The Journal of Rheumatology Volume 28, no.

Physical therapy is effective for patients with osteoarthritis of the knee: a

randomized controlled clinical trial.
M Fransen, J Crosbie and J Edmonds

J Rheumatol 2001;28;156-164
http://www.jrheum.org/content/28/1/156

1. Sign up for our monthly e-table of contents

http://www.jrheum.org/cgi/alerts/etoc

2. Information on Subscriptions
http://jrheum.com/subscribe.html

3. Have us contact your library about access options

[email protected]

4. Information on permissions/orders of reprints

http://jrheum.com/reprints.html

The Journal of Rheumatology is a monthly international serial edited by Earl D.

Silverman featuring research articles on clinical subjects from scientists working in
rheumatology and related fields.

Downloaded from www.jrheum.org on May 12, 2016 - Published by The Journal of

Rheumatology
Physical Therapy Is Effective for Patients with
Osteoarthritis of the Knee: a Randomized Controlled
Clinical Trial
MARLENE FRANSEN, JACK CROSBIE, and JOHN EDMONDS

ABSTRACT. Objective. To assess the effectiveness of physical therapy, given either as an individually attended treat-
ment or in a small group format, in terms of pain, physical function, and health related quality of life
for patients with osteoarthritis (OA) of the knee.
Methods. After 2 baseline assessments, 126 patients were randomized into one of 3 allocation arms:
individual treatments (n = 43), small group format program (n = 40), and waiting list control (n = 43).
After reassessment at 8 weeks, patients allocated to waiting list control were randomized into one of the
2 active treatment arms. Assessments included both self-report measures (WOMAC, SF-36) and objec-
tive measures of physical performance (gait analysis and muscle strength).
Results. Both physical therapy treatment allocations resulted in significant improvements in pain, phys-
ical function, and health related quality of life above the control group (standardized response mean 0.36
to 0.65). Improvements in the self-report measures were substantiated by significant correlated
improvements in knee extensor strength and fast walking speed (rho 0.36–0.42). There were no signif-
icant differences in effectiveness between the 2 physical therapy allocations for any of the measured
outcomes. Improvements gained were maintained for at least 2 months. Responsiveness to treatment
was modified by loss of medial joint space width, the interaction being significant for physical function,
gait, and knee extensor strength.
Conclusion. Physical therapy, either as an individually delivered treatment or in a small group format,
is an effective intervention for patients with knee OA. Responsiveness to this 8 week intervention was
modified by loss of medial joint space width. (J Rheumatol 2001;28:156–64)

Key Indexing Terms:

OSTEOARTHRITIS KNEE EXERCISE GAIT STRENGTH FUNCTION

Symptomatic osteoarthritis (OA) of the knee occurs in about knee OA11 could identify only 5 studies with “acceptable
6.1% of adults aged 30 and over1, with prevalence increasing validity”12-16. Three further randomized clinical trials with
with age2,3. A large community based survey of noninstitu- possible acceptable validity have been published since this
tionalized elders revealed that knee OA accounted for the review17-19. Unfortunately, half of these 8 studies had insuffi-
highest percentage of disability in walking, stair climbing, and cient power to establish even a medium effect13,14,16,17.
housekeeping4. The aging of the population will result in Further, studies with high internal validity and sufficient
exponential growth in the global burden of pain, physical dis- power can suffer from limited generalizability by assessing
ability, and dependency5-7, which will be particularly marked either relatively costly programs not easily accessible even in
in “young” countries such as USA, Canada, and Australia8. developed countries12,15 or assessing programs delivered by a
It is generally accepted that exercise potentially reduces single treating physical therapist14,19. It is also suggested that
knee pain and limits decline of physical function in people studies using volunteer samples would have limited applica-
with knee OA9,10. A systematic review of randomized clinical bility to the clinical situation, as it has been shown that vol-
trials examining the effectiveness of exercise for people with unteers from the community are unrepresentative of the pop-
ulation seeking treatment20,21.
Symptomatic knee OA progresses with a pattern of disease
From the Department of Rheumatology, St.George Hospital, Gray Street, related impairments such as joint pain, loss of lower limb
Kogarah NSW 2217, Australia.
muscle strength22-24, gait disability25-27, and reduced aerobic
Supported by the National Health and Medical Research Council of
Australia and the Arthritis Foundation Australia. fitness28,29. Treatment intensity is often limited by these dis-
M. Fransen, MPH, PhD; J. Crosbie, PhD, Associate Professor, School of ease related impairments together with significant comorbidi-
Physiotherapy, University of Sydney; J. Edmonds, MB, BS, FRACP, ty in this aging population. An effective treatment “dosage”
Director, Department of Rheumatology, St. George Hospital.
may therefore require lengthy, but often economically prohib-
Address reprint requests to M. Fransen, Department of Rheumatology,
St.George Hospital, Gray Street, Kogarah NSW 2217, Australia. itive, treatment duration. Due to the fairly predictable pattern
E-mail: [email protected] of disease related impairments, knee OA would appear to be a
Submitted December 31, 1999 revision accepted July 13, 2000. condition suited to group format intervention programs. In

Personal non-commercial use only. The Journal of Rheumatology Copyright © 2001. All rights reserved.
156
Downloaded from www.jrheum.org on May 12, 2016 - Published byTheThe Journal of
Journal of Rheumatology 2001; 28:1
Rheumatology
theory, a small group format has the potential to allow time for radiographs were magnification controlled using a 3/8" stainless steel ball
sufficient individual monitoring as well as achieve a more mounted in a perspex tube taped to head of fibula. JSW was later digitally
measured, a mean of 3 readings on one day providing the final measurement.
efficient use of health care resources. The less obvious poten- Test-retest (1 month) reliability on 18 randomly selected radiographs (previ-
tial may lie in increased patient access and longterm adher- ous markings removed) was calculated as ICC 2,1 = 0.98 (95% confidence
ence due to the influence of group association. interval, CI, 0.94–0.99).
This study follows on from an uncontrolled pilot study at At each assessment, as well as collecting information concerning medica-
this center investigating the effectiveness of an 8 week small tions and usual physical activity level, participants were asked to complete
both the Western Ontario and McMaster Universities Arthritis Index
group format program for patients referred for physical thera- (WOMAC) and the Medical Outcomes Study Short Form (SF-36). The
py treatment30. The current study contrasts in many ways to WOMAC is a validated disease-specific self-report questionnaire using 100
most randomized controlled clinical trials studies evaluating mm visual analog scales (VAS) to assess “currently experienced” pain (5
exercise for people with knee OA published to date. The cur- questions) and physical disability (17 questions)35. The SF-3636,37 is a vali-
rent pragmatic study extended generalizability by using a dated, extensively used self-report HRQOL questionnaire measuring 8
dimensions of health status38,39. To increase precision and reduce the number
large number of physical therapists to provide treatment, of statistical comparisons needed, the originators of the SF-36 have devel-
recruited patients initially seeking treatment, assessed 2 feasi- oped algorithms to calculate 2 psychometrically based summary measures:
ble programs as routinely provided in the clinic, used widely the Physical Component Summary Scale Score (PCS) and the Mental
validated self-report and objective outcome measures with Component Summary Scale Score (MCS)40,41. The PCS and MCS are norm-
established normative population data, and has assessed treat- based scores so that each has a mean of 50 and a standard deviation of 10 in
the general US population.
ment sustainability. Furthermore this study was designed to At each assessment, patients were also required to participate in a quanti-
analyze certain patient characteristics that were deemed by a tative gait analysis and isometric muscle testing. The quantitative gait vari-
group of clinicians to be plausible predictors of treatment ables of speed, cadence, and stride length were analyzed using an 8 meter
responsiveness. electric footswitch walkway. The system and the standardized testing proto-
The primary hypothesis was that physical therapy (individ- col were the same as that for which validity and reliability had been investi-
gated in earlier studies42,43. The gait variables used in this study were mean
ual treatments or group format) can effect improvement in speed (cm•s–1), mean cadence (steps·min–1), and mean stride length (cm)
self-reported pain, physical function, and health related qual- extrapolated from 2 trials, after a familiarization trial, at a fast self-selected
ity of life (HRQOL) in patients with knee OA referred for speed42,43. For the isometric muscle strength testing, patients were seated on
treatment. The secondary hypothesis was that the group for- a high metal frame chair with the thigh well supported, the foot free, and the
mat is more beneficial than individual treatments in terms of knee passively drawn into 90˚ flexion by gravity. Bilateral isometric knee
extensor and flexor muscle strength were tested in this position using an
self-reported pain, physical function, and HRQOL for patients Xtran load cell (Model S1W, Applied Measurement Australia Pty. Ltd.) fixed
with knee OA. The tertiary hypotheses we tested were onto the metal framework of the chair and connected to a software program
whether physical therapy can significantly improve objective sampling at 80 Hz. Both muscle groups were tested 3 times on each limb in a
measures of physical performance, whether certain baseline set sequence at each assessment, the final score being the mean peak force
characteristics [age, body mass index (BMI), symptom dura- attained for each muscle group. One week test-retest (prerandomization
assessments) measurement reliability was calculated for the knee extensors as
tion, or medial joint space width (JSW)] can predict treatment intraclass correlation coefficient (2,1) = 0.93 (95% CI 0.90–0.95) and for the
responsiveness and if improvements could be maintained 2 knee flexors as ICC(2,1) = 0.87 (95% CI 0.82–0.91).
months after completion of formal treatment. After the 2 baseline assessments, the patients were randomly allocated by
concealed ballot in blocks of 18, according to a random numbers table and
MATERIALS AND METHODS with a clear audit trail, by hospital administrative staff. Allocations were
All patients, living in the community and referred by physicians for physical sealed in numbered opaque envelopes prior to recruitment. The 3 allocations
therapy treatment at a large hospital outpatient department from May 1997 were as follows. (1) Individual treatments. The choice, frequency, and dura-
until February 1999, with a diagnosis of knee pain or knee arthritis, were con- tion of individual treatments within an 8 week period were at the discretion
tacted to assess eligibility. Patients were invited to participate if they were of the treating physical therapist. Treatment procedures and duration were
aged 50 years and over, had knee pain on most days of the past month, and recorded and verified. (2) Group format program. The group program ran,
had evidence of radiographic disease31. Patients were excluded if they had under the supervision of a physical therapist, for 1 hour twice a week for 8
intraarticular cortisone injections within the past 2 months, lower limb joint weeks and was supplemented with a home exercise program. For safety and
arthroplasty, unstable cardiac comorbidity precluding exercise at 50–60% individual supervision reasons, the group size was restricted to a maximum of
maximum heart rate, or other comorbidity affecting gait. More than 90% of 6 patients. The program content is outlined in Appendix 1 and was the same
the patients considered eligible agreed to participate in the study as participa- as that for which efficacy was described in a recent uncontrolled trial30. (3)
tion resulted in a 67% chance of evading the usual 4 to 8 week physical ther- Control. Patients allocated to remain on the waiting list were assessed before
apy waiting list for chronic conditions. All participants were required to sign and after an 8 week nonintervention period. These patients were then ran-
an informed consent. domly allocated to one of the 2 active treatments and reassessed at Week 16.
All participants were assessed 4 times, using a strictly standardized pro- Participants were not informed that there were 2 different delivery modes
tocol: twice with an interval of one week prior to randomization (Week 00); of physical therapy involved in the allocation process, and individual treat-
postintervention or waiting list control period (Week 8); followup, or postin- ments and group exercise sessions were scheduled when possible on different
tervention for controls (Week 16). The chief investigator, who was not days of the week. Patients allocated to the waiting list were asked to contin-
involved in any of the treatments and remained mostly masked to treatment ue their usual prestudy medication and physical activity regime as far as was
allocation, carried out the assessments. At baseline, demographic and radi- ethically possible.
ographic data were collected. All participants were required to obtain a To absorb statistical regression and subject adaptability to the assessment
weight-bearing, semiflexed radiograph of their most painful knee32-34. The measures or equipment, mean data derived from the 2 prerandomization

Personal non-commercial use only. The Journal of Rheumatology Copyright © 2001. All rights reserved.
Downloaded from www.jrheum.org on May 12, 2016 - Published by The Journal of
Fransen, et al: Exercise and knee OA 157
Rheumatology
assessments were used as the baseline. Sample size estimates were based on RESULTS
independent T tests of self-reported pain on the 100 mm VAS of the WOMAC Radiographs were obtained of 114 of the 126 participants
with a 2:1 treatment:control allocation ratio. The clinically significant differ-
ence (15 mm), as well as the standard deviation (22 mm), was based on evi-
(90.5%). Attrition numbers during the course of the study are
dence from the literature and results of a previous study30,44. At an overall sig- given in Figure 1. One hundred twenty-eight patients agreed
nificance level of 2 tailed p = 0.05 and allowing for a 10% loss to followup, to participate in the study. Two withdrew prior to randomiza-
it was calculated that 116 subjects were needed for the study to have a 90% tion because of unrelated general poor health and minor
probability of finding a treatment effect. Data were analyzed per intention-to- abdominal trauma. Five patients dropped out of the 2 physical
treat, assuming no change for subjects unavailable for followup assessment.
Analyses consisted primarily of mean changes with 95% CI and standardized
therapy treatment groups (individual and group format) at var-
response means (SRM). Multiple linear regressions were used to analyze the ious stages due to acceptance of cortisone injection, accep-
significance of group allocation on self-report and physical performance tance of knee arthroplasty, family circumstances, severe asth-
changes scores adjusted for the associated baseline score. Correlation analy- ma related symptoms, and not responding to appointments.
sis was used to establish if changes in self-report measures were plausibly Two waiting list control subjects were unavailable for the
associated with changes in objective measures of physical performance (iso-
metric muscle strength and gait). Split median stratification by age, body
Week 8 assessment: not responding to appointments. After 8
mass index (BMI), symptom duration, and medial JSW was used to assess weeks on the waiting list, controls were randomly allocated to
possible predictors of treatment responsiveness. one of the 2 forms of physical therapy treatment. Three wait-

Figure 1. The procedure of randomization and withdrawal during the study.

Personal non-commercial use only. The Journal of Rheumatology Copyright © 2001. All rights reserved.
158
Downloaded from www.jrheum.org on May 12, 2016 - Published byTheThe Journal of
Journal of Rheumatology 2001; 28:1
Rheumatology
ing list controls were unavailable for randomization: moving Table 1. Baseline 1 characteristics, individual treatments, group format, and
to another region, acceptance of hydrotherapy, acceptance of control (Week 00).
arthroscopy. After physical therapy (Week 16), 6 of the wait- Individual, Group, Control,
ing list patients did not attend for posttreatment assessment: n = 43, n = 40, n = 43,
intraarticular cortisone injection, cardiac problems, accep- mean (SD) mean (SD) mean (SD)
tance of total knee arthroplasty, remission of severe back pain,
ankle injury, and not responding to appointments. Age, yrs 68.5 (8.7) 65.3 (7.1) 66.1 (10.3)
Sex, % female 74 78 67
To increase generalizability, 24 physical therapists were BMI 30.0 (4.6) 29.9 (5.9) 28.3 (4.5)
involved in the individual treatments and 4 different physical Symptoms 1,2–5, > 5 yrs 5, 18, 20 6, 18, 16 5, 21, 17
therapists supervised the group format program. Individual Medial JSW, mm 2.0 (1.4) 2.2 (1.3) 2.0 (1.4)
treatments consisted almost universally of at least 20 minutes Lateral JSW, mm 4.9 (1.8) 5.0 (2.1) 4.9 (1.7)
of muscle strengthening exercise or manual techniques aimed WOMAC pain, 100–0 59.5 (20.0) 61.4 (18.9) 65.8 (19.4)
WOMAC function, 100–0 58.5 (18.8) 63.1 (21.0) 60.0 (20.5)
at increasing range of motion and 5–10 minutes of an electro- SF-36 PCS, mean = 50 31.4 (6.6) 34.3 (9.1) 34.8 (8.2)
physical agent such as heat, ultrasound, laser, or interferential SF-36 MCS, mean = 50 42.7 (7.2) 44.5 (7.8) 42.9 (7.0)
therapy. The mean number of half-hour individual treatments Knee extensors, N 167.9 (72.9) 171.0 (63.4) 173.3 (67.1)
attended was 7 (range 2–4). About 90% of the patients allo- Knee flexors, N 89.5 (38.2) 99.8 (35.6) 100.1 (40.9)
cated to the group format program attended at least 12 of the Fast speed, cm·s–1 121.5 (28.5) 135.7 (23.1) 127.8 (24.1)
Fast cadence, steps·min–1 117.6 (14.3) 122.2 (13.6) 117.6 (10.3)
16 sessions. Fast stride length, cm 123.2 (21.1) 133.2 (17.8) 130.3 (22.2)
The WOMAC scores were reverse scored (100 = no pain
or difficulty, 0 = extreme pain or difficulty), so that for all out- N: newtons, BMI: body mass index, JSW: joint space width, PCS: physical
come measures higher scores are better scores. component score, MCS: mental component score.
Primary hypothesis. The initial 3 allocation groups were com-
parable at baseline 1 (Week 00, Figure 1) for age, sex, BMI, of treatment above control, as assessed by the SRM, would be
symptom duration, medial JSW, and self-report measures rated as medium for pain, physical function, and the SF-36
(Table 1). The primary hypothesis, that physical therapy (indi- MCS, and small for the SF-36 PCS.
vidual treatments or group format) can effect improvements in Secondary hypothesis. After the waiting list controls were ran-
pain, physical function, and HRQOL, is substantiated by the domized to one of the 2 forms of physical therapy treatment
results of this study. Patients originally allocated to physical delivery (Figure 1), the total individual treatment group (n =
therapy had significantly decreased pain and physical dys- 62) was comparable with the total group format group (n =
function (WOMAC) as well as improved HRQOL (SF-36) at 59) (Table 3). The secondary hypothesis, that the group format
Week 8 (Table 2). In contrast, patients allocated to remain on program is more beneficial than individual treatments, could
the waiting list had no significant changes in any of these not be substantiated by our results (Table 4). Both forms of
measures at Week 8. To avoid a Type I error at the overall sig- physical therapy achieved significant improvements. For pain
nificance level of 2P < 0.05, the required significance level for and physical function (WOMAC), the immediate effects of
each of the 4 outcomes was adjusted to p < 0.01. Only the treatment (SRM) would be rated as medium for the group for-
generic SF-36 PCS failed to achieve statistical significance mat program and small for the individual treatments. For
for the effect of treatment above control (Table 2). The effects HRQOL (SF-36), the immediate effects of treatment would be

Table 2. Treatment outcomes, treatment (individual and group) vs control (Week 8–Week 00).

Treatment (n = 83), Control (n = 43), p* SRM

mean change (95% CI) mean change (95% CI) (Treatment (Treatment
vs Control) over Control)

WOMAC pain, 100–0 10.6 (6.3, 15.0) –1.5 (–5.5, 2.4) < 0.01 0.65
WOMAC function, 100–0 7.7 (4.2, 11.2) –0.1 (–3.9, 3.7) < 0.01 0.49
SF-36 PCS, mean = 50 3.6 (1.9, 5.3) 0.5 (–1.5, 2.3) 0.05 0.36
SF-36 MCS, mean = 50 2.0 (0.8, 3.3) –0.7 (–1.8, 0.5) < 0.01 0.51
Knee extensors, N 10.8 (4.3, 17.3) –2.4 (–9.2, 4.5) 0.01 0.46
Knee flexors, N 8.4 (4.0, 12.7) –0.6 (–5.5, 5.2) 0.02 0.46
Fast speed, cm·s–1 7.1 (4.7, 9.4) 0.4 (–1.3, 2.1) < 0.01 0.58
Fast cadence, steps·min–1 1.9 (0.7, 3.2) 0.3 (–0.6, 1.3) 0.05 0.26
Fast stride length, cm 4.7 (3.0, 6.3) 0.4 (–1.3, 2.0) < 0.01 0.55

*Significance adjusted for baseline differences in the variable. SRM: standardized response mean, N: newtons.

Personal non-commercial use only. The Journal of Rheumatology Copyright © 2001. All rights reserved.
Downloaded from www.jrheum.org on May 12, 2016 - Published by The Journal of
Fransen, et al: Exercise and knee OA 159
Rheumatology
Table 3. Baseline 2 characteristics, individual treatments and group format, Tertiary hypotheses. This study also revealed that: (1) Both
Week 00 (active treatment) and Week 8 (former controls). forms of physical therapy treatment resulted in significantly
Individual, Group,
increased isometric muscle strength, gait speed, and stride
n = 62, n = 59, length above controls (Table 2). After inclusion of the waiting
mean (SD) mean (SD) list controls into one of the 2 forms of physical therapy, the
group format program appeared to result in consistently supe-
Age, yrs 66.7 (10.1) 66.8 (7.5) rior gains in these measures of physical performance (Table
Sex, % female 75 71
Height, cm 162.8 (8.6) 163.0 (8.5)
4); however, the difference between the 2 active treatments
BMI 29.7 (4.6) 29.0 (5.6) did not reach statistical significance. (2) Changes in self-
Symptoms 1,2–5, > 5 yrs 9, 27, 27 7, 27, 23 reported pain were correlated with changes in isometric exten-
Medial JSW, mm 2.0 (1.4) 2.1 (1.3) sor strength (rho = 0.42) and fast walking speed (rho = 0.36).
Lalteral JSW, mm 5.1 (1.7) 4.7 (2.0) Changes in self-reported physical function were similarly cor-
WOMAC pain, 100–0 60.7 (21.3) 62.7 (18.4)
WOMAC function, 100–0 58.7 (21.1) 62.1 (19.7)
related with changes in isometric extensor strength (rho =
SF-36 PCS, mean = 50 33.2 (8.8) 34.3 (9.1) 0.38) and fast walking speed (rho = 0.38). All reported asso-
SF-36 MCS, mean = 50 43.1 (8.0) 43.5 (7.3) ciations were significant at the p < 0.01 level. (3) A median-
Knee extensors, N 173.7 (75.2) 168.5 (67.1) split stratification according to medial JSW revealed a consis-
Knee flexors, N 97.0 (42.1) 96.9 (39.1) tent trend in treatment effectiveness between the stratified
Fast speed, cm·s–1 125.4 (28.1) 133.3 (22.7)
Fast cadence, steps·min–1 118.1 (12.5) 121.1 (12.8)
groups (Table 5). That the group with greater loss of medial
Fast stride length, cm 126.8 (22.6) 132.4 (20.6) JSW had higher baseline extensor strength and comparable
gait variables is attributed to the significantly greater propor-
N: newtons. tion of men in this group (39% vs 16%). Subjects in the group
with a medial JSW < 1.9 mm (mean 0.9 mm, range 0.2–1.8)
improved markedly less than subjects with a medial JSW >
Table 4. Treatment outcomes, individual treatment vs group format, Week
1.9 mm (mean 3.2 mm, range 1.9–6.7). Indeed the group with
8–Week 00 (active treatment); Week 16–Week 8 (former controls).
more severe loss of medial JSW consistently showed small
Individual, Group, effect sizes, with significant treatment effect only in self-
mean change (95% CI), mean change (95% CI), reported pain. The group with less severe loss of medial JSW
SRM SRM showed moderate to large effect sizes with significant treat-
ment effect for all the measured outcomes. The statistical sig-
WOMAC pain, 100–0 7.7 (3.0, 12.4) 11.4 (6.7, 16.0)
0.42 0.65
WOMAC function, 100–0 6.6 (2.7, 10.5) 8.5 (4.5, 12.5)
0.42 0.55
SF-36 PCS, mean 50 2.7 (0.9, 4.5) 2.3 (0.6, 4.1)
0.38 0.34 Table 5. Outcomes stratified by medial joint space width, individual treat-
SF-36 MCS, mean 50 1.6 (0.2, 2.9) 2.1 (0.8, 3.4) ments or group format.
0.28 0.41
Knee extensors, N 6.5 (0.8, 13.0) 5.9 (0.3, 11.6) Baseline, Change, SRM
0.25 0.27 mean (SD) mean (95% CI)
Knee flexors, N 6.5 (1.9, 11.0) 7.1 (2.4, 11.8)
0.36 0.39 Medial JSW < 1.9 mm, n = 57
Fast speed, cm·s–1 1.6 (0.5, 6.7) 6.3 (3.7, 8.9) WOMAC pain, 100–0 62.4 (20.3) 5.6 (0.6, 10.6) 0.30
0.30 0.63 WOMAC function, 100–0 59.9 (19.3) 2.6 (–1.6, 6.8) 0.17
Fast cadence, steps·min–1 1.1 (–0.4, 2.6) 2.2 (0.8, 3.7) SF-36 PCS, mean = 50 33.6 (9.3) 1.4 (–0.5, 3.3) 0.20
0.19 0.40 SF-36 MCS, mean = 50 44.0 (7.6) 1.3 (–0.0, 2.6) 0.27
Fast stride length, cm 2.3 (–0.1, 4.7) 3.6 (1.8, 5.5) Knee extensors, N 185.1 (76.2) 4.0 (–2.5, 10.5) 0.16
0.24 0.51 Knee flexors, N 98.9 (40.8) 4.9 (–0.0, 9.8) 0.26
Fast velocity, cm·s–1 127.7 (25.2) 1.5 (–1.0, 4.0) 0.16
SRM: standardized response mean, N: newtons. Fast stride length, cm 129.3 (22.0) 1.1 (–0.9, 3.1) 0.14
Medial JSW > 1.9 mm, n = 57
WOMAC pain, 100–0 61.2 (20.3) 11.0 (6.3, 15.5) 0.63
WOMAC function, 100–0 62.1 (22.6) 9.1 (5.7, 12.5) 0.72
SF-36 PCS, mean = 50 33.2 (9.8) 4.5 (2.5, 6.5) 0.59
SF-36 MCS. mean = 50 42.4 (9.8) 2.8 (1.4, 4.3) 0.51
rated as small for both forms of physical therapy. Further, Knee extensors, N 153.0 (60.3) 11.7 (5.9, 17.5) 0.53
increased levels of physical activity and decreased medication Knee flexors, N 91.0 (38.1) 10.3 (5.7, 14.9) 0.59
Fast velocity, cm·s–1 130.1 (26.4) 8.9 (6.0, 11.3) 0.82
use after treatment were similar in both groups. There were no
Fast stride length, cm 128.1 (21.7) 5.2 (3.1, 7.3) 0.67
statistically significant differences between the effects of the
2 modes of physical therapy treatment. JSW: joint space width; SRM: standardized response mean.

Personal non-commercial use only. The Journal of Rheumatology Copyright © 2001. All rights reserved.
160
Downloaded from www.jrheum.org on May 12, 2016 - Published byTheThe Journal of
Journal of Rheumatology 2001; 28:1
Rheumatology
Table 6. Two month followup data, individual treatments and group format. would be more clinically effective than individual treatments,
could not be substantiated. However, data collected during
Week 8, Week 16, this study indicated that the group format program was less
n = 83 mean (SD) mean (SD)
human-resource intensive than the individual treatments. For
WOMAC pain, 100–0 71.1 (18.8) 70.7 (21.3) the individual treatments, 7.02 half-hour treatments extrapo-
WOMAC function, 100–0 68.2 (21.0) 68.7 (21.9) lates conservatively (missed appointments were not included)
SF-36 PCS, mean = 50 36.4 (8.8) 36.8 (9.4) to 3.5 hours of 1:1 treatment time. For the group format pro-
SF-36 MCS, mean = 50 45.5 (7.5) 44.9 (7.8) gram, 16 hours with 6 patients per group extrapolates to 2.7
Knee extensors, N 178.2 (74.5) 179.6 (76.0)
Knee flexors, N 102.1 (38.1) 104.1 (40.1)
hours of 1:1 treatment time. Furthermore, the equipment costs
Fast speed, cm·s–1 135.1 (27.4) 135.0 (27.9) of each delivery mode would be comparable. The group for-
Fast stride length, cm 132.4 (19.2) 132.3 (19.9) mat used 3 stationary bicycles, 3 simple heart rate monitors
(Polar Pacer, Polar Electro Oy), some weights, an exercise
machine allowing both eccentric and concentric lower limb
strengthening, a set of stairs, and a stepper machine
nificance levels of the interactions were: WOMAC physical (Appendix 1). Physical therapists providing individual treat-
function (p = 0.04), SF-36 PCS (p < 0.01), fast gait speed ments at times used various electro-physical agents to supple-
(p < 0.01), fast stride length (p = 0.02), and isometric knee ment exercise: laser (6 patients), interferential (12 patients),
extensor strength (p = 0.05). (4) In contrast, a median-split ultrasound (18 patients), and local heat treatment (10
stratification on age, BMI, and reported symptom duration patients).
(log transformed to attain normal distribution) did not reveal In retrospect, this study was not sufficiently powered to
trends in treatment effectiveness. (5) Followup data collected establish statistical significance for the smaller differences in
at Week 16 (Figure 1) showed that improvements gained in clinical effect realistically anticipated between 2 active treat-
both self-report questionnaires and objective measures of ments compared with the difference between an active treat-
physical performance did not deteriorate over this period ment and a waiting list control group. For example, at an over-
(Table 6). all significance level of p = 0.05, it is calculated that roughly
500 subjects would be needed for the study to have 80% prob-
DISCUSSION ability of establishing a 5 point difference in the WOMAC
The main results of this randomized clinical study are that scores as statistically significant. However, if the secondary
physical therapy, for this sample of referred patients with hypothesis is viewed purely as a pragmatic trial to aid clinical
mostly chronic symptomatic and definite radiographic OA decisions47, then the study would appear to show that the
knee, had a moderate effect on pain and physical function and small group format program is sufficiently effective to pro-
a small effect on health related quality of life. These results vide a cost-effective alternative to the usual individual treat-
are in broad agreement with randomized controlled trials of ments for knee OA.
acceptable validity and power11. There were, however, impor- Some interesting results emerged from the tertiary
tant differences with previous studies relating to the popula- hypotheses of this study. While the self-reported improve-
tion sampled. ments were substantiated by improvements in objective mea-
Most methodologically sound studies reporting on exercise sures of physical performance, there were only small absolute
for people with knee OA have used community volunteers or changes in the measures of physical performance despite no
patients with more recent and less severe symptomatic dis- evidence of a possible ceiling effect. Fast gait speed reached
ease12,15,18. In our sample, 44% reported symptom duration of by women in this sample was only 124 cm•s–1 (167 cm•s–1 for
greater than 5 years, 76% had bilateral symptomatic knee OA, age matched controls) and by men only 136 cm•s–1 (177
and 71% reported a minimum one comorbidity for which they cm•s–1 for age matched controls)42. The patients also demon-
were daily taking prescription medication. Not unexpectedly, strated muscle strength substantially below matched norma-
the study sample had SF-36 scores (Table 1) well below both tive data for both lower limbs48,49. In fact, baseline interquar-
stratified United States (65 years and over) and Australian tile range (IQR) for knee extensor strength in the current study
(65–74 years) population norms45,46. The Australian National was 34–46% for the weaker limb (46–75% for the stronger
Health Survey of 1995 found a PCS score of 42.8 and a MCS limb) of reported normative values. Similarly, but in contrast
score of 51.3 in persons aged 65–74 years (n = 1658). Using to a recent population based study23, the patients in this study
the derived Australian factor score coefficients, the current also showed markedly decreased knee flexor strength com-
sample of patients with knee OA gave a PCS score of 32.0 and pared to normative controls. Baseline IQR for knee flexor
a MCS score of 42.9, indicating that both physical and mental strength was only 32–60% for the weaker limb (41–73% for
HRQOL are affected in these older patients with knee OA the stronger limb)48,49. However, most of the patients in this
seeking treatment. study had moderate to severe radiological and symptomatic
The second hypothesis, that the group format program disease, suggesting that loss of knee flexor strength is a late

Personal non-commercial use only. The Journal of Rheumatology Copyright © 2001. All rights reserved.
Downloaded from www.jrheum.org on May 12, 2016 - Published by The Journal of
Fransen, et al: Exercise and knee OA 161
Rheumatology
disease related impairment associated with disuse atrophy. position provides a better indicator of cartilage thickness com-
The nonlinear relationship between muscle strength and phys- pared with the fully extended position59. Furthermore, large
ical function, or “why small changes in physiological capaci- cross sectional community studies have shown that the pres-
ty may produce relatively large effects on performance in frail ence of radiographic knee OA is significantly associated with
adults,” has been described in large population based samples the presence or absence of knee pain5,60. If symptoms are pre-
of older adults50,51. Two randomized controlled studies evalu- sent, however, our results suggest that radiographic disease
ating exercise for people with knee OA have also shown only severity does not have a linear association with symptom
small absolute and relative changes in isometric knee muscle severity (Table 5). Self-reported pain, physical function, and
strength compared with changes in measures of physical func- HRQOL were comparable between the groups stratified by
tion12,18. Our sample of referred patients had moderate to medial JSW. It may be that these results are confounded by
severe loss of medial JSW, 54% having a medial JSW < 2 differences between the stratified groups in patello-femoral
mm. A previous study found a strong correlation between joint involvement, or radiographic or symptomatic disease
reduced medial JSW and increased varus-valgus laxity at the severity of the contralateral knee but many studies have clear-
semiflexed knee joint52. The current finding of small absolute ly shown the significant influence of psychological distress
changes in isometric muscle strength would support the and social and behavioral variables on self-report mea-
hypothesis that “strengthening may have a smaller impact in sures7,61,62. It has been suggested that people with chronic dis-
lax knees”53. Further, about 73% of this sample were women ease simply adapt their expectations, lifestyle, and environ-
and it is claimed that “older women gain only about half as ment over time. Our study does, however, strongly suggest
much strength as older men under the same exercise proto- that medial JSW has important relevance for short term treat-
col”54. It seems clear that particularly patients with moder- ment responsiveness, substantiating the general recommenda-
ate–severe loss of medial JSW, relatively poor muscle tion that physical therapy is particularly indicated in relative-
strength, and unable to perform high intensity training due to ly early disease.
age and/or comorbidity may have potential to benefit This study deals with tertiary prevention or attempting to
from lengthier treatments than the 8 weeks assessed by our limit disability in established symptomatic disease. The inten-
study. sity of physical treatment possible in older people with
For most people, healthy aging is accompanied by a grad- marked chronic joint disease is often limited, suggesting
ual loss of muscle strength48,49, kinesthetic acuity55, and bio- lengthy treatment duration may be needed to reach an ade-
logical quality of the cartilage, resulting in decreased ability quate treatment dosage. Due to future health care resource
of the joint to safely absorb the repetitive impulse loading constraints in many countries, financial support for lengthy
associated with walking. Peak loading rate increases with treatments may only be feasible with a more cost-effective
increasing walking speed, accounting for the finding that gait strategy than provided by the current usual individual physi-
at a fast self-selected speed has higher discriminative validity cal therapy treatment mode. A more clinically effective strat-
than gait at a normal self-selected speed for people with lower egy may be secondary prevention or screening persons for
limb disability42. The knee extensors function to attenuate early disease and providing an easily accessible and effective
peak loading rate at heel strike56. Indeed, this study showed intervention. It is hypothesized that people with early disease
that changes in knee extensor strength were more highly asso- will be better able to tolerate an intensive program aimed at
ciated with reduced knee pain and improved physical function controlling damaging impulse loading of the knee joint com-
compared with changes in knee flexor strength. It is suggest- pared with patients with late disease. A longitudinal study is
ed that limiting appropriate neuromuscular compensatory needed to establish the effectiveness of this secondary pre-
responses by reducing nociceptive stimuli during weight- vention strategy.
bearing activities with regular analgesia is not an optimal This randomized controlled clinical study confirms the
strategy in early disease57. It is of concern, therefore, that effectiveness of physical therapy for patients with knee OA
roughly 50% of rheumatologists referred patients with knee seeking treatment in terms of self-reported pain, physical
OA for physical therapy “sometimes,” “rarely,” or “never”58. function, and HRQOL. Improvements revealed by self-report
This reported poor referral to physical therapy compared with questionnaires were significantly associated with improve-
the prescription of pharmacologic agents may be due to uncer- ments in objective measures of physical performance, and
tainty concerning the effectiveness provided by physical ther- treatment effectiveness was still apparent 2 months after for-
apy services or to economic constraints of either the health- mal treatment stopped. The sample size was insufficient to
care funder or the patient. We have tried to address both these show a statistically significant difference in clinical effective-
concerns. ness between individual treatments and a small group-format
This study provides initial evidence that radiographic dis- program. Treatment effectiveness was not modified by age,
ease severity will modify physical therapy treatment respon- sex, BMI, or symptom duration, but patients with a severe loss
siveness. Radiographic severity was measured by medial JSW of medial JSW were less responsive to this relatively short
with the knee in a semiflexed weight-bearing position, as this intervention.

Personal non-commercial use only. The Journal of Rheumatology Copyright © 2001. All rights reserved.
162
Downloaded from www.jrheum.org on May 12, 2016 - Published byTheThe Journal of
Journal of Rheumatology 2001; 28:1
Rheumatology
Appendix. Group Format Program. 8. Badley EM, Crotty M. An international comparison of the
estimated effect of the aging of the population on the major cause
Gymnasium: 8 weeks, attendance twice weekly for about 1 h. Group sizes of disablement, musculoskeletal disorders. J Rheumatol
were limited to 6 and were supervised by a physical therapist. The initial visit 1995;22:1934-40.
consisted of an education session outlining the benefits of exercise for people 9. Hochberg MC, Altman RD, Brandt KD, et al. Guidelines for the
with arthritis and the importance of appropriate footwear and weight control. medical management of osteoarthritis. Part II. Osteoarthritis of the
In the gymnasium, subjects were requested to perform all exercises bilateral- knee. Arthritis Rheum 1995;38:1541-6.
ly, start the session with stretches and then proceed with the remaining exer- 10. Puett DW, Griffin MR. Published trials of nonmedicinal and
cises in random order. Subjects were advised to adjust the weights used so noninvasive therapies for hip and knee osteoarthritis. Ann Intern
that the exercises were performed with some effort but with a minimum of Med 1994;121:133-40.
pain during the session. Subjects were to note any adverse reactions to exer- 11. van Baar ME, Assendelft WJ, Dekker J, Oostendorp RA, Bijlsma
cise and seek advice from the supervising physical therapist. JW. Effectiveness of exercise therapy in patients with osteoarthritis
of the hip or knee. Arthritis Rheum 1999;42:1361-9.
Exercise Repetition/(weight range) 12. van Baar ME, Dekker J, Oostendorp RA, et al. The effectiveness of
exercise therapy in patients with osteoarthritis of the hip or knee: a
Stretches: quadriceps, hamstrings, randomized clinical trial. J Rheumatol 1998;25:2432-9.
gastrocnemius 3 × 30 s hold each muscle group 13. Borjesson M, Robertson E, Weidenhielm L, Mattsson E, Olsson E.
Stationary bicycle 20 min/50–60% maximum heart Physiotherapy in knee osteoarthrosis: effect on pain and walking.
rate Physiotherapy Res Inter 1996;1:89-97.
Non-weight-bearing quadriceps muscle 20–40/(0–6 lbs) 14. Callaghan MJ, Oldham JA, Hunt J. An evaluation of exercise
strengthening: inner range with weight regimes for patients with osteoarthritis of the knee: a single blind
attached to ankle randomized controlled trial. Clin Rehabil 1995;9:213-8.
Weight-bearing quadriceps muscle 100 steps/(0 setting) 15. Ettinger WH, Burns R, Messier SP, et al. A randomized trial
strengthening: Tunturi 401 Variable comparing aerobic exercise and resistance exercise with a health
Resistance Climber education program in older adults with knee osteoarthritis. The
Non-weight-bearing concentric/eccentric 20–40/(10–30 lbs) each muscle Fitness Arthritis and Seniors Trial (FAST). JAMA 1997;277:25-31.
quadriceps and knee flexors: full range group 16. Minor MA, Hewett JE, Webel RR, Anderson SK, Kay DR. Efficacy
with Isolator bench (Chattanooga Corp.) of physical conditioning exercise in patients with rheumatoid
Weight-bearing eccentric quadriceps: 20–40 arthritis and osteoarthritis. Arthritis Rheum 1989;32:1396-405.
Controlled stepdown from 10–15 cm step. 17. Rogind H, Bibow-Nielsen B, Jensen B, Moller HC, Frimodt-Moller
Patella taping applied by physical H, Bliddal H. The effects of a physical training program on patients
therapist if required to reduce pain. with osteoarthritis of the knees. Arch Phys Med Rehabil
1998;79:1421-7.
Home program: 3 days per week: Stretches as per group exercise sessions fol- 18. O’Reilly SC, Muir KR, Doherty M. Effectiveness of home exercise
lowed by 20 min of continuous outdoor walking or indoor stationary bicycle. on pain and disability from osteoarthritis of the knee: a randomised
controlled trial. Ann Rheum Dis 1999;58:15-9.
19. Hurley MV, Scott DL. Improvements in quadriceps sensorimotor
function and disability of patients with knee osteoarthritis following
a clinically practicable exercise regime. Br J Rheumatol
1998;37:1181-7.
REFERENCES 20. Dexter P, Brandt K. Distribution and predictors of depressive
1. Felson DT, Zhang Y. An updata on the epidemiology of knee and symptoms in osteoarthritis. J Rheumatol 1994;21:279-86.
hip osteoarthritis with a view to prevention. Arthritis Rheum 21. Macfarlane GJ, Morris S, Hunt IM, et al. Chronic widespread pain
1998;41:1343-55. in the community: the influence of psychological symptoms and
2. Felson DT, Naimark A, Anderson J, Kazis L, Castelli W, Meenan mental disorder on healthcare seeking behavior. J Rheumatol
RF. The prevalence of knee osteoarthritis in the elderly. The 1999;26:413-9.
Framingham Osteoarthritis Study. Arthritis Rheum 1987;30:914-8. 22. O’Reilly SC, Jones A, Muir KR, Doherty M. Quadriceps weakness
3. Hart DJ, Doyle DV, Spector TD. Incidence and risk factors for in knee osteoarthritis: the effect on pain and disability. Ann Rheum
radiographic knee osteoarthritis in middle-aged women. The Dis 1998;57:588-94.
Chingford Study. Arthritis Rheum 1999;42:17-24. 23. Slemenda C, Brandt KD, Heilman DK, et al. Quadriceps weakness
4. Guccione AA, Felson DT, Anderson JJ, et al. The effects of specific and osteoarthritis of the knee. Ann Intern Med 1997;127:97-104.
medical conditions on the functional limitations of elders in the 24. Fisher NM, Pendergast DR. Reduced muscle function in patients
Framingham Study. Am J Public Health 1994;84:351-8. with osteoarthritis. Scand J Rehabil Med 1997;29:213-21.
5. Hochberg MC, Lawrence RC, Everett DF, Cornoni-Huntley J. 25. Andriacchi TP, Ogle JA, Galante JO. Walking speed as a basis for
Epidemiologic associations of pain in osteoarthritis of the knee: normal and abnormal gait measurements. J Biomech 1977;
data from the National Health and Nutrition Examination Survey 10:261-8.
and the National Health and Nutrition Examination-I 26. Murray MP, Gore DR, Sepic SB, Mollinger LA. Antalgic
Epidemiologic Followup Survey. Semin Arthritis Rheum maneuvers during walking in men with unilateral knee disability.
1989;18:4-9. Clin Orthop 1985;199:192-200.
6. Boult C, Kane RL, Louis TA, Boult L, McCaffrey D. Chronic 27. Stauffer RN, Chao EY, Gyory AN. Biomechanical gait analysis of
conditions that lead to functional limitation in the elderly. the diseased knee joint. Clin Orthop 1977;126:246-55.
J Gerontol 1994;49:M28-36. 28. Minor MA, Hewett JE, Webel RR, Dreisinger TE, Kay DR.
7. O’Reilly SC, Muir KR, Doherty M. Knee pain and disability in the Exercise tolerance and disease related measures in patients with
Nottingham community: association with poor health status and rheumatoid arthritis and osteoarthritis. J Rheumatol 1988;
psychological distress. Br J Rheumatol 1998;37:870-3. 15:905-11.

Personal non-commercial use only. The Journal of Rheumatology Copyright © 2001. All rights reserved.
Downloaded from www.jrheum.org on May 12, 2016 - Published by The Journal of
Fransen, et al: Exercise and knee OA 163
Rheumatology
29. Philbin EF, Groff GD, Ries MD, Miller TE. Cardiovascular fitness 45. Ware JE, Kosinski M, Keller SD. SF-36 Physical and Mental
and health in patients with end-stage osteoarthritis. Arthritis Rheum Health Summary Scales: A User’s Manual. Boston: The Health
1995;38:799-805. Institute, New England Medical Center; 1994.
30. Fransen M, Margiotta E, Crosbie J, Edmonds J. A revised group 46. Australian Bureau of Statistics. National Health Survey. SF-36
exercise program for osteoarthritis of the knee. Physiotherapy Res population norms. Belconnen: Australian Bureau of Statistics;
Int 1997;2:30-41. 1995.
31. Altman R, Asch E, Bloch D, et al. Development of criteria for the 47. Schwartz D, Lellouch J. Explanatory and pragmatic attitudes in
classification and reporting of osteoarthritis. Classification of therapeutic trials. J Chronic Dis 1967;20:637-48.
osteoarthritis of the knee. Arthritis Rheum 1986;29:1039-49. 48. Andrews AW, Thomas MW, Bohannon RW. Normative values for
32. Buckland-Wright J, Macfarlane D, Williams S, Ward R. Accuracy isometric muscle force measurements obtained with hand-held
and precision of joint space width measurements in standard and dynamometers. Phys Ther 1996;76:248-59.
macroradiographs of osteoarthritic knees. Ann Rheum Dis 49. Skelton DA, Grieg CA, Davies JM, Young A. Strength, power and
1995;54:872-80. related functional ability of healthy people aged 65-89 years. Age
33. Buckland-Wright C. Protocols for precise radio-anatomical Ageing 1994;23:371-7.
positioning of the tibiofemoral and patellofemoral compartments of 50. Buchner DM, Larson EB, Wagner EH, Koepsell TD, de Lateur BJ.
the knee. Osteoarthritis Cart 1995;3:71-80. Evidence for a non-linear relationship between leg strength and gait
34. Mazzuca S. Plain radiography in the evaluation of knee speed. Age Ageing 1996;25:386-91.
osteoarthritis. Curr Opin Rheumatol 1997;9:263-7. 51. Ferrucci L, Guralnik JM, Buchner D, et al. Departures from
35. Bellamy N, Buchanan WW, Goldsmith CH, Campbell J, Stitt LW. linearity in the relationship between measures of muscular strength
Validation study of WOMAC: A health status instrument for mea- and physical performance of the lower extremities: the Women’s
suring clinically important patient relevant outcomes to Health and Aging Study. J Gerontol 1997;52A:M275-85.
antirheumatic drug therapy in patients with osteoarthritis of the hip 52. Sharma L, Lou C, Felson DT, et al. Laxity in healthy and
or knee. J Rheumatol 1988;15:1833-40. osteoarthritis knees. Arthritis Rheum 1999;42:861-70.
36. Ware J, Snow K, Kosinski M, et al. SF-36 Health Survey: manual 53. Sharma L, Hayes KW, Felson DT, et al. Does laxity alter the
and interpretation guide. Boston: The Health Institute, New relationship between strength and physical function in knee
England Medical Center; 1993. osteoarthritis? Arthritis Rheum 1999;42:25-32.
37. Ware J, Sherbourne D. The MOS 36-item Short-Form Health 54. Buchner DM. Understanding variability in studies of strength
Survey (SF-36): I. Conceptual framework and item selection. Med training in older adults: a meta-analytic perspective. Top Geriatr
Care 1992;30:473-83. Rehabil 1993;8:1-21.
38. Kosinski M, Keller SD, Ware JE, Hatoum HT, Kong SX. The SF- 55. Pai Y-C, Rymer WZ, Chang RW, Sharma L. Effect of age and
36 Health Survey as a generic outcome measure in clinical trials of osteoarthritis on knee proprioception. Arthritis Rheum
patients with osteoarthritis and rheumatoid arthritis: relative 1997;40:2260-5.
validity of scales in relation to clinical measures of arthritis 56. Jefferson RJ, Collins JJ, Whittle MW, Radin EL, O’Connor JJ. The
severity. Med Care 1999;37:MS23-39. role of the quadriceps in controlling impulsive forces around the
39. Kosinski M, Keller SD, Hatoum HT, Kong SX, Ware JE. The SF- heel. J Engineer Med 1990;204:21-8.
36 Health Survey as a generic outcome measure in clinical trials of 57. Schnitzer TJ, Popovich JM, Andersson GB, Andriacchi TP. Effect
patients with osteoarthritis and rheumatoid arthritis: tests of data of piroxicam on gait in patients with osteoarthritis of the knee.
quality, scaling assumptions and score reliability. Med Care Arthritis Rheum 1993;36:1207-13.
1999;37:MS10-22. 58. Hochberg MC, Perlmutter DL, Hudson JI, Altman R. Preferences in
40. Ware JE, Kosinski M, Bayliss MS, McHorney CA, Rogers WH, the management of osteoarthritis of the hip and knee: results of a
Raczek A. Comparison of methods for the scoring and statistical survey of community-based rheumatologists in the United States.
analysis of SF-36 health profile and summary measures: summary Arthritis Care Res 1996;9:170-6.
of results from the Medical Outcomes Study. Med Care 59. Messieh SS, Fowler PJ, Munro T. Anteroposterior radiographs of
1995;33:AS264-79. the osteoarthritis knee. J Bone Joint Surg (Br) 1990;72-B:639-40.
41. Jenkinson C, Layte R, Lawrence K. Development and testing of the 60. Lethbridge-Cejku M, Scott WW, Reichle R, et al. Association of
Medical Outcomes Study 36-Item Short Form Health Survey radiographic features of osteoarthritis of the knee with knee pain:
summary scale scores in the United Kingdom. Med Care data from the Baltimore Longitudinal Study of Aging. Arthritis
1997;35:410-6. Care Res 1995;8:182-8.
42. Fransen M, Heussler J, Margiotta E, Edmonds J. Quantitative gait 61. Dekker J, Boot B, Van der Woude L, Bijlsma J. Pain and disability
analysis — comparison of rheumatoid arthritic and non-arthritic in osteoarthritis: a review of biobehavioural mechanisms. J Behav
subjects. Australian J Physiotherapy 1994;40:191-9. Med 1992;15:189-212.
43. Fransen M, Crosbie J, Edmonds J. Reliability of gait measurements 62. Turk DC, Okifuji A. Assessment of patients’ reporting of pain: an
in people with osteoarthritis of the knee. Phys Ther 1997; integrated perspective. Lancet 1999;353:1784-8.
77:944-53.
44. Bellamy N, Carette S, Ford PM, et al. Osteoarthritis antirheumatic
drug trials. II. Tables for calculating sample size for clinical trials.
J Rheumatol 1992;19:444-50.

Personal non-commercial use only. The Journal of Rheumatology Copyright © 2001. All rights reserved.
164
Downloaded from www.jrheum.org on May 12, 2016 - Published byTheThe Journal of
Journal of Rheumatology 2001; 28:1
Rheumatology