear regression for repeated measure-
ments with the same adjustments as above. For these anal-
yses, in case the patient was already lost to follow-up or
underwent surgical intervention before the first 3-month fol-
low-up, only baseline values were entered. Other measure-
ments were not entered.
Standardized effect sizes (adjusted mean difference in
outcome divided by the pooled SD) were calculated for all
outcomes. Effect sizes between 0.2 and 0.5 represent small
effects, between 0.5 and 0.8 moderate effects, and above
0.8 large effects24.
In addition, we performed explorative subgroup analyses
on primary overall outcomes (using the methods stated
above) stratified for alignment (varus vs valgus), degree of
OA (mild: Ahlbäck 1 vs severe: Ahlbäck 2 and 3), and origin
of OA (primary vs secondary: post-meniscectomy or cruci-
ate lesion), and age (younger than 60 years vs 60 years
and older).
The SPSS and SAS programs were used for the statisti-
cal analyses and a P-value of 0.05 was considered statisti-
cally significant.

Fig. 2. Photograph showing the fitted brace. Results

In the period January 2001 to January 2003, 118 patients
21
consists of a questionnaire and a physical examination . In were randomized. One patient withdrew immediately
the present study physical examination for the HSS knee because of dissatisfaction with the randomization outcome
function score was determined by one un-blinded assessor. (no brace) and refused any further participation; this patient
Secondary outcome measures were walking distance (in was excluded from analysis, resulting in a total sample of
kilometers) and quality of life (measured with the EuroQol- 117 patients.
5D)22. Table I shows that the mean age of the total group was
59.2 (SD 13.7) years, 50% was male and mean BMI was
SAMPLE SIZE
28.5 (SD 4.8). There were 60 patients in the intervention
group and 57 in the control group; four patients in the control
The sample size calculation was based on the study of group were lost to follow-up. In total 95 patients had varus
Magyar et al. who reported a standard deviation (SD) of 9 alignment and 22 had valgus alignment. At baseline, com-
in the HSS knee score in their study population23. For the pared with controls, scores on pain severity, HSS knee func-
present study, with a difference between two groups of tion and walking distance were worse in the brace group.
five points we would reach clinically relevant differences (ef-
fect size 0.55). To detect such a difference with two-sided
PRIMARY OUTCOME MEASURES
testing (a ¼ 0.05 and a power of 80%) we needed to include
51 patients in each study group. Over sizing by 15% al- Compared with controls, pain severity (VAS) was less in
lowed us to reach this power also in the largest subgroup the brace group at each of the three assessment points as
of patients with unicompartmental OA with varus alignment. well as overall during the 12 months follow-up; the largest
difference was at 12 months (
0.81; 95% CI:
1.76; 0.14)
(Table II). At 12 months and overall the difference in VAS
STATISTICAL ANALYSES
score was borderline significant (P < 0.1). Effect sizes at
All data were analyzed according to an intention-to-treat the three assessment points ranged from 0.3 to 0.4.
principle, implying that all patients who were randomized Knee function (HSS) in the brace group was better at
were included in the analyses, and that they were analyzed each assessment point; the largest difference was seen at
according to the group to which they were allocated. 3 months follow-up (3.5 points; 95% CI:
0.24; 7.24). Bor-
Outcome assessments at 3, 6 and 12 months (pain se- derline significance (P < 0.1) was observed at 3 months
verity, HSS knee function score, walking distance and qual- (P ¼ 0.07), 6 months (P ¼ 0.10), and overall (P ¼ 0.09).
ity of life) were analyzed using multiple linear regression The effect size at the three assessment points was 0.3.
analysis. These analyses were adjusted for the baseline
value of the outcome measure. Further, those variables SECONDARY OUTCOME MEASURES
which changed the relationship (slope) between the inde-
pendent variable (treatment group) and one of the depen- The reported walking distances at 3 months (1.21 km;
dent variables (outcomes) by more than 10% were 95% CI: 0.12; 2.28), 12 months (1.34 km; 95% CI: 0.05;
considered as confounders and were included in the 2.63) and overall (1.25 km; 95% CI: 0.15; 2.35) were signif-
models. For patients who were lost to follow-up or were icantly longer in the brace group (P ¼ 0.03, P ¼ 0.04 and
780 R. W. Brouwer et al.: Brace treatment for OA of the knee

Table I Table II
Baseline characteristics of the study population Differences between the intervention and control groups for primary
and secondary outcomes at 3, 6 and 12 months
Total Brace Control
group group group Analysis in total group (N ¼ 117)
N ¼ 117 N ¼ 60 N ¼ 57
Mean difference Effect
Male, n (%) 59 (50) 31 (52) 28 (49) (95% CI) size
BMI (kg/m2), 28.5 (4.8) 27.8 (4.3) 29.4 (5.2)
mean (SD) 3 months follow-up
Duration of 69.9 (90.2) 80.3 (101.1) 59.0 (76.6) Pain severity (VAS, 0e10)
0.73 (
1.62;0.16) 0.3
complaints Knee function (HSS, 0e100) 3.5 (
0.24;7.24)* 0.3
(months), Walking distance (km) 1.21 (0.12;2.28)** 0.3
mean (SD) Quality of life (EQ-5D, 0e1) 0.03 (
0.05;0.12) 0.1
Severe OA, n (%)* 6 months follow-up
Grade 1 74 (63) 41 (68) 33 (58) Pain severity (VAS, 0e10)
0.58 (
1.48;0.32) 0.3
Grade 2 43 (37) 19 (32) 24 (42) Knee function (HSS, 0e100) 3.2 (
0.58;6,98)* 0.3
Walking distance (km) 0.79 (
0.40;1.98) 0.2
Varus alignment, n 95 48 47 Quality of life (EQ-5D, 0e1) 0.01 (
0.08;0.10) 0.0
HKA-angle, 188.2 (4.1) 187.9 (3.4) 188.5 (4.4)
mean (SD)y 12 months follow-up
Pain severity (VAS, 0e10)
0.81 (
1.76;0.14)* 0.4
Valgus alignment, n 22 12 10 Knee function (HSS, 0e100) 3.0 (
1.05;7.05) 0.3
HKA-angle, 174.3 (3.7) 174.3 (3.9) 174.3 (3.6) Walking distance (km) 1.34 (0.05;2.63)** 0.4
mean (SD)y Quality of life (EQ-5D, 0e1) 0.01 (
0.08;0.10) 0.0
Pain severity, 6.0 (2.2) 6.6 (2.4) 5.5 (2.0) Overall
mean (SD) Pain severity (VAS, 0e10)
0.63 (
1.38;0.12)* 0.3
HSS score, 66.9 (10.9) 64.9 (12.0) 69.0 (9.5) Knee function (HSS, 0e100) 3.0 (
0.41;6.41)* 0.3
mean (SD) Walking distance (km) 1.25 (0.15;2.35)** 0.4
Walking distance 3.3 (3.7) 2.6 (3.1) 4.0 (4.0) Quality of life (EQ-5D, 0e1) 0.02 (
0.05;0.09) 0.1
(km), mean (SD)
Quality of life, 0.53 (0.28) 0.50 (0.30) 0.56 (0.26) *P < 0.1, **P < 0.05.
mean (SD) The mean difference is adjusted for baseline values for age, gen-
Analgesic use der, BMI, duration of complaints, severity of knee OA, (alignment),
None, n (%) 47 (40.5) 28 (47) 19 (34) baseline pain severity, knee function, walking distance, medication,
When needed, 18 (15.5) 9 (15) 9 (16) and quality of life.
n (%)
Daily, n (%) 51 (44) 23 (38) 28 (50) secondary OA (n ¼ 47) for knee function (estimate HSS
4.87; P ¼ 0.06) compared to the effect of the brace in pa-
*OA according to Ahlbäck.
tients with primary OA (n ¼ 70) (estimate HSS 1.59;
yHipeKneeeAnkle angle: an angle of more than 180( denoted
P ¼ 0.51). The effect for pain severity showed a similar
a varus alignment.
trend, but not as pronounced as for knee function.
Explorative subgroup analyses stratified for age showed
P ¼ 0.02, respectively) (Table II). Effect sizes at the three a slightly better effect of the brace in patients younger
assessment points ranged from 0.2 to 0.4. No significant dif- than 60 years (n ¼ 60) for knee function (estimate HSS
ferences in quality of life evaluations were found between 3.38; P ¼ 0.13) compared to the effect of the brace in pa-
the intervention and control groups. tients aged 60 years and older (n ¼ 57) (estimate HSS
All our analyses were adjusted for baseline use of anal- 2.48; P ¼ 0.38). The effect for pain severity showed a similar
gesics (none, when needed, and daily). Also during the fol- trend, but not as pronounced as for knee function.
low-up we scored the analgesic use: there was increasingly
lower medication use for each follow-up period in the brace DISCONTINUATION OF TREATMENT DURING FOLLOW-UP
group compared to the control group.
During the 12-month follow-up period, 25 patients in the
brace group and 14 patients in the control group changed
SUBGROUP ANALYSIS
their initial treatment, mostly at around 3 months; in both
Explorative subgroup analyses stratified for the alignment groups the main reason for this was no effect of treatment
showed a better and significant effect of the brace in the (74%) (Table III). Other reasons for stopping brace treat-
varus group (n ¼ 95) for the knee function score (estimate ment were skin irritation and bad fit, and three patients
HSS 4.15; P ¼ 0.03) compared to the effect of the brace stopped because the symptoms strongly reduced. Change
in the valgus group (n ¼ 22) (estimate HSS 0.20; in treatment during follow-up included surgery (e.g., high tib-
P ¼ 0.96). The effect for the pain severity showed a similar ial osteotomy, n ¼ 8; knee arthroplasty, n ¼ 16). Thirteen
trend, but not as pronounced as for knee functions. patients changed brace treatment for standard conservative
Explorative subgroup analyses stratified for degree of OA treatment (Fig. 1).
showed a better effect of the brace in patients with severe
OA (n ¼ 43) for pain severity (estimate VAS
1.31;
Discussion
P ¼ 0.10) compared to the effect of the brace in patients
with mild OA (n ¼ 74) (estimate VAS
0.21; P ¼ 0.65). The results of this study indicate that a brace intended to
The effect for the knee functions showed a similar trend, reduce load offers small additional beneficial effect in knee
but not as pronounced as for pain severity. OA compared with conservative treatment alone.
Explorative subgroup analyses stratified for the origin of Many of the measured outcomes showed only a borderline
OA showed a better effect of the brace in patients with significant difference. We decided in advance to perform
Osteoarthritis and Cartilage Vol. 14, No. 8 781

Table III Although the kind of intervention did not allow blinding of
Data on patients who stopped the treatment to which they were patients, methodological strength would have been gained
originally assigned by blinding the assessor for the functional outcome mea-
Brace group Control group surement (HSS knee score), e.g., by using an independent
N ¼ 60 N ¼ 57 assessor. However, because the same effects were found
Stopped with treatment (total) 25 14
for the self-evaluated functional outcome (i.e., walking dis-
Within 3 months 16 6 tance), and because the caregiver had no definite opinion
Between 3 and 6 months 6 6 about the effectiveness of the brace, we assume that the
Between 6 and 12 months 3 2 assessments made by the caregiver had minimal or no bias.
Second, several patients stopped brace treatment during
Alternative treatment
High tibial osteotomy 5 3 the 12-month follow-up, mainly due to noneffectiveness.
Unicompartment knee 3 0 Moreover, most of these patients stopped brace treatment
prosthesis before the first 3-month assessment point; this may be
Total knee prosthesis 3 10 too short a period (in the absence of adverse side-effects)
(Other) brace 1 1 for a beneficial effect to emerge.
Only usual conservative 13 0 Third, although we used the HSS knee function score
care (frequently used in orthopedic research), the WOMAC-func-
Reason for stopping tion seems to have become the function score of
treatment choice21,29. Nevertheless, in view of the very high correla-
No effect 15 14 tion between the WOMAC-pain and WOMAC-function,
Skin irritation 2 e some have suggested that the WOMAC-function measures
Bad fit 2 e pain rather than function30.
Minimal symptoms 3 e
Several reasons 3 e
CLINICAL IMPLICATIONS

two-sided testing. However, looking at the comparison (stan- Although a brace intended to reduce load indicates
dard care vs standard care in combination with brace treat- a small additional beneficial effect in conservative treatment
ment), one-sided testing would have been allowed of knee OA during a 12-month follow-up, many patients do
because one expects an additional beneficial effect of the not adhere to the brace treatment in the long run, either
additional treatment. Had we used one-sided testing, almost because the positive effects are too small or because the
all of our primary outcomes would have been statistically adverse effects are too large.
significant. Based on explorative subgroup analysis in the present
Studies comparing the effectiveness of braces to treat OA study, a brace intended to reduce load seems to be a treat-
of the knee are scarce: only one randomized controlled trial ment option for younger patients with unicompartmental OA
has evaluated the effectiveness of braces for patients with with varus alignment, because few conservative alterna-
unicompartmental OA of the knee with varus alignment16. tives have proven effective31,32. Correction osteotomy in
The results of the present study confirm those of the latter relatively young patients with unicompartmental OA has
study, which included 119 patients who were followed for 6 good results, but this surgery can present complica-
months. In that study, a valgus brace was compared with tions33,34. Knee arthroplasty for younger patients is not
a neoprene sleeve and with standard medical treatment (con- recommended because the degree of patient activity and
trol group); the brace group showed greater improvement life expectancy means that the arthroplasty may wear out
compared with the sleeve group, which showed greater im- and/or loosen35. For older (aged >60 years) less active
provement compared with the control group. patients, however, brace treatment seems less effective
Also a cross-over study showed in 12 patients with OA of and therefore standard conservative treatment is recom-
the medial compartment and a varus alignment significant mended. If symptoms persist in this older group, a knee
improvements gait with a valgus corrective brace compared arthroplasty (nowadays a routine procedure with good
with a neutral brace25. long-term results) can be considered36,37.
In our study valgisation bracing in medial compartment
OA was more effective than varisation bracing in lateral
FUTURE RESEARCH
compartment OA. This might indicate that the unloading
theory does not apply in patients with lateral compartment Besides the above-mentioned practical considerations,
and a valgus alignment. a larger study is needed to identify predictive factors for
Moreover, the knee adduction moment during the stance the success of brace treatment. Particularly for the valgus
phase of walking causes mainly medial loading6,26. Possi- group a larger study population is needed to identify what
bly, a simple sleeve will show the same or more effect in pa- type of brace will benefit this group. In addition, brace treat-
tients with lateral compartment OA due to increased ment should be compared with using a neoprene sleeve
proprioception27,28. This was also discussed by Kirkley with possibly better treatment adherence.
et al. who reported an effect of a neoprene sleeve in unicom-
partmental OA with varus alignment16. Therefore, in general Acknowledgment
OA of the knee where there is no specific compartment to
unload, a sleeve or a neutral brace may also be beneficial We thank RMD Bernsen for the statistical analysis. Contrib-
due to possible increased proprioception and stability27. utors: RWB, JANV, and LNJEMC had the idea of the study.
RWB and SMABZ designed the study. RWB and TMR
STUDY LIMITATIONS
collected the data. SMABZ was responsible for the interpre-
tations of the data. RWB wrote the manuscript. All authors
First, the assessor was also the caregiver as well as the critically commented on the paper and gave their final
one who informed the patient about the aims of the study. approval of the version to be published. RWB is the
782 R. W. Brouwer et al.: Brace treatment for OA of the knee

guarantor. Funding: This study was supported by the Re- 16. Kirkley A, Webster-Bogaert S, Litchfield R, Amendola A,
volving Fund (RF01-12) of the Erasmus University Medical MacDonald S, McCalden R, et al. The effect of bracing
Centre Rotterdam, The Netherlands. Competing interests: on varus gonarthrosis. J Bone Joint Surg Am 1999;81:
None declared. Ethical approval: Medical Ethics Commit- 539e48.
tees of the Erasmus Medical Centre Rotterdam and Leyen- 17. Komistek RD, Dennis DA, Northcut EJ, Wood A,
burg Hospital, The Hague, The Netherlands. Parker AW, Traina SM. An in vivo analysis of the effec-
tiveness of the osteoarthritic knee brace during heel-
strike of gait. J Arthroplasty 1999;14:738e42.
18. Lindenfeld TN, Hewett TE, Andriacchi TP. Joint loading
References with valgus bracing in patients with varus gonarthrosis.
Clin Orthop Relat Res 1997;344:290e7.
1. Felson DT, Zhang Y. An update on the epidemiology of 19. Brouwer RW, Jakma TC, Verhagen AP, Verhaar JAN,
knee and hip osteoarthritis with a view to prevention. Bierma-Zeinstra SMA. Braces and orthoses for treat-
Arthritis Rheum 1998;41:1343e55. ing osteoarthritis of the knee. Cochrane Database
2. Healy WL, Iorio R, Ko J, Appleby D, Lemos DW. Impact Syst Rev 2005;25(1): CD004020.
of cost reduction programs on short-term patient out- 20. Ahlbäck S. Osteoarthrosis of the knee. A radiographic in-
come and hospital cost of total knee arthroplasty. vestigation. Acta Radiol Diagn 1968;(Suppl 277):7e72.
J Bone Joint Surg Am 2002;84:348e53. 21. Insall JN, Ranawat CS, Aglietti P, Shine J. A compari-
3. Grelsamer RP. Unicompartmental osteoarthrosis of the son of four models of total knee-replacement prosthe-
knee. J Bone Joint Surg Am 1995;77:278e92. ses. J Bone Joint Surg Am 1976;58:754e65.
4. Tetsworth K, Paley D. Malalignment and degenerative 22. Nord E. EuroQol: health-related quality of life measure-
arthropathy. Orthop Clin North Am 1994;25:367e77. ment. Valuations of health states by the general public
5. Sharma L, Cahue S, Song J, Hayes K, Pai YC, in Norway. Health Policy 1991;18:25e36.
Dunlop D. Physical functioning over three years in 23. Magyar G, Toksvig-Larsen S, Lindstrand A. Open
knee osteoarthritis: role of psychosocial, local wedge tibial osteotomy by callus distraction in
mechanical, and neuromuscular factors. Arthritis gonarthrosis. Operative technique and early results
Rheum 2003;48:3359e70. in 36 patients. Acta Orthop Scand 1998;69:147e51.
6. Cooke TD, Harrison L, Khan B, Scudamore A, 24. Jordan KM, Arden NK, Doherty M, Bannwarth B,
Chaudhary MA. Analysis of limb alignment in the path- Bijlsma JW, Dieppe P, et al. EULAR recommendations
ogenesis of osteoarthritis: a comparison of Saudi 2003: an evidence based approach to the manage-
Arabian and Canadian cases. Rheumatol Int 2002; ment of knee osteoarthritis: report of a Task Force of
22:160e4. the Standing Committee for International Clinical Stud-
7. Foley A, Halbert J, Hewitt T, Crotty M. Does hydrotherapy ies Including Therapeutic Trials (ESCISIT). Ann
improve strength and physical function in patients with Rheum Dis 2003;62:1145e55.
osteoarthritis e a randomised controlled trial comparing 25. Richards JD, Sanchez-Ballester J, Jones RK, Darke N,
a gym based and a hydrotherapy based strengthening Livingstone BN. A comparison of knee braces during
programme. Ann Rheum Dis 2003;62:1162e7. walking for the treatment of osteoarthritis of the medial
8. Fransen M, Crosbie J, Edmonds J. Physical therapy is compartment of the knee. J Bone Joint Surg Br 2005;
effective for patients with osteoarthritis of the knee: 87:937e9.
a randomized controlled clinical trial. J Rheumatol 26. Johnson F, Leitl S, Waugh W. The distribution of load
2001;281:156e64. across the knee. A comparison of static and dynamic
9. Fransen M, Crosbie J, Edmonds J. Isometric muscle force measurements. J Bone Joint Surg Br 1980;62:
measurement for clinicians treating patients with osteo- 346e9.
arthritis of the knee. Arthritis Rheum 2003;49:29e35. 27. Birmingham TB, Kramer JF, Kirkley A, Inglis JT,
10. Goorman SD, Watanabe TK, Miller EH, Perry C. Func- Spaulding SJ, Vandervoort AA. Knee bracing for me-
tional outcome in knee osteoarthritis after treatment dial compartment osteoarthritis: effects on propriocep-
with hylan G-F 20: a prospective study. Arch Phys tion and postural control. Rheumatology (Oxford)
Med Rehabil 2000;81:479e83. 2001;40:285e9.
11. Hoffmann S, Theiler R. Physiotherapy in osteoarthritis e 28. Hassan BS, Mockett S, Doherty M. Influence of elastic
a review of literature on conservative therapy of knee bandage on knee pain, proprioception, and postural
and hip osteoarthritis. Ther Umsch 2001;58:480e6. sway in subjects with knee osteoarthritis. Ann Rheum
12. Huang MH, Chen CH, Chen TW, Weng MC, Wang WT, Dis 2002;61:24e8.
Wang YL. The effects of weight reduction on the reha- 29. Bellamy N. WOMAC: a 20-year experiential review of
bilitation of patients with knee osteoarthritis and a patient-centered self-reported health status ques-
obesity. Arthritis Care Res 2000;13:398e405. tionnaire. J Rheumatol 2002;29:2473e6.
13. Hurley MV, Scott DL. Improvements in quadriceps sen- 30. Stratford PW, Kennedy DM. Does parallel item content
sorimotor function and disability of patients with knee on WOMAC’s pain and function subscales limit its
osteoarthritis following a clinically practicable exercise ability to detect change in functional status? BMC
regime. Br J Rheumatol 1998;37:1181e7. Musculoskleletal Disorders 2004;5:1e9.
14. Hewett TE, Noyes FR, Barber-Westin SD, Heckmann TP. 31. Lo GH, LaValley M, McAlindon T, Felson DT. Intra-
Decrease in knee joint pain and increase in function in articular hyaluronic acid in treatment of knee osteo-
patients with medial compartment arthrosis: a prospec- arthritis: a meta-analysis. JAMA 2003;290:3115e21.
tive analysis of valgus bracing. Orthopedics 1998;21: 32. Richy F, Bruyere O, Ethgen O, Cucherat M, Henrotin Y,
131e8. Reginster JY. Structural and symptomatic efficacy of
15. Katsuragawa Y, Fukui N, Nakamura K. Change of bone glucosamine and chondroitin in knee osteoarthritis:
mineral density with valgus knee bracing. Int Orthop a comprehensive meta-analysis. Arch Intern Med
1999;23:164e7. 2003;163:1514e22.
Osteoarthritis and Cartilage Vol. 14, No. 8 783

33. Coventry MB, Ilsrup DM, Wallrichs SL. Proximal tibial study from the Norwegian arthroplasty register
osteotomy. A clinical long-term study of 87 cases. 1994e2000. Acta Orthop Scand 2002;73:117e29.
J Bone Joint Surg Am 1993;75:196e201. 36. Robertsson O, Knutson K, Lewold S, Lidgren L. The
34. Stukenborg-Colsman C, Wirth CJ, Lazovic D, Wefer A. Swedish Knee Arthroplasty Register 1975e1997: an
High tibial osteotomy versus unicompartmental joint update with special emphasis on 41,223 knees oper-
replacement in unicompartmental knee joint osteo- ated on in 1988e1997. Acta Orthop Scand 2001;72:
arthritis: 7e10-year follow-up prospective randomised 503e13.
study. Knee 2001;8:187e94. 37. Watanabef H, Akizuki S, Takizawa T. Survival analysis of
35. Furnes O, Espehaug B, Lie SA, Vollset SE, a cementless, cruciate-retaining total knee arthroplasty.
Engesaeter LB, Havelin LI. Early failures among Clinical and radiographic assessment 10 to 13 years
7,174 primary total knee replacements: a follow-up after surgery. J Bone Joint Surg Br 2004;86:824e9.