Scolaris Content Display Scolaris Content Display

Cochrane Database of Systematic Reviews Review - Intervention

Self‐management education programmes for osteoarthritis

Authors' declarations of interest

Version published: 15 January 2014 Version history

https://doi.org/10.1002/14651858.CD008963.pub2
Collapse all Expand all

Abstract

available in

Background

Self‐management education programmes are complex interventions specifically targeted at patient education and behaviour modification. They are designed to encourage people with chronic disease to take an active self‐management role to supplement medical care and improve outcomes.

Objectives

To assess the effectiveness of self‐management education programmes for people with osteoarthritis.

Search methods

The Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, EMBASE, PyscINFO, SCOPUS and the World Health Organization (WHO) International Clinical Trial Registry Platform were searched, without language restriction, on 17 January 2013. We checked references of reviews and included trials to identify additional studies.

Selection criteria

Randomised controlled trials of self‐management education programmes in people with osteoarthritis were included. Studies with participants receiving passive recipients of care and studies comparing one type of programme versus another were excluded.

Data collection and analysis

In addition to standard methods we extracted components of the self‐management interventions using the eight domains of the Health Education Impact Questionnaire (heiQ), and contextual and participant characteristics using PROGRESS‐Plus and the Health Literacy Questionnaire (HLQ). Outcomes included self‐management of osteoarthritis, participant's positive and active engagement in life, pain, global symptom score, self‐reported function, quality of life and withdrawals (including dropouts and those lost to follow‐up). We assessed the quality of the body of evidence for these outcomes using the GRADE approach.

Main results

We included twenty‐nine studies (6,753 participants) that compared self‐management education programmes to attention control (five studies), usual care (17 studies), information alone (four studies) or another intervention (seven studies). Although heterogeneous, most interventions included elements of skill and technique acquisition (94%), health‐directed activity (85%) and self‐monitoring and insight (79%); social integration and support were addressed in only 12%. Most studies did not provide enough information to assess all PROGRESS‐Plus items. Eight studies included predominantly Caucasian, educated female participants, and only four provided any information on participants' health literacy. All studies were at high risk of performance and detection bias for self‐reported outcomes; 20 studies were at high risk of selection bias, 16 were at high risk of attrition bias, two were at high risk of reporting bias and 12 were at risk of other biases. We deemed attention control as the most appropriate and thus the main comparator.

Compared with attention control, self‐management programmes may not result in significant benefits at 12 months. Low‐quality evidence from one study (344 people) indicates that self‐management skills were similar in active and control groups: 5.8 points on a 10‐point self‐efficacy scale in the control group, and the mean difference (MD) between groups was 0.4 points (95% confidence interval (CI) ‐0.39 to 1.19). Low‐quality evidence from four studies (575 people) indicates that self‐management programmes may lead to a small but clinically unimportant reduction in pain: the standardised mean difference (SMD) between groups was ‐0.26 (95% CI ‐0.44 to ‐0.09); pain was 6 points on a 0 to 10 visual analogue scale (VAS) in the control group, treatment resulted in a mean reduction of 0.8 points (95% CI ‐0.14 to ‐0.3) on a 10‐point scale, with number needed to treat for an additional beneficial outcome (NNTB) of 8 (95% CI 5 to 23). Low‐quality evidence from one study (251 people) indicates that the mean global osteoarthritis score was 4.2 on a 0 to 10‐point symptom scale (lower better) in the control group, and treatment reduced symptoms by a mean of 0.14 points (95% CI ‐0.54 to 0.26). This result does not exclude the possibility of a clinically important benefit in some people (0.5 point reduction included in 95% CI). Low‐quality evidence from three studies (574 people) showed no signficant difference in function between groups (SMD ‐0.19, 95% CI ‐0.5 to 0.11); mean function was 1.29 points on a 0 to 3‐point scale in the control group, and treatment resulted in a mean improvement of 0.04 points with self‐management (95% CI ‐0.10 to 0.02). Low‐quality evidence from one study (165 people) showed no between‐group difference in quality of life (MD ‐0.01, 95% CI ‐0.03 to 0.01) from a control group mean of 0.57 units on 0 to 1 well‐being scale. Moderate‐quality evidence from five studies (937 people) shows similar withdrawal rates between self‐management (13%) and control groups (12%): RR 1.11 (95% CI 0.78 to 1.57). Positive and active engagement in life was not measured.

Compared with usual care, moderate‐quality evidence from 11 studies (up to 1,706 participants) indicates that self‐management programmes probably provide small benefits up to 21 months, in terms of self‐management skills, pain, osteoarthritis symptoms and function, although these are of doubtful clinical importance, and no improvement in positive and active engagement in life or quality of life. Withdrawal rates were similar. Low to moderate quality evidence indicates no important differences in self‐management , pain, symptoms, function, quality of life or withdrawal rates between self‐management programmes and information alone or other interventions (exercise, physiotherapy, social support or acupuncture).

Authors' conclusions

Low to moderate quality evidence indicates that self‐management education programmes result in no or small benefits in people with osteoarthritis but are unlikely to cause harm.

Compared with attention control, these programmes probably do not improve self‐management skills, pain, osteoarthritis symptoms, function or quality of life, and have unknown effects on positive and active engagement in life. Compared with usual care, they may slightly improve self‐management skills, pain, function and symptoms, although these benefits are of unlikely clinical importance.

Further studies investigating the effects of self‐management education programmes, as delivered in the trials in this review, are unlikely to change our conclusions substantially, as confounding from biases across studies would have likely favoured self‐management. However, trials assessing other models of self‐management education programme delivery may be warranted. These should adequately describe the intervention they deliver and consider the expanded PROGRESS‐Plus framework and health literacy, to explore issues of health equity for recipients.

PICOs

Population
Intervention
Comparison
Outcome

The PICO model is widely used and taught in evidence-based health care as a strategy for formulating questions and search strategies and for characterizing clinical studies or meta-analyses. PICO stands for four different potential components of a clinical question: Patient, Population or Problem; Intervention; Comparison; Outcome.

See more on using PICO in the Cochrane Handbook.

Plain language summary

available in

Self‐management education programmes for osteoarthritis

This review shows that in people with osteoarthritis:

Self‐management education programmes may not improve self‐management skills, osteoarthritis symptoms, function, quality of life and dropout rates but may reduce pain modestly compared with attention control. Active and positive engagement in life was not reported.

Self‐management education programmes may slightly improve self‐management skills, pain and function but may not improve active and positive engagement in life, osteoarthritis symptoms, quality of life and dropout rates compared with usual care.

Self‐management education programmes probably do not improve outcomes compared with provision of information alone or compared with other interventions (exercise, physiotherapy, social support or acupuncture).

What is osteoarthritis and what are self‐management education programmes?

Osteoarthritis (OA) is a disease of the joints, such as your knee or hip, or the joints in your hands. The joint cartilage that lines the joint gradually thins, narrowing the joint space. In severe cases, no cartilage remains between the bones, and the bones rub together when the joint is moved, making the joint painful and sometimes unstable.

Self‐management education programmes are behavioural interventions designed to encourage people with chronic disease to take an active role in the management of their own condition. These programmes aim to improve outcomes for patients by supporting, not replacing, medical care. The content used to educate patients about their condition and to explain how they can best manage their symptoms varies between programmes.

Best estimate of what happens to people with osteoarthritis who undergo self‐management programmes:

People who completed a self‐management programme rated their self‐management skills to be 0.4 points better (0.4 points worse to 1.2 points better) on a scale of 1 to 10 (higher score means better self‐management) after 12 months (4% absolute improvement; 4% worse to 12% better).

‐ People who completed a self‐management programme rated their self‐management skills as 6.2 points on a scale of 1 to 10.

‐ People who received attention control rated their self‐management skills as 5.8 points on a scale of 1 to 10.

People who completed a self‐management programme rated their pain to be 0.8 points lower (0.3 to 0.14 points lower) on a scale of 0 to 10 (lower score means less pain) after 12 months (8% absolute improvement).

‐ People who completed a self‐management programme rated their pain as 5 points on a scale of 0 to 10.

‐ People who received attention control rated their pain as 5.8 points on a scale of 0 to 10.

People who completed a self‐management programme rated their osteoarthritis symptoms to be 0.14 points lower (0.54 points lower to 0.26 points higher) on a scale of 0 to 10 (lower score means fewer symptoms) after 12 months (1% absolute improvement).

‐ People who completed a self‐management programme rated their symptoms as 4.1 points on a scale of 0 to 10.

‐ People who received attention control rated their symptoms as 4.2 points on a scale of 0 to 10.

People who completed a self‐management programme rated their function to be 0.04 points lower (0.02 points lower to 0.10 points higher) on a scale of 0 to 3 (lower score means better function) after 12 months (4% absolute improvement).

‐ People who completed a self‐management programme rated their function as 1.25 points on a scale of 0 to 3.

‐ People who received attention control rated their function as 1.29 points on a scale of 0 to 3.

People who completed a self‐management programme rated their quality of life to be 0.01 points lower (0.03 points lower to 0.01 points higher) on a scale of 0 to 1 (higher score means better quality of life) after 12 months (1% absolute worsening).

‐ People who completed a self‐management programme rated their quality of life as 0.56 points on a scale of 0 to 1.

‐ People who received attention control rated their quality of life as 0.57 points on a scale of 0 to 1.

One more person out of 100 dropped out of self‐management programmes (1% absolute improvement).

‐ 13 out of 100 people who received a self‐management programme dropped out.

‐ 12 out of 100 people who received attention control dropped out.

Authors' conclusions

Implications for practice

We found low to moderate evidence suggesting that self‐management education programmes, as delivered in the studies included in this review, result in no or small benefits in patients with OA. Compared with attention control, the comparator with the least risk of bias, as it controls for any effect of contact time with programme providers, self‐management education programmes probably do not improve self‐management skills, pain, osteoarthritis symptoms, function or quality of life in people with OA, and their effects on positive and active engagement in life are unknown. Compared with usual care, self‐management programmes do not result in more positive and active engagement in life but may improve self‐management skills, pain, osteoarthritis symptoms and function. However, all apparent benefits were small and unlikely to be of clinical importance. Compared with provision of information alone or alternative interventions (e.g. exercise, physiotherapy, social support, acupuncture), self‐management education programmes do not improve self‐management of OA, positive and active engagement in life, pain, global OA scores, function or quality of life. We found no evidence that self‐management programmes cause harm.

Implications for research

Although we downgraded the evidence to moderate or low for most outcomes, we believe that further studies investigating the effects of self‐management education programmes, as delivered in the trials in this review, are not likely to substantially change the conclusions of this review. Confounding from biases across studies would have likely favoured self‐management; thus it is unlikely that correcting for these biases would overturn the direction of the results and may, in fact, drive some of the small but clinically unimportant improvements seen with self‐management over usual care, towards the null (i.e. no significant between‐group differences).

However, it is possible that other models of self‐management education programmes that differ in mode of delivery, type of audience, duration and frequency of sessions, personnel used to teach self‐management skills, etc., may enhance self‐management of OA. Further trials of different self‐management education programmes, particularly those that are tailored to the individual, may therefore be warranted.

We found a mismatch between the aims of self‐management education programmes and the outcomes used to measure success. Studies often sought to measure pain as a main outcome. Although this is certainly a desirable outcome, OA is a chronic condition, and conservative interventions, including self‐management education programmes, are not intended to 'cure' pain but rather to enable people to have a reasonable life despite their ongoing pain. The stated aim of self‐management education programmes for people with OA is to educate people about their condition and teach them how best to manage their symptoms. It seems reasonable therefore that indicators of knowledge, self‐management skills and self‐efficacy should be included as key outcomes in studies assessing the effects of a self‐management education programme in OA. We suggest that a measure of fatigue should be considered as an outcome in the next updated version of this review. Most studies (27/29 studies) did not include fatigue as an outcome, although it has been reported to be a major problem in patients suffering from chronic diseases such as OA (Snijders 2011).

Comprehensive evaluation of self‐management education programmes using robust and well‐validated tools will improve this field. The heiQ was developed from patients’ and clinicians’ perspectives on what are valued outcomes of self‐management programmes (Osborne 2007), and it has been a useful evaluation tool in many settings (e.g. Cadhilac 2011; Francis 2009; Greenhalgh 2009; Osborne 2011; Packer 2012; Wanitkun 2011). Although self‐efficacy scales are often applied, the interpretation, validity and value of the derived scores have been debated; these require further rigorous testing in relevant clinical settings in well‐defined populations (Brady 1997; Brady 2011).

Further trials should adequately describe the self‐management education programme that they deliver to enable better assessment of the therapeutic quality of the programme and to enable an assessment of the likely impact of intended outcomes. This will also allow between‐study comparisons. The methods we employed to describe the self‐management education programmes included in this review may serve as a good starting point. We suggest that trials include a more detailed description of study participants, taking into account the expanded PROGRESS‐Plus framework and health literacy, as well as other potential predictors and moderators of treatment outcome. We detected differences between subgroups in several outcomes, suggesting that it might be important for future studies to explore issues of health equity.

Summary of findings

Open in table viewer
Summary of findings for the main comparison. SMP compared to Attention control for osteoarthritis

SMP compared with attention control for osteoarthritis

Patient or population: patients with osteoarthritis
Settings: primary care, or outpatient
Intervention: SMP
Comparison: attention control

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No of participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Attention control

SMP

Self‐management of OA
Arthritis self‐efficacy scale (ASES). Scale from 1 to 10, higher better
Follow‐up: 12 months

Mean self‐management of osteoarthritis in the control groups was
5.8 points

Mean self‐management of osteoarthritis in the intervention groups was
0.4 points higher
(0.4 lower to 1.2 higher)

344
(one study)

⊕⊕⊝⊝
low1,2

MD 0.4 (‐0.39 to 1.19)

Absolute mean improvement 4% (4% worse to 12% improved) Relative improvement 7% (7% worse to 21% improved)3

Positive and active engagement in life —not measured

See comment

See comment

Not estimable

‐

See comment

No studies measured this outcome

Pain
Multiple tools4. Scale from: 0 to 10, 0 = no pain
Follow‐up: six to 12 months

Mean pain ranged across control groups from
5.67 to 6.19 points

Mean pain in the intervention groups was
0.8 points lower
(0.3 to 0.14 lower)

575
(three studies)

⊕⊕⊝⊝
low1,2

SMD ‐0.26 (‐0.44 to ‐0.09)

Absolute reduction in pain 8% (3% to 14% reduction). Relative reduction in pain 13% (5% to 22% reduction). NNTB = 8 (5 to 23)5

Global OA scores
AIMS2 (average of physical, affect, and pain subscales). Scale from 0 to 10, lower better
Follow‐up: nine months

Mean global osteoarthritis symptom score in the control group was
4.22 points

Mean global osteoarthritis symptom score in the intervention group was
0.14 points lower
(0.54 lower to 0.26 higher)

251
(one study)

⊕⊕⊝⊝
low2,6

Absolute reduction 1.4% (5.4% reduction to 2.6% increase). Relative reduction 3% (11% reduction to 5% increase)

Self‐reported function
Multiple tools.7 Lower score better
Follow‐up: 12 months

Mean self‐reported function in the control groups was
1.29 points on 0 to 3 scale8

Mean self‐reported function in the intervention groups was
0.04 points lower
(0.02 lower to 0.10 higher)

574
(three studies)

⊕⊕⊝⊝
low1,2

SMD ‐0.19 (‐0.5 to 0.11)

Absolute improvement in function 4% (2% reduction to 11% improvement). Relative improvement 11% (6% reduction to 30% improvement)8

Quality of life
Quality of well‐being scale. Scale from 0 to 1, higher better
Follow‐up: 12 months

Mean quality of life in the control groups was
0.57 units

Mean quality of life in the intervention groups was
0.01 lower
(0.03 lower to 0.01 higher)

165
(one study)

⊕⊕⊝⊝
low2,9

Absolute mean reduction 1% (95% CI 3% lower to 1% higher) Relative reduction 2% (95% CI 5% lower to 1% higher)

Withdrawals
Follow‐up: six to 12 months

117 per 1,000

130 per 1,000
(91 to 183)

RR 1.11
(0.78 to 1.57)

937
(five studies)

⊕⊕⊕⊝
moderate2

Absolute risk difference 1% increase (95% CI 3% decrease to 5% increase). Relative percentage change 11% increase (95% CI 22% decrease to 57% increase)

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: Confidence interval; RR: Risk ratio.

GRADE Working Group grades of evidence.
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1One large study was conducted in Veteran population that was mostly men (92.7%), limiting applicability of findings.

2Design flaws, including participants were not blind to group allocation in all trials, other trials had unclear randomisation method or concealment of allocation and unbalanced withdrawals across treatment groups, render the evidence susceptible to bias.

3Estimated relative changes based on mean (SD) ASES score in attention control group at baseline 5.8 (2.0) from Allen 2010.

4Pain VAS, pain on walking VAS and pain subscale of the arthritis impact measurement scale (AIMS).
5Estimated using mean (SD) for control group VAS pain on walking at baseline 6.28 (3.18) from Mazzuca 1997, and an assumed minimal clinically important difference of 1.5 points in 10‐point pain scale.
6Approximately half of total study population in trial had rheumatoid arthritis (data not included); data are presented for the OA subgroup; small sample size and wide CIs reduce precision.
7AIMS physical disability subscale, AIMS2 function subscale and HAQ disability subscale.
8Assumed risk from Mazzuca 1997 control group at 12 months mean HAQ disability scale:1.29 (SD 0.70); 0 to 3 scale, lower score better. Absolute risk difference estimated from control group SD at baseline from the same study (SD 0.66); and relative percent change using mean control group HAQ score at baseline (1.13).
9Potential imprecision due to data available only from a single study (n = 165).

Open in table viewer
Summary of findings 2. SMP compared with usual care for osteoarthritis

SMP compared with usual care

Patient or population: patients with osteoarthritis
Settings: community, outpatient, primary care
Intervention: SMP
Comparison: usual care or no treatment or wait list control

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No of participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Usual care/No treatment/Wait list

SMP

Self‐management of OA
Multiple tools1. Scale from 1 to 10, higher better
Follow‐up: three to 21 months

Mean self‐management of osteoarthritis in the control groups, using ASES 1 to 10‐point scale (10 is better), was
3.7 points2

Mean self‐management of osteoarthritis in the intervention groups was
0.13 points higher
(0.02 to 0.23 higher)

1,706
(11 studies)

⊕⊕⊕⊝
moderate3

SMD 0.16 (0.03 to 0.29)

Absolute mean improvement 1.3% (0.2% to 2.3% improvement). Relative improvement 3.5% (0.65% to 6.3% improvement). NNTB 13 (7 to 69)2

Positive and active engagement in life
Multiple tools4
Follow‐up: six to 12 months

Mean positive and active engagement in life in the control groups, based on SF‐36 subscale for role emotional, 0 to 100 scale (100 best), was
57 points5

Mean positive and active engagement in life in the intervention groups was

0.4 points higher
(8 lower to 8.4 higher)

357
(three studies)

⊕⊕⊕⊝
moderate3

SMD 0.01 (‐0.2 to 0.21)

Absolute mean improvement 0.4% (8% worsening to 8.8% improvement). Relative improvement 0.5% (10% worsening to 10% improvement)

Pain
Multiple tools6
Follow‐up: three to 21 months

Mean pain in the control groups, based on 0 to 10 VAS scale (0 is no pain), was
3.5 points7

Mean pain in the intervention groups was
0.5 points lower
(0.23 to 0.1 lower)

2,083
(14 studies)

⊕⊕⊕⊝
moderate3

SMD ‐0.19 (‐0.28 to ‐0.10)

Absolute mean reduction 5.5% (8% to 3% reduction). Relative reduction 16% (23% to 8% reduction). NNTB 11 (7 to 21)

Global OA scores
Multiple tools8
Follow‐up: six to 21 months

Mean global osteoarthritis scores in the control groups, based on 0 to 96 point WOMAC scale (lower is better), was
35 points9

Mean global osteoarthritis score in the intervention groups was
5.0 points lower
(7.6 to 2.7 lower)9

1,957
(seven studies)

⊕⊕⊕⊝
moderate3

SMD ‐0.25 (‐0.37 to ‐0.13)

Absolute mean improvement 5% (3% to 8% improvement). Relative improvement 13% (7% to 19% improvement). NNTB 10 (7 to 19)9

Function—Self‐reported
Scale from 0 to 68. Lower score is better
Follow‐up: six to 21 months

Mean function self‐reported in the control groups, based on 0 to 68 WOMAC subscale (lower is better), was
25 points10

Mean function self‐reported in the intervention groups was
2.6 points lower
(3.9 to 1.3 lower)10

2,254
(13 studies)

⊕⊕⊕⊝
moderate3

SMD ‐0.18 (‐0.27 to ‐0.09)

Absolute improvement 4% (2% to 6% improvement). Relative improvement 10% (5% to 15% improvement). NNTB 14 (9 to 27)10

Quality of life
Multiple tools11
Follow‐up: six to 21 months

Mean quality of life in the control groups, based on ‐0.11 to 1.0 EQ‐5D scale (higher score is better), was
0.66 points12

Mean quality of life in the intervention groups was
0.006 points higher
(0.03 lower to 0.04 higher)

1,383
(eight studies)

⊕⊕⊕⊝
moderate3

SMD 0.02 (‐0.09 to 0.13)

Absolute improvement 0.6% (2.7% worsening to 3.9% improvement). Relative improvement 1% (4.5% worsening to 6.5% improvement)

Withdrawals
Losses to follow‐up
Follow‐up: three to 21 months

172 per 1,000

171 per 1,000
(128 to 229)

RR 0.99
(0.74 to 1.33)

3,738
(16 studies)

⊕⊕⊝⊝
low3,13

Absolute risk difference 0% (3% lower to 4% higher). Relative difference 10% lower (26% lower to 33% higher)

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: Confidence interval; RR: Risk ratio.

GRADE Working Group grades of evidence.
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1Arthritis self‐efficacy scale (ASES), ASES subscale for pain, arthritis helplessness index (AHI) and the health education impact questionnaire (heiQ).
2Self‐management measured using ASES 1 to 10 scale; 10 is best score, taken from Heuts 2005: mean (SD) baseline ASES score in control group was 3.7 (0.8), and mean (SD) ASES final score in control group was 3.7 (0.9); for number needed to treat for an additional beneficial outcome (NNTB) calculation, the minimal clinically important difference not known, assumed as 0.5.
3Participants and study personnel were not blind to group allocations; other issues included unclear randomisation and concealment of allocation; thus trials were at risk of selection, performance and detection biases. One trial, Victor 2005, had inconsistent results compared with the other 10 trials, possibly related to a high risk of bias in that trial, but we do not believe it was significant enough to downgrade the evidence further.
4Quality of life short form 36 (SF‐36) subscale for role emotional and heiQ subscale for positive and active engagement in life.

5Positive and active engagement in life calculated from Victor 2005, using SF‐36, 0 to 100 scale (100 is highest score): mean (SD) baseline score in control group was x (y); and mean final score in control group was 57 points.
6Arthritis impact measurement scale (AIMS), visual analogue scale (VAS) and Western Ontario McMaster Universities Arthritis Index (WOMAC).

7Pain calculated from Heuts 2005, using VAS 0 to 10 scale (0 is no pain): Mean (SD) baseline hip pain score in control group was 3.5 (2.9); and mean final score in control group was 3.5 (2.7).
8Western Ontario McMasters University Arthritis Index (WOMAC), Arthritis Impact Measurement Scale (AIMS2) and self‐rated global health questionnaires.
9Global disease scores taken from Hurley 2007, using WOMAC 0 to 96 point scale (lower score better): Mean (SD) baseline score in the control group was 38.4 (19.82); and mean final score in control group was 35 points. For number needed to treat for an additional beneficial outcome (NNTB) calculation, the minimal clinically important difference not known, assumed as 0.5.

10Self‐reported function based on Hurley 2007, using WOMAC function 0 to 68 point scale (lower score better): Mean (SD) baseline score in the control group was 27.2 (14.6); and mean final score in the control group was 25 points. For number needed to treat for an additional beneficial outcome (NNTB) calculation, the minimal clinically important difference not known, assumed as 0.5.
11Quality of life short form 36 (SF‐36), quality of life, quality of well‐being scale and EQ‐5D.

12Quality of life taken based on Hurley EQ‐5D (0 to 1 scale; higher score better): Control group mean (SD) at baseline was 0.6 (0.3) and at follow‐up was 0.66 (0.3).
13Inconsistency across studies regarding whether greater number of withdrawals in the self‐management group or control group.

Open in table viewer
Summary of findings 3. SMP compared with information only for osteoarthritis

SMP compared with information only for osteoarthritis

Patient or population: patients with osteoarthritis
Settings: community, outpatient, primary care
Intervention: SMP
Comparison: information only

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No of participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Information only

SMP

Self‐management of OA
Multiple tools1. Scale from 5 to 35 points (higher better)
Follow‐up: 12 months

Mean self‐management of osteoarthritis in the control groups, based on 5 to 35 point ASES self‐efficacy pain scale, was
19.2 points2

Mean self‐management of osteoarthritis in the intervention groups was
1.3 points higher
(0.26 lower to 2.82 higher)

760
(three studies)

⊕⊕⊕⊝
moderate3

SMD 0.20 (‐0.04 to 0.44)

Absolute mean improvement 4% (1% reduction to 9% improvement). Relative improvement 7% (1% reduction to 15% improvement)2

Positive and active engagement in life
heiQ subscale for positive and active engagement in life. Scale from 1 to 6 (higher better)
Follow‐up: 12 months

Mean positive and active engagement in life in the control group, on 1 to 6 point scale (higher better), was
4.76 points4

Mean positive and active engagement in life in the intervention group was
0.2 points lower
(0.59 lower to 0.18 higher)

93
(one study)

⊕⊕⊝⊝
low3,5

Absolute mean worsening 3% (10% worse to 3% improved). Relative mean worsening 4% (12% worse to 4% improved)4

Pain
WOMAC subscale for pain . Scale from 0 to 20 (lower better)
Follow‐up: 12 months

Mean pain in the control group, based on 0 to 20 WOMAC pain subscale (lower is better), was
8.5 points2

Mean pain in the intervention groups was
0.3 points lower
(0.8 lower to 0.3 higher)

751
(three studies)

⊕⊕⊕⊝
moderate3

SMD ‐0.07 (‐0.21 to 0.08)

Absolute mean reduction in pain 1.3% (4.0% reduction to 1.5% increase). Relative mean reduction 3% (9% reduction to 3% increase)

Global OA scores
Multiple tools6. Scale from 0 to 96 (lower score better)
Follow‐up: 12 months

Mean global OA scores in the control group, based on 0 to 96 total WOMAC score (lower better), was
41.1 points2

Mean global OA score in the intervention group was
0.8 points lower
(3.7 lower to 2.1 higher)

751
(three studies)

⊕⊕⊕⊝
moderate3

SMD ‐0.06 (‐0.28 to 0.16)

Absolute mean improvement 0.8% (2% worse to 4% improved). Relative mean improvement 2% (5% worse to 9% improvement)2

Self‐reported function
Multiple tools7. Scale from 0 to 68 (lower better)
Follow‐up: six to 12 months

Mean self‐reported function in the control group, based on 0 to 68 WOMAC function scale (lower better), was
28.9 points2

Mean self‐reported function in the intervention groups was
1.1 points lower
(2.7 lower to 0.6 higher)

854
(four studies)

⊕⊕⊕⊝
moderate3

SMD ‐0.09 (‐0.22 to 0.05)

Absolute mean improvement 2% (4% improved to 1% worse). Relative mean improvement 4% (10% improved to 2% worse)2

Quality of life
Multiple tools8. Scale from 0 to 100 (higher better)
Follow‐up: 12 months

Mean quality of life in the control group, based on 0 to 100 point scale (higher better), was
55.9 points2

Mean quality of life in the intervention group was
0.5 points higher
(1 lower to 2 higher)

648
(two studies)

⊕⊕⊕⊝
moderate3

SMD 0.05 (‐0.1 to 0.21)

Absolute mean improvement 0.5% (1% worsening to 2% improvement). Relative mean improvement 1% (2% worsening to 4% improvement)2

Withdrawals
Losses to follow‐up
Follow‐up: six to 12 months

243 per 1,000

389 per 1,000
(182 to 827)

RR 1.6
(0.75 to 3.4)

1,251
(four studies)

⊕⊕⊝⊝
low3,9

Absolute difference 6% higher withdrawals (8% lower to 19% higher). Relative increase 60% (25% decrease to 240% increase)

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: Confidence interval; RR: Risk ratio.

GRADE Working Group grades of evidence.
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1ASES subscale for pain and heiQ subscale for self‐monitoring and insight.
2Control group baseline and final values taken from Buszewicz 2006, used to estimate the mean difference between groups and absolute and relative changes: mean (SD) self‐management score, based on arthritis self‐efficacy pain 5 to 35 point scale (higher score better), in the control group at baseline was 19.2 (6.4), and at follow‐up was 18.8 (6.5) points; mean (SD) pain score, based on 0 to 20 point WOMAC pain subscale (lower better) in the control group at baseline, was 8.7 (3.7) points, and at follow‐up was 8.5 (3.9) points; mean (SD) global OA score, based on WOMAC 0 to 96 point scale (lower better), in the control group at baseline was 41.6 (13.32) and at follow‐up was 41.4 points; mean (SD) function, based on WOMAC 0 to 68 (lower better), in the control group at baseline was 29.1 (12.7) and at follow‐up was 28.9 points; mean (SD) quality of life, based on SF‐36 mental component score 0 to 100 (higher better), in the control group at baseline was 50.6 (10.6) and at follow‐up was 55.9 points.
3Design flaws, including participants were not blind to group allocation in all trials, some trials had unclear randomisation method or concealment of allocation and unbalanced withdrawals across treatment groups; thus the evidence is susceptible to selection, performance, detection or attrition biases.

4Baseline and final value control group heiQ scores (1 to 6 point scale, higher better) from Ackerman 2012: Mean (SD) at baseline was 4.8 (0.8); mean at follow‐up was 4.76 points.
5Findings based on a single study.
6WOMAC and the hip and knee multi‐attribute priority tool (MAPT).
7Function subscales of WOMAC and the Dutch AIMS‐SF.
8Mental health component of the short form 36 (SF‐36) and assessment of quality of life (AQoL).

9Inconsistency across studies regarding whether greater number of withdrawals in the self‐management group or the control group.

Open in table viewer
Summary of findings 4. SMP compared with non‐SMP intervention for osteoarthritis

SMP compared with non‐SMP intervention for osteoarthritis

Patient or population: patients with osteoarthritis
Settings: community, outpatient or physiotherapy clinic, age care facility
Intervention: SMP
Comparison: non‐SMP intervention

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No of participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Non‐SMP intervention

SMP

Self‐management of OA
Arthritis self‐efficacy scale. Scale from 30 to 300.
Follow‐up: one to 12 months

Mean self‐management of osteoarthritis in the control groups, based on 30 to 300 ASES scale (higher score better), was
220.46 points1

Mean self‐management of osteoarthritis in the intervention groups was
12 points higher
(0 to 24 higher)1

175
(three studies)

⊕⊕⊕⊝
moderate2

SMD 0.33 (0 to 0.66)

Absolute mean improvement 4% (0% to 9% improvement). Relative improvement 5.7% (0% to 11.3% improvement)

Positive and active engagement in life

Not measured

See comment

See comment

Not estimable

0
(0)

See comment

No studies measured this outcome

Pain
Multiple tools3. Scale from 0 to 20. Lower score is better.
Follow‐up: one to 12 months

Mean pain in the control groups, based on 0 to 20 WOMAC pain subscale (lower better), was

4.2 points4

Mean pain in the intervention groups was
0.3 points lower
(1.2 lower to 0.5 higher)4

321
(five studies)

⊕⊕⊕⊝
moderate2

SMD ‐0.09 (‐0.36 to 0.17)

Absolute mean reduction 1.4% (‐5.8 to 2.7%). Relative reduction 6% (‐20% 9.5%)4

Global OA scores
WOMAC. Scale from 0 to 240. Lower score is better.
Follow‐up: 12 weeks

Mean global osteoarthritis scores in the control group, based on 0 to 240 WOMAC scale (lower better), was
66.8 points5

Mean global osteoarthritis score in the intervention group was
11.6 points higher
(5.6 lower to 28.7 higher)

98
(one study)

⊕⊕⊕⊝
moderate2

SMD 0.27 (‐0.13 to 0.67)

Absolute mean worsening 4.8% (12% worsening to 2.3% improvement). Relative worsening 12.7% (31.4% worsening to 6.1% improvement)5

Function—Self‐reported
Multiple tools3. Scale from 0 to 68. Lower score is better.
Follow‐up: one to 12 months

Mean function self‐reported in the control groups, based on 0 to 68 point WOMAC scale (lower better), was
12.2 points4

Mean function self‐reported in the intervention groups was
0.04 standard deviations higher
(0.34 lower to 0.42 higher)

216
(three studies)

⊕⊕⊕⊝
moderate2

SMD 0.04 (‐0.34 to 0.42)

Absolute worsening 0.6% (6.4% worsening to 5.2% improvement). Relative worsening 2.6% (27.4% worsening to 22.2% improvement)

Quality of life
Multiple tools6. Scale from 0 to 1. Higher score is better.
Follow‐up: 12 to 36 months

Mean quality of life in the control groups was
0.734

Mean quality of life in the intervention groups was
0.06 standard deviations lower
(0.49 lower to 0.36 higher)

226
(two studies)

⊕⊕⊕⊝
moderate2

SMD ‐0.06 (95% CI ‐0.49 to 0.36)

Absolute worsening 0.54% (4.4% worsening to 3.2% improvement). Relative worsening 0.7% (5.8% worsening to 4.3% improvement)4

Withdrawals
Losses to follow‐up.
Follow‐up: one to 36 months

243 per 1,000

209 per 1,000
(168 to 265)

RR 0.86
(0.69 to 1.09)

919
(seven studies)

⊕⊕⊕⊝
moderate2

Absolute difference of 2% fewer withdrawals (7% fewer to 12% more). Relative percentage change of 14% fewer withdrawals (31% fewer to 9% more)

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: Confidence interval; RR: Risk ratio.

GRADE Working Group grades of evidence.
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1Baseline and final values for comparison group taken from Keefe 2004, ASES 30 to 300 self‐efficacy scale (higher better): Control group baseline mean (SD) was 215 (36.95); control group follow‐up score was 220.46.
2Trials had design flaws, making results susceptible to bias, including participants and study personnel were not blind to group allocations; unclear randomisation and concealment of allocation; and selective reporting.
3Arthritis impact measurement scale (AIMS) and Western Ontario McMasters University Arthritis Index (WOMAC).
4Control group baseline and final values taken from Jessep 2009. WOMAC 0 to 20 point pain subscale (lower better): Control group mean (SD) pain at baseline was 5.7 (3.2) and at 12‐month follow‐up was 4.2 (4.0). Control group WOMAC 0 to 68 point function subscale (lower better): Control group mean (SD) at baseline was 15.9 (10.4) and at follow‐up was 12.2. Control group quality of life 0 to 1 point EQ‐5D scale (higher is better): Control group mean (SD) at baseline was 0.76 (0.09) and at follow up was 0.73.
5Baseline and final control group means from Maurer 1999: Control group mean at baseline was 91.5 points on 0 to 240 WOMAC scale (0 is better) and at 12 weeks was 66.8 points; control group SD taken from Wolfe 1999 (42.9).
6Quality of well‐being scale and EQ‐5D.

Background

Description of the condition

Osteoarthritis (OA) is a degenerative joint condition that affects primarily the weight‐bearing joints such as the hips, knees, ankles and spine but may also affect the hands. Worldwide, OA affects approximately 10% of men and 18% of women 60 years of age or older (Woolf 2003), and incidence increases with age.

The impact of OA includes pain, difficulty in performing activities of daily living, dependency on family and friends for assistance, reduced quality of life, lost productivity and personal economic impact associated with ongoing care and management (ABS 2004; March 2004). The major costs of OA within the health sector are related to joint replacement, visits to general practitioners (GPs) and specialists, prescription and over‐the‐counter medications and allied health care. As the condition is not reversible, the growing number of people with OA will result in a greater burden of disease; current estimates indicate that OA will be the fourth leading cause of disability by the year 2020 (Woolf 2003).

Description of the intervention

Support is available for coordinated delivery of patient self‐management education programmes to improve healthcare outcomes for people with OA (Osborne 2004). Self‐management education programmes are distinct from simple patient education or skills training, as they encourage people with chronic disease to take an active role in the management of their own condition. Self‐management education programmes aim to improve outcomes for patients by supporting, not replacing, medical care (Walker 2003).

Self‐management education programmes are complex behavioural interventions comprising a package of interventions specifically targeted at patient education and behaviour modification. Programmes vary in the content used to educate patients about their condition and to explain how they can best manage their symptoms. Some programmes specifically focus on managing the chronic condition itself, whereas other programmes may take a more holistic approach to managing the overall general well‐being of the individual. Substantial variation exists in the delivery of self‐management education programmes, such as the mode (face‐to‐face, Internet, telephone), the audience (group, individual), the duration (single session, several months, ongoing), the frequency (once a week, once every two months) and the personnel (healthcare professionals, lay leaders).

How the intervention might work

The pathology associated with joints affected by OA is typically irreversible. Aside from joint replacement, interventions are usually targeted at maintaining or improving life with the condition rather than improving the condition itself. A variety of terms in the literature describe self‐management, including self‐care, self‐monitoring, self‐help and social support (Walker 2003). Several models of self‐management are known (Osborne 2004); however, the core steps involve (1) engaging in activities that promote health and prevent adverse sequelae; (2) interacting with healthcare providers; (3) performing improved self‐monitoring of physical and emotional status; and (4) managing the effects of illness on a person’s ability to function in important roles and on emotions, self‐esteem and relationships with others (Von Korff 1997). The skills required for these tasks include problem solving, decision making, finding and utilising resources, forming partnerships with healthcare workers and taking action (Lorig 2003). 

Assessing the characteristics and impact of self‐management education programmes

Studies of self‐management education programmes have varied widely in their attempts to quantify the potential impact of these programmes on participant health and well‐being. This has resulted in significant heterogeneity in outcome assessment across studies and has contributed to inconsistencies in reported effectiveness of programmes. Understanding which outcomes are most relevant to assessment of the effectiveness of self‐management education programmes is required, so that programmes can be assessed systematically on the basis of outcomes that we know are important to participants.

The Arthritis Self‐Efficacy Scale (ASES) was the first arthritis‐specific instrument developed to measure the effects of arthritis self‐management programmes (Lorig 1989). It consists of three subscales (pain, function and other symptoms) and includes efficacy expectation items that ask individuals how certain they are that they can perform a specific activity, for example, walking 100 feet on flat ground in seven seconds; as well as performance attainment items, for example, how certain they are that they can control their fatigue or deal with the frustration of arthritis. Although these items capture an individual's ability to self‐manage and therefore are useful in measuring outcomes of self‐management education programmes, the validity of the ASES as a true self‐efficacy measure has been questioned (Brady 1997; Brady 2011). Although the ASES includes items pertaining to efficacy expectations, it does not ask about an individual's confidence that different behaviours will produce the desired outcome (outcome expectations)—an integral component of Bandura's theory of self‐efficacy (Bandura 1977). In addition, the function subscale items appear to capture perceived physical function rather than self‐efficacy belief.

Recent research has been undertaken to identify key indicators of effective self‐management interventions from the patient perspective (Osborne 2007). Development of the Health Education Impact Questionnaire (heiQ) involved extensive engagement and consultation with consumers and healthcare professionals regarding the outcomes they consider to be valuable and direct benefits of self‐management programmes. Eight independent domains were described and form the basis of the constructs of the questionnaire. Domains identified as key indicators of effective self‐management programmes include health‐directed behaviour; positive and active engagement in life; emotional well‐being; self‐monitoring and insight; constructive attitudes and approaches; skill and technique acquisition; social integration and support; and health service navigation. The constructs used in the heiQ have been shown as robust across a range of settings (Nolte 2007; Nolte 2009).

Why it is important to do this review

Several previous systematic reviews have summarised the effects of self‐management programmes (Chodosh 2005; Devos‐Comby 2006; Warsi 2003) or arthritis patient education interventions (Hirano 1994; Lorig 1987; Superio‐Cabuslay 1996) for people with OA or for mixed populations (including people with chronic diseases such as OA). However, some of these previous reviews have (1) combined data from studies that included both people with OA and individuals with rheumatoid arthritis (Warsi 2003); (2) restricted inclusion of studies to those in which participants have OA affecting the knee only (Devos‐Comby 2006); (3) restricted inclusion of studies to those in which the comparison intervention was a non‐steroidal anti‐inflammatory drug treatment only (Superio‐Cabuslay 1996); or (4) did not employ rigorous and systematic methods of searching, appraising and synthesising the evidence as necessary to produce reliable systematic reviews on the effects of healthcare interventions (Hirano 1994; Lorig 1987).

Existing clinical practice guidelines uniformly recommend self‐management for OA of the hip and knee (March 2010; Zhang 2007 ). If broad implementation of self‐management education programmes for OA are to be considered, a strong evidence base of support is needed. Evidence must show that self‐management education programmes improve functional, psychological and/or social outcomes for people with OA. Furthermore, it is important to identify any variables that may significantly affect outcomes of the intervention (e.g. age, stage of disease, comorbidities), as well as contextual issues around content, settings and methods of delivery of self‐management education programmes. A systematic review of all randomised controlled trials (RCTs) to date would determine whether self‐management education programmes are effective in improving outcomes for people with OA.

Objectives

To assess the effectiveness of self‐management education programmes for people with osteoarthritis.

Methods

Criteria for considering studies for this review

Types of studies

We included RCTs or quasi‐randomised trials in which group assignments were determined through methods other than true randomisation (i.e. alternate assignment).

Types of participants

We included studies of people of all age groups diagnosed with OA as defined in the included trials. Studies involving people with conditions other than OA (i.e. mixed populations) were included only if outcomes for people with OA were presented separately, or if a high proportion of participants (90% or greater) had OA.. We planned to contact trial authors to obtain separate data for people with OA if these had not been reported.

Types of interventions

We included structured self‐management education programmes that were judged as being primarily educational and that addressed self‐management of OA, arthritis in general or living with chronic disease. Programme components that directly address self‐management may include fostering skills in managing OA, such as problem solving, goal setting, decision making, self‐monitoring and coping with the condition, as well as providing interventions to manage pain or improve physical and psychological functioning.

Structured programmes delivered by healthcare professionals, lay leaders or both were included, irrespective of whether the programme was delivered to a group of participants or on an individual basis. Studies describing interventions for self‐management that are not delivered within a structured format or by some form of organised content delivery were excluded. All modes of delivery, such as face‐to‐face or interventions delivered by post, Internet, or telephone, were included, provided they incorporated an iterative process of interaction between the participant and programme facilitators. Studies were not excluded on the basis of duration or location of self‐management education programmes. We excluded interventions that were judged as treating participants as passive recipients of care (e.g. provision of information alone). Studies that focused solely on exercise were not included.

Programmes that incorporate education for carers or relatives were included, provided the intervention was primarily intended for the person with OA.

Studies that compared outcomes of people with OA assigned to a self‐management education programme versus those of individuals who did not receive a self‐management educational intervention (i.e. information only, no treatment, usual care, waiting list control, or alternative interventions not considered self‐management education programmes) were eligible for inclusion. Studies comparing different self‐management education programmes without an appropriate comparison group (i.e. no programme) were excluded. Co‐interventions were considered, provided the comparison group received the same co‐intervention.

Studies that compared one type of self‐management programme versus another were excluded.

Types of outcome measures

No studies were excluded on the basis of outcomes reported. All immediate (up to six weeks from the start of the intervention), intermediate (up to and including one year after the intervention) and longer‐term outcomes (longer than one year after the intervention) are reported.

Main outcomes

The following outcomes were selected as the most relevant and are included in the Summary of findings tables.

  • Self‐management of OA (participant's self‐monitoring and insight into living with OA).

  • Participant's positive and active engagement in life (including return to work, fulfilling his or her role within the family).

  • Pain,

  • Global OA scores (e.g. Western Ontario and McMaster Universities Arthritis Index (WOMAC), Lequesne Osteoarthritis Index global score).

  • Self‐reported function (e.g. function as measured on the WOMAC function subscale).

  • Quality of life (including participant‐reported general health status).

  • Withdrawals (including withdrawals related or unrelated to the study intervention (dropouts) and individuals lost to follow‐up). Originally, we planned to assess adverse events or withdrawals (when reported reasons for withdrawal are related to the intervention); however, none of the included trials reported this outcome.

Other outcomes

We also included the following outcomes, which are relevant to the impact of self‐management.

  • Performance measures (e.g. the six‐minute walking distance test, the timed up‐and‐go test).

  • Emotional distress (including depression, anxiety, stress).

  • Health‐directed activity (including adherence, exercise, diet, relaxation).

  • Social integration and support (including social participation, social network, social input).

  • Health service navigation (visits to healthcare professionals, emergency room visits, hospital admissions, length of stay).

  • Skill and technique acquisition (including knowledge about the condition and how symptoms can be managed).

  • Constructive attitude and approaches (including changes in perceived impact of OA on participant's life).

  • Participant satisfaction.

In the light of the common use of arthritis‐specific self‐efficacy measures, particularly the ASES (Lorig 1989), as outcome measures in trials of self‐management education programmes, we considered whether to include self‐efficacy as a separate outcome. However, as these tools may not comprehensively capture all aspects of self‐efficacy theory and may include items measuring performance attainment, we considered an important or meaningful distinction between the outcome of 'self‐management' and 'self‐efficacy' to be insufficient to justify treating these as separate outcomes in the review. We therefore included arthritis self‐efficacy scales within the primary outcome of self‐management.

Search methods for identification of studies

We searched the following electronic databases for primary studies, up to January 17, 2013.

  • The Cochrane Central Register of Controlled Trials (CENTRAL via The Cochane Library, Issue 1, 2013).

  • MEDLINE (Ovid 2005 to January 17, 2013).

  • EMBASE (Ovid 2010 to January 2013).

  • CINAHL (EBSCOHost).

  • PsycINFO (1806 to January 2013).

  • Dissertation Abstracts (ProQuest January 2013).

  • SCOPUS (January 2013).

The Database of Abstracts of Reviews of Effects (DARE) via The Cochane Library was searched to identify potentially relevant reviews and the reference lists screened to identify primary studies. Reference lists of relevant studies were also screened to identify potential studies for inclusion in the review.

We also searched the World Health Organization International Clinical Trial Registry Platform (ICTRP) (www.who.int/ictrp/en) to identify trials in progress.

Search strategies and time periods for each database are listed in the appendices (Appendix 1; Appendix 2; Appendix 3; Appendix 4; Appendix 5; Appendix 6; Appendix 7; Appendix 8). The search strategy combined text words and controlled vocabulary for describing arthritis/OA and self‐management programmes using the Cochrane highly sensitive methodological filter for controlled trials (Lefebvre 2011). We applied no restrictions based on language or publication status. The MEDLINE and EMBASE searches were limited to 2005 to 2013 and 2010 to 2013, respectively, to reduce duplication, as RCTs from both of these databases are included in CENTRAL for the years before these time periods.

A flow diagram for the search results and for selection of studies is presented in Figure 1.

Figure 1
Flowchart.

Flowchart.

Data collection and analysis

Selection of studies

Two review authors (VP and FPBK) assessed the titles and available abstracts of all studies identified by the initial search and excluded any clearly irrelevant studies. Studies were not excluded on the basis of the language of published articles. We included both published and unpublished reports. Two review authors (FPBK and LRAB) independently applied the selection criteria to full‐text reports of potentially eligible studies. The review authors resolved any disagreements through discussion until consensus was reached or by arbitration by a third review author (VP, RB or RHO) when required. Reasons for exclusion are provided in cases where studies could be considered plausible for inclusion but were excluded from the review.

Data extraction and management

Two review authors (FPBK and LRAB) independently extracted data from the included trials, including information about the study population, interventions, analyses, outcomes and sources of funding, using a standardised data extraction form specifically designed and piloted for this review.

Contextual factors and characteristics of the population relevant for addressing potential issues in health equity were extracted using the PROGRESS‐Plus concept (place of residence; race, ethnicity and culture; occupation; sex; religion; education; socioeconomic status; social capital; age; disability; and sexual orientation) (Bambas 2004; Borkhoff 2011).

Health literacy of the population may be another potentially important issue of relevance to health equity that is not currently captured in the PROGRESS‐Plus framework. The World Health Organization describes health literacy as “the cognitive and social skills which determine the motivation and ability of individuals to gain access to, understand and use information in ways which promote and maintain good health” (World Health Organization 1998). In both developing and developed countries, health and social policies are emerging that highlight health literacy as a key determinant of a person’s ability to optimally manage his or her health and ensure equitable access to and use of services (Committee on Health Literacy 2004; Commonwealth of Aus 2009; United Nations Economic and Social Council 2010). We therefore also extracted information regarding health literacy of the study population, if available, using the nine domains of the Health Literacy Questionnaire (HLQ) (Osborne 2013).

  • Feeling understood and supported by healthcare providers.

  • Having sufficient information to manage my health.

  • Actively managing my health.

  • Social support for health.

  • Appraisal of health information.

  • Ability to actively engage with healthcare providers.

  • Navigating the healthcare system.

  • Ability to find good health information.

  • Understand health information well enough to know what to do.

The following information was collected for each self‐management education programme.

  • Intended audience (people with OA, arthritis or chronic disease).

  • Mode (delivered on a one‐to‐one basis or to groups of participants).

  • Personnel (led by healthcare professionals or by trained facilitators).

  • Delivery method (face‐to‐face, written, audio, video, phone, Internet).

  • Language (English or other languages).

  • Format (tailored to the individual's needs or delivered in standard format).

  • Location (hospital, GP clinic, community setting, home).

  • Duration (number and frequency of sessions, hours per session, total duration of programme).

We also extracted information about the components of each self‐management education intervention using the eight domains described in the Health Education Impact Questionnaire (heiQ) (Osborne 2007). Each of these domains have been identified as an independent outcome indicator of effective self‐management interventions and has been found to be robust across settings (Osborne 2007). We assessed whether interventions were developed on the basis of an explicit theoretical framework (e.g. models of behavioural theory) or a set of principles (e.g. principles of adult education), and whether each of the following components was addressed within each programme.

  • Health‐directed activity.

  • Positive and active engagement in life.

  • Emotional distress.

  • Self‐monitoring and insight.

  • Constructive attitudes and approaches.

  • Skill and technique acquisition.

  • Social integration and support.

  • Health service navigation.

To assess the effects of an intervention, we extracted raw data for outcomes of interest (means and standard deviations for continuous outcomes and number of events for dichotomous outcomes) when available in the published reports.

We contacted the authors of all studies to obtain more information as needed.

If a study reported multiple time points within immediate, intermediate or longer‐term outcomes, only the longest time point was extracted.

Assessment of risk of bias in included studies

Two review authors (FPBK and LRAB) independently assessed the risk of bias in each included study against key criteria: random sequence generation; allocation concealment; blinding of outcomes; incomplete outcome data; and selective outcome reporting. We also considered other sources of bias, such as whether the intervention was delivered as intended, whether groups were comparable at baseline and whether contamination between groups was possible. Assessments were conducted in accordance with methods recommended by The Cochrane Collaboration (Higgins 2011).

Selective outcome reporting was judged on the basis of whether all outcomes assessed in a trial have been reported. When possible, we obtained trial protocols for comparison of planned outcome assessment versus the outcome data available from each trial. For trials published after July 1, 2005, we searched for trial protocols using the Clinical Trial Register at the International Clinical Trials Registry Platform of the World Health Organization (http://apps.who.int/trialssearch; DeAngelis 2004). We also constructed outcome matrices listing primary outcomes reported across the included studies to identify inconsistencies in outcome reporting to indicate possible selective outcome reporting. We planned to use the Outcome Reporting Bias In Trials (ORBIT) classification system to describe whether selective outcome reporting was suspected and the potential reason for it (Kirkham 2010).

Each of the domains assessed for risk of bias is explicitly judged as follows: Yes (low risk of bias); No (high risk of bias); or Unclear (lack of information or uncertainty over the potential for bias). Review authors resolved disagreements through discussion until consensus was reached, or consulted a third review author (RB) to resolve disagreements if necessary.

Measures of treatment effect

Point estimates and 95% confidence intervals (CIs) were calculated for outcomes of individual RCTs whenever possible. Point estimates for dichotomous outcomes are expressed as risk ratios (RRs). For continuous outcomes, results are summarised as mean difference (MDs) if the same tool has been used to measure the same outcome across separate studies. Alternatively, we summarised treatment effects using the standardised mean difference (SMD) when studies measured the same outcome but employed different tools. If results could not be summarised as point estimates with 95% CIs, we tabulated results for each outcome.

Studies included in forest plots were listed in order of the weight (from lowest to highest weight) that each individual study contributed to the overall summary estimate.

Unit of analysis issues

When appropriate, we incorporated results of cluster‐randomised trials into meta‐analyses using the generic inverse variance method in RevMan (Deeks 2011). Effect estimates (e.g. RR and 95% CI) for relevant outcomes were extracted from cluster trials that had appropriately accounted for the cluster design. When trials had not appropriately accounted for the design effect, we corrected the standard errors of effect estimates by using an intraclass correlation coefficient that was obtained from the trial report or estimated from similar studies.

Dealing with missing data

We contacted trial authors if the type of arthritis had not been specified or when studies involved mixed populations, in an attempt to obtain separate data for people with OA. Clarification was also sought for descriptions of interventions (e.g. setting, mode of delivery, format, duration), trial conduct (e.g. method of random sequence generation, method of allocating participants to treatment groups, blinding of trial personnel) and availability of unpublished data for outcomes that were measured.

For outcomes assessed using standard scales (e.g. WOMAC, quality of life scales), we attempted to present overall scores when possible. If results were presented only for separate subscales, we used results of subscales considered to be most relevant to the outcome of interest and recorded instances where this applied in the Notes section of the Characteristics of included studies.

When the number of people assessed for an outcome was unclear, we imputed this on the basis of the number of people originally randomly assigned to the study groups. For continuous measures, missing standard deviation (SD) values were estimated from other measures such as standard error (SE), P values or confidence intervals whenever possible, or they were imputed on the basis of SD values in similar trials (Higgins 2011). For dichotomous outcomes, percentages were used to estimate the number of events or the number of people assessed for an outcome. All data imputations are recorded in the Notes section of the Characteristics of included studies.

Assessment of heterogeneity

Before a meta‐analysis was conducted, studies were assessed for similarities with respect to characteristics of the self‐management education programmes, comparison groups and outcomes. Studies judged by the review authors as being too different from each other were not combined in the analysis but were described separately in the text of the review.

For studies judged as sufficiently similar, statistical heterogeneity was assessed visually by looking at the scatter of effect estimates on the forest plots and by determining the I2 statistic (Higgins 2003). The I2 statistic was used as an indication of the proportion of heterogeneity, with higher values indicating a higher proportion of heterogeneity, using the following as a rough guide for interpretation: 0 to 40% might not be important, 30% to 60% may represent moderate heterogeneity, 50% to 90% may represent substantial heterogeneity and 75% to 100% may represent considerable heterogeneity (Deeks 2011). In cases of considerable heterogeneity (defined as I2 ≥ 75%), we explored the data further by comparing characteristics of individual studies and any subgroup analyses and reported any differences when interpreting the results of this review, or we reported I2 values whenever unexplained statistical heterogeneity was present.

Assessment of reporting biases

Selective outcome reporting was assessed using the approach described previously in this protocol (see Assessment of risk of bias in included studies). The potential impact of selective outcome reporting on the overall results of the review was discussed in terms of suspected reasons for the missing outcome data and the size, strength and direction of the effect.

To assess for potential small‐study effects in meta‐analyses (i.e. the intervention effect is more beneficial in smaller studies), we compared effect estimates derived from a random‐effects model and from a fixed‐effect model of meta‐analysis. In the presence of small‐study effects, the random‐effects model will give a more beneficial estimate of the intervention than the fixed‐effect estimate (Sterne 2011).

It was planned that the potential for small‐study effects in the main outcomes of the review would be further explored using funnel plots if at least 10 studies were included in a meta‐analysis.

Data synthesis

Included studies were grouped and assessed according to whether they compared self‐management education programmes versus:

  • an attention control group (i.e. participants received the same contact hours with programme providers, but the content delivered was unrelated to self‐management of their condition);

  • a group that received no treatment or usual care or were placed on a waiting list to attend the self‐management programme at a later date;

  • an information‐only group (i.e. educational materials, programme handbook);

  • a group that received an alternate intervention that was not a self‐management education programme (i.e. exercise or diet plan); or

  • a group that received acupuncture. This intervention was considered separately from other studies comparing self‐management programmes versus alternative interventions (i.e. comparison described in the previous bullet) because, unlike alternative interventions in the other trials, this comparison is not a behavioural intervention.

We considered the first two comparisons to be the most important for addressing the objectives of this review. Evidence from physical therapy trials suggests that the quality of the therapeutic relationship influences clinical outcomes such as pain and physical function (Hall 2010; Pinto 2012). This effect may also apply to self‐management education programme providers. An attention control may control for any effect of contact time with programme providers, and we considered this to be the comparator with the lowest risk of bias in determining the true effect of self‐management education programmes. We considered usual care to be an important comparator as well, as this reflects routine care. However, comparisons versus usual care are generally unblinded (analogous to no treatment), while an attention control allows for blinding of participants (of utmost importance when outcomes are participant assessed) so is closer to a sham/placebo control.

When studies were considered to be sufficiently similar in terms of the self‐management education programme delivered and the comparison intervention provided, we pooled outcomes in a meta‐analysis using the random‐effects method (Deeks 2011). Forest plots display individual study results sorted by weight in ascending order.

To minimise outcome reporting bias, if data from more than one self‐efficacy scale were reported for a trial, we extracted data according to the following hierarchy.

  • Self‐efficacy on a visual analogue scale.

  • Arthritis self‐efficacy scale mean score.

  • Arthritis self‐efficacy subscale (pain or other symptoms).

  • Self‐efficacy on other scales (i.e. Arthritis Helplessness Index, heiQ self‐monitoring and insight).

If data on more than one pain scale were provided for a trial, we referred to a previously described hierarchy of pain‐related outcomes (Jüni 2006) and presented data on the pain scale that was highest on this list.

  • Global pain.

  • Pain on walking.

  • WOMAC pain subscore.

  • Composite pain scores other than WOMAC.

  • Pain on activities other than walking.

  • Rest pain or pain during the night.

If data on more than one global OA scale were provided for a trial, we extracted data according to the following hierarchy.

  • Patient global assessment score.

  • Self‐rated global health.

  • MAPT (Multi‐attribute Arthritis Prioritisation Tool).

  • WOMAC total score.

  • AIMS (Arthritis Impact Measurement Scales) total score.

Similarly, if data on more than one self‐reported function scale were provided for a trial, we extracted data according to the hierarchy presented below (Rutjes 2009).

  • Global disability score.

  • Walking disability.

  • WOMAC disability subscore.

  • Composite disability scores other than WOMAC.

  • Disability other than walking.

If data on more than one quality of life scale were provided for a trial, we extracted data according to the following hierarchy.

  • Short Form (SF)‐36.

  • SF‐12.

  • EuroQoL.

  • SIP (Sickness Impact Profile).

  • NHP (Nottingham Health Profile).

  • Other validated quality of life scores.

If data on a quality of life scale were provided in both a multi‐question format and a visual analogue scale format, we chose the first, as we judged this would provide a more accurate measure of quality of life and patient‐reported global health status.

If data for both anxiety and depression were presented, we chose to extract only the data on depression for the outcome of emotional distress, as we judged that depression overall was more consistently reported as a measure of emotional distress when compared with anxiety.

Summary of findings

We presented the main outcomes of the review in Summary of findings tables (self‐management, positive and active engagement in life, withdrawals, pain, global OA scores, self‐reported function and quality of life) to provide key information concerning the quality of evidence, the magnitude of effect of the interventions examined and the sum of available data on the main outcomes, as recommended by The Cochrane Collaboration (Schünemann 2011a). The 'Summary of findings' tables provide an overall grading of the evidence related to each of the main outcomes based on the GRADE approach (Schünemann 2011b).

Overall outcome data presented in the Summary of findings tables are based on the longest time points measured in each study. Separate analyses (forest plots not shown) were performed to include all possible studies per outcome, as we did not have a prespecified primary time point, and there did not seem to be an effect of time. Outcomes pooled using SMDs were re‐expressed as a mean difference by multiplying the SMD by a representative control group baseline standard deviation from a trial, using a familiar instrument.

In the comments column, we calculated the absolute percentage change and the relative percentage change; and, for outcomes with statistically significant differences between intervention groups, we calculated the number needed to treat for an additional beneficial outcome (NNTB).

For dichotomous outcomes, the absolute risk difference was calculated using the risk difference statistic in RevMan and the result expressed as a percentage; the relative percentage change was calculated as the risk ratio ‐1 and was expressed as a percentage; and the NNT from the control group event rate and the risk ratio were determined using the Visual Rx NNT calculator (Cates 2008).

For continuous outcomes, the absolute risk difference was calculated as the mean difference between intervention and control groups in the original measurement units (divided by the scale), expressed as a percentage; the relative difference was calculated as the absolute change (or mean difference) divided by the baseline mean of the control group from a representative trial. We used the Wells calculator to obtain the NNTB for continuous measures (available at the Cochrane Musculoskeletal Group (CMSG) Editorial office; http://musculoskeletal.cochrane.org/). The minimal clinically important difference (MCID) for each outcome was determined for input into the calculator. We assumed an MCID of 1.5 points on a 10‐point pain scale, 0.5 points on the 10‐point ASES (self‐management) scale and 0.5 points on the WOMAC function 0 to 68‐point scale.

Subgroup analysis and investigation of heterogeneity

We undertook an exploratory analysis of studies assessing the effects of self‐management education programmes and issues of health equity. We compared effect sizes for the major outcomes of the review, as well as self‐reported pain across studies comparing self‐management education programmes versus a control group. Studies were classified according to whether the study population consisted mainly of Caucasian, educated, older females or populations drawn from minority groups within the community (e.g. culturally and linguistically diverse populations).

Sensitivity analysis

A sensitivity analysis was conducted and was based on whether participants were randomly allocated and group assignments had been adequately concealed.

Results

Description of studies

Results of the search

The initial database search identified 2,248 records (see Figure 1). We assessed 151 possibly eligible papers in full text. Of these, 29 studies published between 1990 and 2012, involving 6,753 participants (range 32 to 570), met the inclusion criteria for this review. A further eight trials were identified in an updated search performed on 17 January 2013, and, as they were unlikely to alter the conclusions of the review, these trials will be assessed when the review is updated (see Studies awaiting classification).

Included studies

A full description of all included studies is provided in the Characteristics of included studies table.

Design

Of the 29 included studies, three were cluster‐RCTs (Hurley 2007; Mazzuca 2004; Victor 2005), and the remaining 26 studies were RCTs. All studies were published in English.

Participants

A description of participants using the PROGRESS‐Plus framework is shown in Table 1. Studies were most commonly conducted in the US (17 studies, 58.6%) followed by the UK (four studies, 13.8%), The Netherlands (three studies, 10.3%) and Australia (two studies, 6.9%). Single studies were performed in Spain, Sweden and Hong Kong (China). Most participants (68%) were female, and the average age of participants was 64.8 years. Race and ethnicity were reported in 15 studies (Allen 2010; Berman 2004; Blixen 2004; Buszewicz 2006; Calfas 1992; Cronan 1997; Hughes 2004; Lorig 2008; Maisiak 1996; Mazzuca 1997; Mazzuca 2004; McKnight 2010; Murphy 2008; Victor 2005; Yip 2007), and 70.2% of participants were described as white or Caucasian. PROGRESS‐Plus domains that were least described across studies included occupation (nine studies) and socioeconomic status (six studies). Only four trials provided any information related to health literacy, and this was limited to the domains of social support for health (three trials), navigating the healthcare system (two trials), actively managing my health (two trials) and ability to actively engage with healthcare providers (one trial).

Open in table viewer
Table 1. PROGRESS‐Plus

Study

P

R

O

Gr

E

S

S

A

D

HL

Allen 2010

US

54% white,

46% non‐white,

2% Hispanic

38.3% employed

92.7% male,

7.3% female

33.3% high school education or less

25.7% inadequate income

69.3% married

60.1 (10.4)

32.3% fair or poor health

‐

Berman 2004

US

69.1% white,

28.6% African American,

2.3% other

‐

35.9% male,

64.1% female

31.1% no college, 68.9% some college

‐

‐

65.5 (8.6)

75.1% moderate to lesser pain, 24.9% severe to extreme pain

‐

Blixen 2004

US

72% white,

28% African American

‐

62.5% male,

37.5% female

6% grade 7 to 9,

3% grade 10 and 11, 22% high school, 50% 1 to 4 year college, 9.5% college graduate, 9.5% professional/graduate school

22% < $10,000/y, 13% $10,000 to $19,999/y, 19% $20,000 to $29,999/y, 17.5% $30,000 to $39,999/y, 13.5% $40,000 to $49,999/y, 17.5% > $50,000/y

56.5% married, 25% divorced,

13% single,

25% widowed

70.8 (6.1)

16% joint replacement surgery

Domain social support for health

Buszewicz 2006

UK

> 99% white,

< 1% black Caribbean,

< 1% black African

‐

37% male,

63% female

72.5% no higher education,

27.5% higher education

‐

81% house owner, 1% staying with friends or family, 18.5% rented accommodation

68.6 (8.4)

‐

‐

Calfas 1992

US

97.5% white,

2.5% non‐white

82.5% retired, 5% full‐time employed, 12.5% part‐time employed

27.5% male,

72.5% female

‐

‐

70% married, 20% divorced, 10% widowed,

5% other

67.0 (7.3)

‐

‐

Cronan 1997

US

92.3% white,

2.5% African American,

2.2% Hispanic,

1.7% Asian,

1.4% other

‐

36.3% male, 64.0% female

31.2% high school, 37.7% some college, 28.7% college degree

27.2% < $20,000, 43.2% $20,000 to $40,000, 19.7% $40,000 to $60,000, 10.0% > $60,000

‐

69.2 (5.6)

69.9% other medical conditions present

‐

Crotty 2009

Australia

‐

11.6% employed, 5.4% home duties, 82.7% retired

39.5% male,

60.5% female

3.5% none to some primary school, 15.0% primary school, 30.6% high school to year 8, 27.2% high school to year 12, 17.4% TAFE/trade, 6.2% university to above

‐

48.3% live alone

67.5 (10.8)

32.9% on waiting list for hip replacement

Domains social support for health, navigating health system, actively managing my health

Hansson 2010

Sweden

‐

‐

‐

‐

‐

‐

62.5 (9.4)

‐

‐

Heuts 2005

The Netherlands

‐

40.5% paying job, 35.5% no paying job

40.5% male,

59.5% female

25.5% low,

30.5% middle, 20% high

‐

‐

51.6 (5.1)

‐

‐

Hopman‐Rock 2000

The Netherlands

‐

‐

17% male,

83% female

21.5% primary, 49.5% secondary, 23.5% college/university

‐

67.5% living together/married, 27% living alone

65.3 (5.5)

2.5 (1.6) other chronic conditions

‐

Hughes 2004

US

72.2% white,

22.1% African American, 2.6% Hispanic,

2.1% Asian‐Pacific Islander, 1.1% other

‐

16.8% male, 83.3% female

10.5% less than high school, 20.0% high school, 69.6% more than high school

33.1% income less than $20,000

‐

73.4

22.4% ARA class I, 64.4% ARA class II, 13.3% ARA class III

‐

Hurley 2007

UK

‐

‐

29.7% male, 70.3% female

‐

‐

‐

67 (range 50 to 91)

‐

‐

Jessep 2009

UK

‐

‐

30.5% male, 69.5% female

‐

‐

‐

66.5 (range 51 to 81)

‐

‐

Keefe 1990

US

‐

‐

28.3% male, 71.7% female

‐

‐

‐

63.9 (11.5)

‐

‐

Keefe 1996

US

‐

‐

40% male,

60% female

‐

‐

‐

62.6

‐

‐

Keefe 2004

US

‐

‐

46.9% male, 53.4% female

‐

‐

‐

59.5 (11.1)

‐

‐

Lorig 2008

US

92.3% non‐Hispanic white

‐

9.9% male, 90.2% female

15.7 (3.1) years of education

‐

68.3% married

52.4 (11.6)

‐

Domain ability to actively engage with healthcare

providers

Maisiak 1996

US

85% white

‐

7.7% male, 92.3% female

12.1 years of schooling

‐

‐

60.4

48% see a specialist

‐

Martire 2007

US

‐

‐

27.3% male, 72.7% female

‐

‐

‐

68.6 (7.6)

‐

‐

Maurer 1999

US

‐

‐

58.5% male, 41.5% female

‐

‐

‐

65.4 (8.6)

‐

‐

Mazzuca 1997

US

69% African American

53% unemployed, 15% employed, 31.5% retired

15% male, 85% female

9.7 (3.1) years of education

96.5% annual income ≤ $20,000

73.5% living alone

62.4 (11.6)

1.6 (1.1) comorbid conditions

‐

Mazzuca 2004

US

70.5% white

‐

28.5% male, 71.5% female

87.5% 12 or more years of education

‐

62% married

61.8 (12.2)

‐

‐

McKnight 2010

US

91.7% white

‐

23.4% male, 77.0% female

63.0% college educated

‐

‐

52.6 (7.2)

‐

‐

Murphy 2008

US

91% white

‐

11% male, 89% female

66.5% some college to advanced degree

‐

22% married

75.3 (7.1)

4.5 (2.1) painful or stiff joints

‐

Nunez 2006

Spain

‐

84% retired or housewives, 13% permanently disabled, 3% active (sick leave)

24.5% male, 75.5% female

‐

‐

67.5% have family, 30.5% alone, 3% living with carer

71.1 (6.7)

33% prior prostheses

‐

Ackerman 2012

Australia

68.5% Australian‐born

26% paid employment, 62% retired, 8.5% not working because of medical condition, 3.5% unemployed

40% male, 60% female

12% primary school or less, 46% years 7 to 10, 15.5% years 11 and 12, 14.5% trade/technical education, 12% university

‐

63.5% married or living with partner

65.1 (10.9)

‐

Domains social support for health, navigating

health system, actively managing

my health

Victor 2005

UK

64% non‐white ethnic group

30% employed, 33.5% professional or managerial job

28% male, 72% female

37% higher education

66% home owner

35.5% living alone, 51% married

63.1 (11.1)

60% OA in both knees, 64% OA in other joints

‐

Wetzels 2005

The Netherlands

‐

‐

24% male, 76% female

52% primary or lower secondary, 48% upper secondary or further

‐

‐

74.5 (6.4)

‐

‐

Yip 2007

Hong Kong

100% Asian

26.4% housewife, 8.8% professional and administration, 64.9% service provider and non‐professional

16.0% male, 84.1% female

87.4% Form 3 level or below, 12.7% above Form 3

‐

69.8% married and living together, 30.3% single

64.8 (10.0)

10.9% one joint OA, 48.9% two joints OA, 23.8% three joints OA, 16.5% four or more joints OA

‐

P = Place of residence.
R = Race, ethnicity, culture.
O = Occupation.
Gr = Gender.
E = Education.
S = Socioeconomic status.
S = Social capital.
A = Age (in years).
D = Disability.
HL = Health literacy.

If not stated differently, data are ‘mean (SD)’.

y = year; IQR = interquartile range, ‐ = no information available.

No information was available on 'Religion' and 'Sexual orientation'; therefore these domains are not included in this table.

Other characteristics of participants (location and duration of OA, body mass index (BMI)) are listed in Table 2. Location of osteoarthritis was reported in 20 of 29 studies (69%), and the predominant location was the knee. BMI was reported in only nine of 29 studies (31%). The duration of OA was reported in 15 studies and ranged from a few months to longer than 20 years.

Open in table viewer
Table 2. Additional characteristics of included studies

Study

No

Location of OA

BMI

Duration of OA

Allen 2010

523

80% knee, 15% hip, 5% knee and hip

31.8 (6.7)

16.1 (12.2) years

Berman 2004

570

100% knee

‐

50.4% < 5 years,

20.7% 6 to 10 years,

28.7% > 10 years

Blixen 2004

32

‐

‐

8.3 (IQR 2 to 24) years

Buszewicz 2006

812

‐

‐

‐

Calfas 1992

40

‐

‐

10% < 1 year,

5% 1 to 5 years,

92.5% > 5 years

Cronan 1997

363

‐

‐

7.0 (5.5) years since diagnosis

Crotty 2009

152

32.9% hip

Height 167.2 (10.1) cm

Weight 85.8 (21.0) kg

‐

Hansson 2010

114

4.5% hip, 34% knee, 32% hand, 29% more locations

35% BMI 20 to 25,

38.5% BMI 25 to 30,

27% BMI > 30

‐

Heuts 2005

297

‐

28.1 (5.0)

‐

Hopman‐Rock 2000

120

‐

27.6 (4.3)

3% < 1 year,

20.5% 1 to 3 years,

33.5% 3 to 10 years,

18% 10 to 20 years,

17% > 20 years

Hughes 2004

215

‐

‐

‐

Hurley 2007

431

100% knee

30.2 (range 18 to 51)

6 (IQR 3 to 13) years

Jessep 2009

64

100% knee

29.5 (range 1 to 47)

12.5 (range 0.5 to 55) years

Keefe 1990

99

100% knee

24.2 (23.6)% above ideal weight

‐

Keefe 1996

88

100% knee

‐

‐

Keefe 2004

72

100% knee

‐

‐

Lorig 2008

551

‐

‐

‐

Maisiak 1996

405

‐

‐

16.0 years

Martire 2007

242

Hip and knee

‐

15.1 (10.9) years

Maurer 1999

113

100% knee

Weight 187.1 (33.9) lb

11.4 (10.5) years

Mazzuca 1997

211

100% knee

‐

14.0 (15.9) years

Mazzuca 2004

186

100% knee

‐

‐

McKnight 2010

273

100% knee

27.7 (4.2)

‐

Murphy 2008

54

67% knee, 11% hip, 22% hip and knee

30.1 (5.7)

‐

Nunez 2006

100

100% knee

‐

11.9 (10.6) months

Ackerman 2012

120

31% hip, 62.5% knee, 6.5% hip and knee

24.5 (IQR 25 to 35)

‐

Victor 2005

193

100% knee

‐

55% > 3 years

Wetzels 2005

104

53.8% knee, 20.1% hip, 26% hip and knee

‐

‐

Yip 2007

182

100% knee

‐

8.1 (6.8) years

Mean (SD), unless indicated otherwise.
No = Number of participants randomly assigned (total of all groups).
y = year(s).
‐ = no information available.
IQR = interquartile range.
BMI in kg/m2.

Intervention

Although self‐management education programmes differed in mode (individual or group), personnel (healthcare professionals or trained facilitators), delivery method (face‐to‐face, telephone, Internet) and duration, all were considered to include an element of self‐management (see Table 3). Across 29 studies, 34 self‐management education interventions were assessed. Twenty‐two of these interventions were group sessions, eight were individual sessions and four involved a combination of group and individual sessions. Most programmes were delivered face‐to‐face (25 interventions), two were provided over the telephone and one was delivered over the Internet. The remaining six interventions delivered a combination of face‐to‐face and telephone sessions. The total duration of the programmes ranged from four weeks to 12 months, although most programmes lasted six weeks. The frequency of sessions ranged from four per week to one per month, and most programmes were delivered on a weekly basis.

Open in table viewer
Table 3. Summary of comparisons

Study

Country

Setting

Intervention (N)

Mode/Personnel/Delivery method/Duration

Attention control (N)

Mode/Personnel/Delivery method/Duration

Allen 2010

US

Primary care

Self‐management intervention (174)

M: individual

P: health professionals

De: telephone

Du: 12 calls in 12 months

Health education (175)

M: individual

P: health professionals

De: telephone

Du: 12 calls in 12 months

Calfas 1992

US

Outpatients

Cognitive‐behaviour modification (20)

M: group

P: trained facilitator

De: face‐to‐face

Du: 10 weekly sessions

Traditional education intervention (20)

M: group

P: health specialists

De: face‐to‐face

Du: several lectures

Keefe 1990

US

Outpatients

Pain coping skills training (32)

M: group

P: nurse and psychologist

De: face‐to‐face

Du: 10 sessions in 10 weeks (1.5 hours)

and three phone calls

Arthritis education (36)

M: group

P: nurse and psychologist

De: lectures and telephone

Du: 10 sessions in 10 weeks (1.5 hours) and three phone calls

Maisiak 1996

US

Primary care and outpatients

Treatment counselling (135)

M: individual

P: trained counsellors

De: telephone

Du: 11 sessions in nine months

Symptom monitoring (135)

M: individual

P: college students trained for two hours

De: telephone

Du: 11 sessions in nine months

Mazzuca 1997

US

Primary care

Education (105)

M: individual

P: arthritis nurse educator

De: face‐to‐face/telephone

Du: one interview/two calls in one month

Attention control (106)

M: group and individual

P: not specified

De: audiovisual presentation and telephone

Du: 20‐minute presentation, two phone calls (five to 10 minutes)

Study

Setting

Intervention (N)

Information only (N)

Buszewicz 2006

UK

Primary care

Self‐management programme (406)

M: group

P: trained volunteer

De: face‐to‐face

Du: six sessions in six weeks

Education (406)

M: individual

P: none

De: education booklet

Du: ‐

Hughes 2004

US

General population

Fit & Strong (115)

M: group

P: physical therapist

De: face‐to‐face

Du: 24 sessions in eight weeks

Education (100)

M: individual

P: none

De: Arthritis Helpbook and handouts

Du: ‐

Ackerman 2012

Australia

Outpatients

Arthritis self‐management programme (58)

M: group

P: peer leader and health professional

De: face‐to‐face

Du: six sessions in six weeks

Education only (62)

M: individual

P: none

De: Arthritis Helpbook

Du: ‐

Wetzels 2005

The Netherlands

Primary care

Self‐management intervention (51)

M: individual

P: nurse

De: face‐to‐face/telephone

Du: one session/one call

Education (53)

M: individual

P: none

De: educational booklet

Du: ‐

Study

Setting

Intervention (N)

Usual care/Waiting list/No treatment (N)

Allen 2010

US

Primary care

Self‐management intervention (174)

M: individual

P: health professionals

De: telephone

Du: 12 calls in 12 months

Usual care (174)

 

Blixen 2004

US

Outpatients

Telephone health education strategy (16)

M: individual

P: advanced practice nurse

De: telephone

Du: six sessions in six weeks

Usual care (16)

 

Cronan 1997

US

General population

Education group (97)

M: group

P: professional health educator

De: face‐to‐face

Du: 10 weekly sessions and 10 monthly sessions

No treatment (90)

 

Cronan 1997

US

General population

Combination education + social support (89)

M: group

P: professional health educator (first hour)

De: face‐to‐face

Du: 10 weekly sessions and 10 monthly sessions

No treatment (90)

Crotty 2009

Australia

Outpatients

Patient education (75)

M: individual

P: peer support volunteer/research nurse

De: telephone/face‐to‐face

Du: six calls in six months/one interview

Usual care (77)

Hansson 2010

Sweden

Primary care

PEPOA (61)

M: group

P: health professionals

De: face‐to‐face

Du: five sessions in five weeks

Usual care (53)

Heuts 2005

The Netherlands

Primary care

Self‐management programme (149)

M: group

P: trained physiotherapists

De: face‐to‐face

Du: six sessions

Usual care (148)

Hopman‐Rock 2000

The Netherlands

General population

Health educational and exercise programme (60)

M: group

P: peer educator and health professionals

De: face‐to‐face

Du: six sessions in six weeks

No treatment (60)

Hurley 2007

UK

Primary care

Individual rehabilitation (146)

M: individual

P: physiotherapist

De: face‐to‐face

Du: 12 sessions in six weeks

Usual care (140)

Hurley 2007

UK

Primary care

Group rehabilitation (132)

M: group

P: physiotherapist

De: face‐to‐face

Du: 12 sessions in six weeks

Usual care (140)

Keefe 1990

US

Outpatients

Pain coping skills training (32)

M: group

P: nurse and psychologist

De: face‐to‐face

Du: 10 sessions in 10 weeks (1.5 hours)

and three phone calls

Usual care (31)

Keefe 2004

US

Outpatients and general population

Spouse‐assisted coping skills training (18)

M: group

P: trained psychologist

De: face‐to‐face

Du: 12 sessions in 12 weeks

Usual care (18)

Keefe 2004

US

Outpatients and general population

Spouse‐assisted coping skills training + exercise (20)

M: group

P: trained psychologist/exercise trainer

De: face‐to‐face

Du: 38 sessions in 12 weeks

Usual care (18)

Lorig 2008

US

General population

Internet‐based arthritis self‐management programme (433)

M: individual

P: peer moderators

De: Internet

Du: 18 sessions in six weeks (at least)

Usual care (422)

Maisiak 1996

US

Primary care and outpatients

Treatment counselling (135)

M: individual

P: trained counsellors

De: telephone

Du: 11 sessions in nine months

Usual care (135)

Martire 2007

US

Outpatients

Patient‐oriented education and support (89)

M: group

P: trained facilitator

De: face‐to‐face

Du: six sessions in six weeks

Usual care (54)

Martire 2007

US

Outpatients

Couple‐oriented education and support (99)

M: group

P: trained facilitator

De: face‐to‐face

Du: six sessions in six weeks

Usual care (54)

Mazzuca 2004

US

Primary care

Nurse‐directed intervention (111)

M: individual

P: arthritis nurse educator

De: face‐to‐face/telephone

Du: one interview/five to nine calls inthree to five months

Waiting list (75)

Nunez 2006

US

Outpatients

Self‐management programme (51)

M: group/individual

P: trained health educator

De: face‐to‐face

Du: four sessions in three months

Usual care (49)

Victor 2005

UK

Primary care

Self‐management (120)

M: group

P: research nurse

De: face‐to‐face

Du: four sessions

Waiting list (73)

Yip 2007

China

Outpatients

Arthritis self‐management programme (88)

M: group

P: nurses

De: face‐to‐face

Du: six sessions in six weeks

Usual care (94)

Study

Setting

Intervention (N)

Alternate intervention (N)

Berman 2004

US

General population

Education (189)

M: group

P: patient education specialist

De: face‐to‐face

Du: six sessions in 12 weeks

True/sham acupuncture (190 + 191)

M: individual

P: acupuncturists

De: face‐to‐face

Du: 23 treatments in 26 weeks

Cronan 1997

US

General population

Education group (97)

M: group

P: professional health educator

De: face‐to‐face

Du: 10 weekly sessions and 10 monthly sessions (two hours)

Social support group (87)

M: group

P: none

De: face‐to‐face

Du: 10 weekly sessions and 10 monthly sessions (two hours)

Cronan 1997

US

General population

Combination education + social support (89)

M: group

P: professional health educator (first hour)

De: face‐to‐face

Du: 10 weekly sessions and 10 monthly sessions (two hours)

Social support group (87)

M: group

P: none

De: face‐to‐face

Du: 10 weekly sessions and 10 monthly sessions (two hours)

Jessep 2009

UK

Outpatients

ESCAPE‐knee pain (29)

M: group

P: physiotherapist

De: face‐to‐face

Du: 10 sessions in five weeks

Physiotherapy (35)

M: not specified

P: physiotherapist

De: face‐to‐face

Du: maximum of 10 sessions

Keefe 1996

US

Not specified

Coping skills training (29)

M: group

P: nurse and psychologist

De: face‐to‐face

Du: 10 sessions in 10 weeks (two hours)

Arthritis education—spousal support (29)

M: group

P: nurse and psychologist

De: discussions and educational material

Du: one time a week (two hours) for 10 weeks

Keefe 1996

US

Not specified

Spouse‐assisted coping skills training (30)

M: group

P: nurse and psychologist

De: face‐to‐face

Du: 10 sessions in 10 weeks (two hours)

Arthritis education—spousal support (29)

M: group

P: nurse and psychologist

De: discussions and educational material

Du: one time a week (two hours) for 10 weeks

Keefe 2004

US

Outpatients and general population

Spouse‐assisted coping skills training (18)

M: group

P: trained psychologist

De: face‐to‐face

Du: 12 sessions in 12 weeks

Exercise (16)

M: group

P: exercise physiologists (BA level or above)

De: face‐to‐face

Du: three times a week (one hour) for 12 weeks

Keefe 2004

US

Outpatients and general population

Spouse‐assisted coping skills training + exercise (20)

M: group

P: trained psychologist/exercise trainer

De: face‐to‐face

Du: 38 sessions in 12 weeks

Exercise (16)

M: group

P: exercise physiologists (BA level or above)

De: face‐to‐face

Du: three times a week (one hour) for 12 weeks

Maurer 1999

US

Outpatients

Education (56)

M: group

P: healthcare professionals

De: face‐to‐face

Du: four sessions in eight weeks

Exercise (57)

M: not specified

P: not specified

De: face‐to‐face

Du: three times a week for eight weeks

McKnight 2010

US

General population

Self‐management group (87)

M: group/individual

P: healthcare professionals

De: face‐to‐face/telephone

Du: 12 sessions in 12 weeks/staggering calls

Strength training (91)

M: not specified

P: expert physical trainers

De: face‐to‐face

Du: three weekly sessions for nine months (phase 1), then contact every two weeks in first six weeks, then monthly for a total of 15 months (phase 2)

McKnight 2010

US

General population

Combination group (95)

M: group/individual

P: healthcare professionals and physiotherapists

De: face‐to‐face/telephone

Du: 48 sessions in 12 weeks/staggering calls

Strength training (91)

M: not specified

P: expert physical trainers

De: face‐to‐face, telephone

Du: three weekly sessions for nine months (phase 1), then contact every two weeks in first six weeks, then monthly for a total of 15 months (phase 2)

Murphy 2008

US

Senior housing facilities

Exercise + activity strategy training (28)

M: group + one individual session

P: occupational therapists

De: face‐to‐face

Du: eight sessions (1.5 hours) in four weeks

Exercise + education (26)

M: group

P: health education interventionists

De: educational materials

Du: two sessions per week (1.5 hours) for four weeks

M = Mode, P = Personnel, De = Delivery method, Du = Duration.

A description of the components of the programmes based on the heiQ framework is shown in Table 4. The mean number of heiQ items included in individual programmes was 4.4 (out of a possible eight components). Most programmes included skill and technique acquisition (32 programmes, 94%), health‐directed activity (29 programmes, 85%) and self‐monitoring and insight (27 programmes, 79%). The least included component was social integration and support (four programmes, 12%).

Open in table viewer
Table 4. heiQ‐components addressed in interventions

Intervention

Health‐directed activity

Positive and active engagement in life

Emotional distress

Self‐monitoring and insight

Constructive attitudes and approaches

Skill and technique acquisition

Social integration and support

Health service navigation

TOTAL + (%)

29/34

85%

9/34

26%

13/34

38%

27/34

79%

15/34

44%

32/34

94%

4/34

12%

11/34

32%

Allen 2010

+

‐

‐

+

+

+

‐

+

Berman 2004

+

U

+

‐

+

+

‐

+

Blixen 2004

+

‐

+

+

‐

+

‐

+

Buszewicz 2006

+

+

+

+

+

+

U

+

Calfas 1992

+

+

+

+

+

+

‐

‐

Cronan 1997 (education)

+

‐

‐

+

+

+

‐

+

Cronan 1997 (combination)

+

‐

‐

+

+

+

‐

+

Crotty 2009

‐

‐

‐

+

+

‐

‐

‐

Hansson 2010

+

‐

‐

+

‐

+

‐

‐

Heuts 2005

+

‐

+

+

‐

+

U

+

Hopman‐Rock 2000

+

‐

‐

+

+

+

‐

‐

Hughes 2004

+

‐

‐

‐

‐

+

‐

‐

Hurley 2007 (Indiv Rehab)

+

+

+

+

+

+

‐

‐

Hurley 2007 (Group Rehab)

+

+

+

+

+

+

‐

‐

Jessep 2009

+

+

+

+

+

+

‐

‐

Keefe 2004

‐

‐

‐

+

+

+

+

‐

Keefe 1996 (CST)

‐

‐

‐

+

+

+

‐

‐

Keefe 1996 (SA‐CST)

‐

‐

‐

+

+

+

+

‐

Keefe 1990

‐

‐

‐

+

+

+

‐

‐

Lorig 2008

+

+

+

+

+

+

‐

+

Maisiak 1996

+

‐

‐

+

‐

+

‐

+

Martire 2007 (PES)

+

+

+

+

+

+

‐

‐

Martire 2007 (CES)

+

+

+

+

+

+

+

‐

Maurer 1999

+

‐

‐

+

‐

+

‐

‐

Mazzuca 1997

+

‐

‐

‐

‐

+

‐

‐

Mazzuca 2004

+

‐

‐

‐

‐

+

‐

‐

McKnight 2010 (SMP)

+

‐

‐

+

+

+

‐

‐

McKnight 2010 (combination)

+

‐

‐

+

+

+

‐

‐

Murphy 2008

+

‐

‐

‐

+

+

‐

‐

Nunez 2006

+

‐

‐

‐

‐

+

‐

‐

Ackerman 2012

+

‐

+

+

+

+

‐

+

Victor 2005

+

‐

‐

‐

+

+

+

‐

Wetzels 2005

+

+

‐

+

+

‐

‐

‐

Yip 2007

+

‐

+

+

+

+

‐

+

+ = heiQ component addressed, ‐ = heiQ component not addressed, U = Unclear whether heiQ component was addressed

Comparator

Self‐management education programmes were compared with attention control (five studies), usual care (17 studies), information‐only (four studies) or alternative interventions that did not include self‐management (seven studies incorporating interventions such as exercise, physiotherapy or social support). All of these studies considered the self‐management education programme to be an active intervention. In contrast, one additional study included self‐management education as an inactive control group and compared it with both acupuncture as the active intervention and sham acupuncture (Berman 2004).

Outcomes

A limited number of studies comparing self‐management programmes versus usual care or information only reported the main outcomes prespecified in the protocol for this review. The main outcome of positive and active engagement in life was not reported in any trials comparing self‐management versus an attention control or alternative intervention.

For one study, the only data that could be extracted consisted of information regarding withdrawals (Calfas 1992).

Excluded studies

One hundred twenty‐two studies assessed in full text were excluded. Ninety‐nine were judged irrelevant, and 23 studies were excluded according to reasons provided in Characteristics of excluded studies. Studies were excluded if they involved mixed arthritis populations without subgroup data available for people with OA (12 studies; Barlow 2000; Ehrlich‐Jones 2001; Goeppinger 1989; Laforest 2008; Laforest 2008a; Lindroth 1989; Lorig 1985; Lorig 1999a; Lorig 1999b; Lorig 2005; Nour 2006; Solomon 2002), if they compared two different self‐management education programmes (six studies; Coleman 2010; Hoogeboom 2010; Lorig 1998; Martire 2003a; Martire 2008; Murphy 2010) or if the intervention was judged as failing to fulfil our inclusion criteria for a self‐management programme (five studies; Bezalel 2010; Ettinger 1997; Fernandes 2009; Fernandes 2010; Focht 2005).

Risk of bias in included studies

The risk of bias was assessed for each study (see Characteristics of included studies), and the results are summarised in Figure 2 and Figure 3.

Figure 2
Summary of the risk of bias across all included studies.

Summary of the risk of bias across all included studies.

Figure 3
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Allocation

Seven studies described adequate sequence generation and allocation concealment and were assessed as being at low risk of selection bias (Ackerman 2012; Allen 2010; Berman 2004; Buszewicz 2006; Crotty 2009; Heuts 2005; Wetzels 2005). Nine trials had adequate random sequence generation, but the risk of bias in the concealment of allocation was unclear (Hansson 2010; Hurley 2007; Keefe 1990; Maisiak 1996; Maurer 1999; McKnight 2010; Murphy 2008; Nunez 2006; Yip 2007). Three trials were assessed as adequately concealing allocation but as having high risk (Martire 2007; Victor 2005) or unclear risk (Jessep 2009) of bias in random sequence generation. Mazzuca 1997 was assessed as having inadequate random sequence generation and unclear risk of bias in allocation concealment. The remaining nine trials were assessed as having unclear risk of selection bias, based on both random sequence generation and allocation concealment.

Blinding

It is not possible to blind personnel delivering self‐management programmes, and all participants were aware of the treatment they were receiving; therefore all studies were at high risk of performance bias and dectection bias (for self‐reported outcomes). Of 14 studies that included assessor‐reported outcomes, eight were assessed as low risk, as the outcome assessors were blinded (Berman 2004; Hansson 2010; Hopman‐Rock 2000; Jessep 2009; Maurer 1999; Murphy 2008; Victor 2005; Wetzels 2005), four were assessed as unclear risk (Cronan 1997; Hurley 2007; Keefe 2004; Yip 2007) and two were assessed as high risk for detection bias (Hughes 2004; McKnight 2010).

Incomplete outcome data

Thirteen trials were assessed as low risk for attrition bias (Blixen 2004; Crotty 2009; Hansson 2010; Heuts 2005; Hopman‐Rock 2000; Hurley 2007; Keefe 1990; Keefe 2004; Maisiak 1996; Maurer 1999; McKnight 2010; Murphy 2008; Wetzels 2005). The remaining 16 studies had unexplained incomplete outcome data and were judged as being at high risk of attrition bias.

Selective reporting

Twenty studies were assessed as being at low risk for reporting bias. Seven trials were assessed as having an unclear risk (Allen 2010; Calfas 1992; Keefe 1990; Maisiak 1996; Martire 2007; Murphy 2008; Victor 2005), as the impact of minor outcomes that were not reported and/or the impact of not reporting outcomes at all time points was unclear. The remaining two studies (Cronan 1997; McKnight 2010) were assessed as being at high risk for reporting bias because they failed to report the primary outcome for this review.

Other potential sources of bias

All three cluster‐RCTs accounted for the effects of clustering in the analysis of their results (Hurley 2007; Mazzuca 2004; Victor 2005).

In 12 trials (Ackerman 2012; Allen 2010; Blixen 2004; Buszewicz 2006; Cronan 1997; Hurley 2007; Keefe 2004; Martire 2007; Mazzuca 2004; McKnight 2010; Wetzels 2005; Yip 2007), another potential source of bias was identified. These included baseline differences between groups (Keefe 2004; Martire 2007), low adherence to the intervention (Buszewicz 2006; Cronan 1997; McKnight 2010), differences in adherence between groups (Hurley 2007), risk of contamination between groups (Mazzuca 2004) and cultural bias in self‐reported measures (Yip 2007).

Effects of interventions

See: Summary of findings for the main comparison SMP compared to Attention control for osteoarthritis; Summary of findings 2 SMP compared with usual care for osteoarthritis; Summary of findings 3 SMP compared with information only for osteoarthritis; Summary of findings 4 SMP compared with non‐SMP intervention for osteoarthritis

Comparison 1. Self‐management programmes versus attention control

Main outcomes

One trial with 344 participants (Allen 2010) found no difference between a self‐management education programme and attention control in terms of improving self‐management skills after 12 months (mean difference (MD) 0.40, 95% confidence interval (CI) ‐0.39 to 1.19) (Analysis 1.1).

Positive and active engagement in life was not reported by any of the trials comparing self‐management versus attention control.

One trial (Keefe 1990) with 68 participants found greater pain reduction post‐treatment in the self‐management intervention group than in the attention control group (SMD ‐0.62, 95% CI ‐1.11 to ‐0.13)—equivalent to a mean difference of 1.55 cm (95% CI 0.33 to 2.78 cm) on a 10‐cm VAS (Analysis 1.2). Three trials with 577 participants (Allen 2010; Keefe 1990; Mazzuca 1997) assessed pain at longer time points (six to 12 months); pooled analysis also found greater pain reduction in the self‐management intervention group than in the attention control group (SMD ‐0.26, 95% CI ‐0.44 to ‐0.09) (Analysis 1.2). This is equivalent to a mean difference of 0.65 cm (95% CI 0.23 to 1.10 cm) on a 10‐cm VAS—a finding that is unlikely to be of clinical significance.

No overall between‐group differences were noted in the other main efficacy outcomes: global OA scores (one trial, SMD ‐0.14, 95% CI ‐0.54 to 0.26), self‐reported function in the short term (one trial, SMD ‐0.13, 95% CI ‐0.49 to 0.23) and in the intermediate term (three trials, SMD ‐0.19, 95% CI ‐0.50 to 0.11) and quality of life (one trial, SMD ‐0.01, 95% CI ‐0.03 to 0.01) (Analysis 1.3; Analysis 1.4; Analysis 1.5).

In five studies with 937 participants (Allen 2010; Calfas 1992; Keefe 1990; Maisiak 1996; Mazzuca 1997), no difference was observed in the rate of withdrawals between self‐management (61/446, 14%) and attention control (55/471, 12%) groups (RR 1.11, 95% CI 0.78 to 1.57) (Analysis 1.6).

Other outcomes

No overall between‐group differences in emotional distress were noted in the short term (one trial, SMD ‐0.37, 95% CI ‐0.85 to 0.11) and in the intermediate term (two trials, SMD 0.02, 95% CI ‐0.18 to 0.21) (Analysis 1.7).

Functional performance, health‐directed activity, social integration and support, health service navigation, skill and technique acquisition and constructive attitudes and approaches were not reported by any of the trials in this comparison.

Comparison 2. Self‐management programmes versus usual care

Main outcomes

Five trials with 721 participants (Hopman‐Rock 2000; Keefe 2004; Martire 2007; Victor 2005; Yip 2007) reported no difference between a self‐management programme and usual care in terms of self‐management skills within one month post‐treatment (SMD 0.22, 95% CI ‐0.00 to 0.45) (Analysis 2.1); this is equivalent to a mean difference of 0.11 points (95% CI ‐0.00 to 0.23) on the 1 to 6‐point (higher score is better) heiQ subscale self‐monitoring and insight. However, moderate statistical heterogeneity was present (I2 = 48%) because findings of one trial were not consistent (Victor 2005) with findings of the other four trials. Victor 2005 was a cluster‐RCT with more withdrawals in the intervention than in the control group (40% vs 27%) and with high risk of selection, reporting and attrition biases.

Results of 10 trials with 1,647 participants (Allen 2010; Blixen 2004; Crotty 2009; Hansson 2010; Heuts 2005; Hopman‐Rock 2000; Lorig 2008; Martire 2007; Victor 2005; Yip 2007) show no difference between a self‐management programme and usual care in improving self‐management skills from three to 12 months (SMD 0.14, 95% CI 0.00 to 0.27) (Analysis 2.1); this is equivalent to a mean difference of 0.07 points (95% CI 0.00 to 0.14) on the heiQ subscale self‐monitoring and insight (1 to 6 scale, higher score is better). Moderate statistical heterogeneity was present (I2 = 41%) because the results of Victor 2005 were not consistent with the findings of the other nine trials. One trial with 195 participants (Heuts 2005) assessed self‐management skills over 21 months and found no difference between participants in a self‐management programme and those receiving usual care (SMD 0.23, 95% CI ‐0.05 to 0.51) (Analysis 2.1); this finding is equivalent to a mean difference of 0.12 (95% CI ‐0.03 to 0.26) on the heiQ subscale self‐monitoring and insight (1 to 6 scale, higher score is better). The impact of the failure of Cronan 1997 to report self‐management skills is likely to be small.

One trial with 143 participants (Victor 2005) found no difference between a self‐management programme and usual care in terms of participants' positive and active engagement in life after one month (SMD ‐0.23, 95% CI ‐0.57 to 0.11) (Analysis 2.2); this is equivalent to a mean difference of ‐0.18 (95% CI ‐0.46 to 0.09) on the heiQ subscale for positive and active engagement in life (1 to 6 scale, higher score is better). Three trials with 357 participants (Crotty 2009; Nunez 2006; Victor 2005) also showed no difference at six to 12 months (SMD 0.01, 95% CI ‐0.20 to 0.21); this is equivalent to a mean difference of 0.01 (95% CI ‐0.16 to 0.17) on the heiQ subscale positive and active engagement in life (1 to 6 scale, higher score is better).

Findings of six trials with 766 participants (Hopman‐Rock 2000; Keefe 1990; Keefe 2004; Martire 2007; Victor 2005; Yip 2007) indicate that pain improved significantly in the self‐management intervention group compared with the usual care group in the time frame post‐treatment to one month (SMD ‐0.26, 95% CI ‐0.41 to ‐0.10) (Analysis 2.3); this is equivalent to a mean difference of 0.65 cm (95% CI 0.25 to 1.00 cm) on a 10‐cm VAS, which is unlikely to be of clinical significance. In 13 trials with 7,447 participants (Allen 2010; Blixen 2004; Crotty 2009; Heuts 2005; Hopman‐Rock 2000; Hurley 2007; Keefe 1990; Lorig 2008; Martire 2007; Mazzuca 2004; Nunez 2006; Victor 2005; Yip 2007), pain improved significantly in the self‐management group in the time frame three to 12 months (SMD ‐0.17, 95% CI ‐0.26 to ‐0.08); this is equivalent to a mean difference of 0.43 cm (95% CI 0.2 to 0.65 cm) on a 10‐cm VAS—a finding also unlikely to be of clinical significance. One trial with 213 participants (Heuts 2005) found no difference in pain between the self‐management programme and usual care after 21 months (SMD ‐0.18, 95% CI ‐0.45 to 0.09).

The findings of two studies with 319 participants (Martire 2007; Yip 2007) show that global OA scores improved significantly in the self‐management intervention compared with the usual care group up to one week post‐treatment (SMD ‐0.34, 95% CI ‐0.59 to ‐0.09). This result could be clinically meaningful (MD 0.71 points, 95% CI 0.19 to 1.23 points on the WOMAC) (Analysis 2.4). In seven studies with 1,351 participants (Heuts 2005; Hurley 2007; Lorig 2008; Maisiak 1996; Martire 2007; Nunez 2006; Yip 2007), global OA scores improved in the self‐management education programme compared with usual care between three and 12 months (SMD ‐0.28, 95% CI ‐0.39 to ‐0.17) (Analysis 2.4). However, this finding is unlikely to be of clinical significance (MD 0.59 points, 95% CI 0.36 to 0.82 points, on the WOMAC; 0 to 96‐point scale, lower score is better). One study with 197 participants (Heuts 2005) reported that global OA scores improved significantly in the self‐management education programme compared with the usual care group after 21 months (SMD ‐0.29, 95% CI ‐0.56 to ‐0.02) (Analysis 2.4), but this finding is unlikely to be of clinical significance (MD 0.61, 95% CI 0.04 to ‐1.17 points on the 0 to 96‐point WOMAC).

Findings of five studies with 714 participants (Hopman‐Rock 2000; Keefe 1990; Martire 2007; Victor 2005; Yip 2007) show no difference in self‐reported function between participants in self‐management programmes and those provided with usual care up to one month post‐treatment (SMD ‐0.01, 95% ‐0.19 to 0.18) (Analysis 2.5). Thirteen studies with 2,176 participants (Allen 2010; Blixen 2004; Crotty 2009; Heuts 2005; Hopman‐Rock 2000; Hurley 2007; Keefe 1990; Lorig 2008; Martire 2007; Mazzuca 2004; Nunez 2006; Victor 2005; Yip 2007) showed that self‐reported function improved significantly in self‐management programmes compared with usual care between three and 12 months (SMD ‐0.16, 95% CI ‐0.25 to ‐0.08). However, this finding is unlikely to be of clinical significance (MD 0.35, 95% CI 0.17 to ‐0.55 points on the 0 to 68‐point WOMAC function subscale) (Analysis 2.5). One study with 199 participants (Heuts 2005) showed no difference in self‐reported function between participants in self‐management education programmes and usual care after 21 months (SMD ‐0.27, 95% CI ‐0.55 to 0.01) (Analysis 2.5). Similarly, no between‐group differences were seen in performance measures of function in the short term (one trial, SMD 0.33, 95% CI ‐0.07 to 0.73) and in the intermediate term (two trials, SMD 0.06, 95% CI ‐0.24 to 0.36) (Analysis 2.6).

No overall between‐group difference in quality of life was noted at any time point (two trials assessed up to six weeks post‐treatment: SMD 0.14, 95% CI ‐0.47 to 0.75; eight trials assessed up to one year: SMD 0.03, 95% CI ‐0.08 to 0.14; and two studies assessed beyond one year: SMD 0.10, 95% CI ‐0.10 to 0.31) (Analysis 2.7).

Findings of 16 trials with 3,738 participants (Allen 2010; Blixen 2004; Cronan 1997; Hansson 2010; Heuts 2005; Hopman‐Rock 2000; Hurley 2007; Keefe 1990; Keefe 2004; Lorig 2008; Maisiak 1996; Martire 2007; Mazzuca 2004; Nunez 2006; Victor 2005; Yip 2007) show that no differences in numbers of withdrawals were found between groups (RR 0.99, 95% CI 0.74 to 1.33) (Analysis 2.8).

Other outcomes

Three studies with 262 participants (Keefe 1990; Keefe 2004; Victor 2005) reported no difference in emotional distress between self‐management programmes and usual care in the time frame post‐treatment and one month (SMD 0.01, 95% CI ‐0.44 to 0.45), although considerable statistical heterogeneity (I2 = 64%) was observed (Analysis 2.9). Eight studies with 1,427 participants (Allen 2010; Blixen 2004; Crotty 2009; Hurley 2007; Keefe 1990; Lorig 2008; Nunez 2006; Victor 2005) reported no difference in terms of emotional distress between self‐management programmes and usual care between six and 12 months (SMD 0.11, 95% CI ‐0.06, 0.28), although moderate statistical heterogeneity was present (I2 = 48%).

One study with 182 participants (Yip 2007) indicated that self‐management programmes improved health‐directed activity compared with usual care after one week (SMD 0.67, 95% CI 0.37 to 0.97) (Analysis 2.10). This result could be clinically meaningful (MD 0.64, 95% CI 0.35‐0.92 points on the heiQ 1 to 6‐point subscale health‐directed activities). Findings of three studies with 626 participants (Crotty 2009; Lorig 2008; Yip 2007) indicate that self‐management education programmes improved health‐directed activity compared with usual care between six and 12 months (SMD 0.25, 95% CI 0.05 to 0.46). However, this finding is unlikely to be of clinical importance (MD 0.24, 95% CI 0.05 to 0.44 points on the heiQ 1 to 6‐point subscale health‐directed activities).

One study with 152 participants (Crotty 2009) reported that skill and technique acquisition improved significantly in participants in the self‐management intervention compared with usual care after six months (MD 0.26, 95% CI 0.01 to 0.51) (Analysis 2.11). However, this finding is unlikely to be of clinical importance (MD 0.22, 95% CI 0.01 to 0.43 points on the heiQ 1 to 6‐point subscale skill and technique acquisition).

Another study with 56 participants (Keefe 2004) reported that constructive attitudes and approaches improved significantly post‐treatment in participants in the self‐management intervention compared with usual care (SMD 1.04, 95% CI 0.44 to 1.63) (Analysis 2.12). This result could be clinically important (MD 0.83, 95% CI 0.35 to 1.30 points on the 1 to 6‐point heiQ subscale constructive attitudes and approaches). One study with 152 participants (Crotty 2009) showed no difference in terms of constructive attitudes and approaches between self‐management programmes and usual care after six months (SMD 0.11, 95% CI ‐0.20 to 0.43).

No between‐group differences were described for any of the other reported outcomes, including social integration and support in the short term (one trial, SMD ‐0.19, 95% CI ‐0.52 to 0.15) and in the intermediate term (three trials, SMD ‐0.08, 95% CI ‐0.30 to 0.14) (Analysis 2.13) and health service navigation (two trials, SMD 0.15, 95% CI ‐0.03 to 0.34) (Analysis 2.14).

Comparison 3. Self‐management programmes versus information only

Main outcomes

One study with 90 participants (Ackerman 2012) found no difference between a self‐management education programmes and information only in improving self‐management skills over six weeks (SMD 0.06, 95% CI ‐0.36 to 0.47) (Analysis 3.1); this is equivalent to a mean improvement of 0.03 (95% CI ‐0.18 to 0.24 cm) points with self‐management on the 1 to 6‐point heiQ subscale self‐monitoring and insight. Three trials with 760 participants (Ackerman 2012; Buszewicz 2006; Hughes 2004) showed no difference between a self‐management programme and information only in terms of improving self‐management skills over 12 months (SMD 0.20, 95% CI ‐0.04 to 0.44); this is equivalent to a mean improvement of 0.10 (95% CI ‐0.02 to 0.22) points on the 1 to 6‐point heiQ subscale self‐monitoring and insight, although moderate statistical heterogeneity was present (I2 = 41%). Hughes 2004 differed from the other two trials in that it had a higher withdrawal rate in the control group than in the intervention group (68% vs 50%), and the study incorporated a minimal number of components considered important for self‐management programmes (2/8 heiQ components compared with 6/8 (Ackerman 2012) and 7/8 (Buszewicz 2006)).

One study with 90 participants (Ackerman 2012) showed no difference between a self‐management programme and information only in improving participants' positive and active engagement in life after six weeks (MD ‐0.08, 95% CI ‐0.41 to 0.26) or 12 months (MD ‐0.20, 95% CI ‐0.59 to 0.18) on the 1 to 6‐point heiQ subscale positive and active engagement in life) (Analysis 3.2).

No overall between‐group differences were noted in the other main efficacy outcomes: pain (intermediate term, three trials, SMD ‐0.07, 95% CI ‐0.21 to 0.08), global OA (short term, one trial, SMD 0.09, 95% CI ‐0.33 to 0.50; and intermediate term, three trials, SMD ‐0.06, 95% CI ‐0.28 to 0.16), self‐reported function (four trials up to 12 months, SMD ‐0.09, 95% CI ‐0.22 to 0.05) and quality of life (intermediate term, two trials, SMD 0.05, 95% CI ‐0.10 to 0.21) (Analysis 3.3; Analysis 3.4; Analysis 3.5; Analysis 3.7).

Findings of four trials with 1,251 participants (Ackerman 2012; Buszewicz 2006; Hughes 2004; Wetzels 2005) showed no between‐group differences with respect to numbers of withdrawals (RR 1.60, 95% CI 0.75, 3.40) (Analysis 3.8). However, considerable statistical heterogeneity (I2 = 89%) was found to be due to larger differences in favour of the intervention in two small trials and small but statistically significant differences in opposite directions in the remaining two large trials.

Other outcomes

No differences between groups were noted for any of the other reported outcomes, including emotional distress (three trials up to 12 months, SMD 0.00, 95% CI ‐0.30 to 0.30) (Analysis 3.9); health‐directed activity (short term, one trial, MD 0.24, 95% CI ‐0.09 to 0.57; and intermediate term, one trial, MD 0.21, 95% CI ‐0.14 to 0.56) (Analysis 3.10); social integration and support (short term, one trial SMD ‐0.02, 95% CI ‐0.44 to 0.40; and intermediate term, two trials, SMD ‐0.02, 95% CI ‐0.39 to 0.35) (Analysis 3.11); health service navigation (short term, one trial, MD ‐0.04, 95% CI ‐0.33 to 0.26; and intermediate term, one trial, MD 0.14, 95% CI ‐0.13 to 0.41) (Analysis 3.12); skill and technique acquisition (short term, one trial, MD 0.15, 95% CI ‐0.13 to 0.44; and intermediate term, one trial, MD ‐0.06, 95% CI ‐0.35 to 0.23) (Analysis 3.13); and constructive attitudes and approaches (short term, one trial, MD 0.16, 95% CI ‐0.12 to 0.44; and intermediate term, one trial, MD ‐0.20, 95% CI ‐0.50 to 0.10) (Analysis 3.14).

Comparison 4. Self‐management programmes versus alternate interventions

Main outcomes

Findings of three trials with 186 participants (Keefe 1996; Keefe 2004; Murphy 2008) showed no difference between a self‐management programme and an alternate intervention in terms of improving self‐management skills immediately post‐treatment (SMD 0.42, 95% CI ‐0.05 to 0.89); this is equivalent to a mean difference of 0.21 (95% CI ‐0.03 to 0.45) on the heiQ subscale self‐monitoring and insight (Analysis 4.1), although moderate statistical heterogeneity was present (I2 = 56%). One trial with 81 participants (Keefe 1996) indicated that self‐management skills improved at 12 months for those attending a self‐management programme compared with arthritis education (SMD 0.54, 95% CI 0.05 to 1.04) (Analysis 4.1); this is equivalent to a mean improvement of 0.27 (95% CI 0.03 to 0.52) points on the 1 to 6‐point heiQ subscale self‐monitoring and insight. This result could be of clinical significance. The failure of McKnight 2010 to report self‐management skills is likely to have had little impact on these results.

Positive and active engagement in life was not reported by any of the trials comparing self‐management versus alternate interventions.

No overall between‐group differences were reported in terms of pain (short term, five trials, SMD 0.03, 95% CI ‐0.29 to 0.36; and intermediate term, two trials, SMD ‐0.18, 95% CI ‐0.56 to 0.19) (Analysis 4.2); global OA scores (one trial, SMD 0.27, 95% CI ‐0.13 to 0.67) (Analysis 4.3); self‐reported function (short term, three trials, SMD 0.23, 95% CI ‐0.03 to 0.48; and intermediate term, two trials, SMD ‐0.17, 95% CI ‐0.54 to 0.20) (Analysis 4.4); functional performance (one trial, SMD ‐0.09, 95% CI ‐0.64 to 0.46) (Analysis 4.5); and quality of life in the short term (one trial, SMD 0.24, 95% CI ‐0.28 to 0.76), the intermediate term (two trials, SMD ‐0.01, 95% CI ‐0.28 to 0.26) and the long term (one trial, SMD ‐0.23, 95% CI ‐0.55 to 0.10) (Analysis 4.6).

Seven trials with 880 participants (Cronan 1997; Jessep 2009; Keefe 1996; Keefe 2004; Maurer 1999; McKnight 2010; Murphy 2008) showed no differences in the numbers of withdrawals between groups (RR 0.86, 95% CI 0.69 to 1.09) (Analysis 4.7).

Other outcomes

No between‐group differences were noted for emotional distress in the short term (three trials, SMD 0.13, 95% CI ‐0.28 to 0.55) or in the intermediate term (two trials, SMD 0.18, 95% CI ‐0.19 to 0.55) (Analysis 4.8).

In two trials with 135 participants (Keefe 1996; Keefe 2004), constructive attitudes and approaches increased in participants in the self‐management intervention compared with an alternate intervention (arthritis education and exercise) immediately post‐treatment (SMD 0.92, 95% CI 0.49 to 1.34) (Analysis 4.9); this is equivalent to a mean improvement of 0.74 (95% CI 0.39 to 1.07) on the 1 to 6‐point heiQ subscale constructive attitudes and approaches, which could be of clinical significance. One trial with 81 participants (Keefe 1996) reported that constructive attitudes and approaches improved significantly in participants in the self‐management intervention group compared with those in an alternate intervention group after 12 months (SMD 0.62, 95% CI 0.12, 1.12); this is equivalent to a mean improvement of 0.50 (95% CI 0.10 to 0.90) points on the 1 to 6‐point heiQ subscale constructive attitudes and approaches. This result could be of clinical significance.

The other outcomes in this review, including health‐directed activity, social integration and support, health service navigation and skill and technique acquisition, were not reported by any of the trials comparing self‐management programmes versus alternative interventions.

Comparison 5. Self‐management programmes versus acupuncture

One study (Berman 2004) with 570 participants compared a self‐management programme versus acupuncture. This study was considered separately from the other studies comparing self‐management programmes versus alternative interventions (Comparison 4) because, unlike in the other trials, the comparison was not a behavioural intervention. This study was judged to be at high risk of bias because of lack of blinding and high attrition (see Characteristics of included studies), increasing the risk of bias in favour of acupuncture.

Main outcomes

Self‐management of OA, self‐monitoring and insight, positive and active engagement in life and quality of life were not reported in this trial.

No difference was found between treatment groups in global OA scores after four weeks (SMD 0.05, 95% CI ‐0.15 to 0.25) and after 26 weeks (SMD ‐0.10, 95% CI ‐0.32 to 0.12) (Analysis 5.1); these differences are equivalent to a decrease of ‐0.10 points (95% CI ‐0.52 to 0.31) on the WOMAC 96‐point scale after four weeks and ‐0.21 points (95% CI ‐0.68 to 0.25) after 26 weeks.

Pain improved more in the acupuncture group than in the self‐management group after four weeks (SMD 0.95, 95% CI 0.74 to 1.16) and after 26 weeks (SMD 1.37, 95% CI 1.13 to 1.61) (Analysis 5.2); this is equivalent to a mean reduction in pain of 2.38 cm (95% CI 1.85 to 2.90 cm) on a 10‐cm VAS after four weeks and a mean reduction of 3.43 cm (95% CI 2.83 to 4.03 cm) after 26 weeks and is likely to represent a clinically meaningful difference.

Self‐reported function increased in the acupuncture group compared with the self‐management education programme group after four weeks (SMD 1.22, 95% CI 1.00 to 1.44) and after 26 weeks (SMD 1.53, 95% CI 1.29 to 1.77) (Analysis 5.3); this is equivalent to a mean improvement in function of 2.67 (95% CI 2.19 to 3.15) points on the 68‐point WOMAC function scale after four weeks and a mean improvement of 3.35 (95% CI 2.83 to 3.88) points after 26 weeks—differences that are likely to be clinically meaningful. Functional performance increased in the acupuncture group compared with the self‐management group after 26 weeks (SMD ‐0.30, 95% CI ‐0.52 to ‐0.08) (Analysis 5.4); this is equivalent to a mean improvement of 97.8 (95% CI 26.1 to 169.5) feet on the six‐minute walk distance test and could show clinical significance.

Significantly more withdrawals were reported in the self‐management group than in the acupuncture group (RR 1.96, 95% CI 1.58, 2.42) (Analysis 5.5).

Other outcomes

The other outcomes in this review, including emotional distress, health‐directed activity, social integration and support, health service navigation, skill and technique acquisition and constructive attitudes and approaches, were not reported in this trial.

Sensitivity analysis

The results were robust to excluding trials that did not randomly allocate participants (results not shown). No trials were judged to be at high risk of bias for treatment allocation concealment. However, the results were robust to excluding trials with unclear treatment allocation concealment (results not shown).

Subgroup analyses

We had planned a subgroup analysis based on whether the study population consisted mainly of Caucasian, educated, older females. However, as the mean age of participants was 64.8 years across all studies, the subgroup analysis considered only whether the study population consisted mainly of Caucasian, educated females. Eight studies included predominantly Caucasian, educated females (Berman 2004; Cronan 1997; Hughes 2004; Lorig 2008; Maisiak 1996; Mazzuca 2004; McKnight 2010; Murphy 2008), and five studies did not (Allen 2010; Blixen 2004; Mazzuca 1997; Victor 2005; Yip 2007). One study (Berman 2004) was judged too different to be included in this analysis, and not enough information was available for review authors to determine subgroup status for the remaining 16 studies.

For participants' positive and active engagement in life, no subgroup analysis was possible, as not enough trials had assessed this outcome. Results for the other subgroup analyses varied by outcome. For self‐management in OA and self‐reported function, self‐management programmes appeared to be more beneficial in trials that primarily included Caucasian, educated females compared with trials that did not primarily include this subgroup (self‐management in OA: SMD 0.29, 95% CI 0.07 to 0.50 vs SMD 0.03, 95% CI ‐0.29 to 0.36; Analysis 6.1; self‐reported function: SMD ‐0.20, 95% CI ‐0.37 to ‐0.02 vs SMD ‐0.06, 95% CI ‐0.21, 0.08; Analysis 6.2). On the other hand, for self‐reported pain, self‐management programmes appeared more beneficial in trials that did not primarily include Caucasian, educated females (SMD ‐0.11, 95%CI ‐0.30 to 0.07 vs SMD ‐0.20, 95% CI ‐0.35 to ‐0.05) (Analysis 6.3). No difference in withdrawals was noted between subgroups (Analysis 6.4).

Because no strong treatment effects were indicated for self‐management programmes in the Results of this review, we did not conduct other subgroup analyses as described in the protocol (see Differences between protocol and review).

Funnel plots

The potential for small‐study effects was explored using funnel plots in outcomes with more than 10 studies; this included self‐management of OA of Analysis 3.1 (Figure 4), withdrawals of Analysis 3.3 (Figure 5) and pain of Analysis 3.4 (Figure 6). Based on the appearance of these funnel plots, we judged that the pooled results for these outcomes were not biased by small‐study effects.

Figure 4
Funnel plot of comparison: 3 SMP versus Usual care/No treatment/Wait list, outcome: 3.1 Self‐management of OA.

Funnel plot of comparison: 3 SMP versus Usual care/No treatment/Wait list, outcome: 3.1 Self‐management of OA.

Figure 5
Funnel plot of comparison: 3 SMP versus Usual care/No treatment/Wait list, outcome: 3.3 Dropouts.

Funnel plot of comparison: 3 SMP versus Usual care/No treatment/Wait list, outcome: 3.3 Dropouts.

Figure 6
Funnel plot of comparison: 3 SMP versus Usual care/No treatment/Wait list, outcome: 3.4 Pain.

Funnel plot of comparison: 3 SMP versus Usual care/No treatment/Wait list, outcome: 3.4 Pain.

Discussion

Summary of main results

This systematic review assessed the effects of self‐management education programmes for OA by employing rigorous and systematic methods of searching, appraising and synthesising the evidence. As overall outcome data presented in the Summary of findings tables are based on the longest time points measured in each study, some differences in effect estimates from results presented will be noted according to immediate, intermediate and longer‐term outcomes. Evidence of low to moderate quality indicated lack of benefit of self‐management education programmes in comparison with attention control, information‐only or alternate interventions. Evidence of low to moderate quality showed a small, statistically significant benefit of self‐management education programmes compared with usual care; however, these findings were unlikely to be of clinical importance.

Compared with attention control, evidence of low to moderate quality indicated that self‐management education programmes may not result in significant benefit at 12 months (summary of findings Table for the main comparison). Although a small difference in pain favoured self management education programmes (low‐quality evidence, three trials, 575 participants; SMD ‐0.26, 95% CI ‐0.44 to ‐0.09), this is unlikely to be of clinical importance, and no between‐group differences were observed for any of the other measured main outcomes (e.g. self‐management skills) (low‐quality evidence, one trial, N = 344; MD 0.4 points, 95% CI ‐0.39 to 1.19) or for withdrawal rates (moderate‐quality evidence, five trials, 937 participants; RR 1.11, 95% CI 0.78 to 1.57). No trial measured positive and active engagement in life.

Compared with usual care, moderate‐quality evidence (11 trials, N = 1706) suggested small but clinically unimportant benefits favouring self‐management education programmes up to 21 months (summary of findings Table 2). Differences favoured self‐management education programmes for self‐management skills (absolute improvement (SMD 3.78%, 95% CI 0.71% to 6.85%), pain (SMD ‐0.19, 95% CI ‐0.28 to ‐0.1), function (SMD 0.08, 95% CI 0.09 to 0.27) and global osteoarthritis symptoms (SMD ‐0.28, 95% CI ‐0.39 to ‐0.17), but no between‐group differences in quality of life (SMD 0.02, 95% CI ‐0.09 to 0.13) or in positive and active engagement in life (SMD 0.01, 95% CI ‐0.2 to 0.21) were noted. Evidence of low quality (16 trials, N = 3,738) suggested similar withdrawal rates (RR 0.99, 95% CI 0.74 to 1.33).

Low to moderate evidence from four studies (up to 1,251 participants in total) comparing self‐management education programmes with provision of information alone showed no differences in outcome in terms of self‐management of OA, positive and active engagement in life, pain, global OA scores, function, quality of life or study withdrawals (summary of findings Table 3).

Evidence of low to moderate quality obtained from seven studies (up to 919 participants in total) showed no differences in outcome with respect to self‐management of OA, positive and active engagement in life, pain, global OA scores, function, quality of life or study withdrawals for self‐management education programmes compared with alternative interventions (e.g. exercise, physiotherapy, social support, acupuncture) (summary of findings Table 4).

This review found several modest beneficial effects of self‐management education programmes although none of these was considered to be of clinical importance. Although we do not have data on patient expectations for self‐management education programs in OA, a recent survey of patients with chronic low back pain has shown that, on average, recipients of care need to see effect sizes (i.e. additional improvement to natural recovery) ranging from 20% to 30% on pain and disability with interventions such as exercise and non‐steroidal anti‐inflammatory drugs to consider them worthwhile (Ferreira 2011). It is therefore likely that individual patients would not consider any of the identified small beneficial effects of the studied self‐management education programmes worthwhile.

It is important to note that all of the efficacy outcomes that we considered were continuous outcome measures. However for chronic pain, patients may report a very good or a very poor response to treatment, making interpretation of average changes in continuous pain measures difficult (Moore 2010). Although average effect sizes in our review were small, we cannot exclude a bimodal (rather than Gaussian) distribution of response, whereby a proportion of participants may have derived a large, clinically relevant benefit of the self‐management intervention, although, by extension, another group would have had to derive a large, clinically relevant harm from the intervention—a scenario that is unlikely given that the intervention of interest was self‐management education programmes. However, although no significant difference in withdrawal rates was noted between self‐management education programmes and control groups, and no differences were evident between trials of short or longer duration of follow‐up, a high withdrawal rate across studies suggests that participant adherence to self‐management education programmes, as delivered in the trials included in this review, may be less than optimal.

Overall completeness and applicability of evidence

This review was limited to evidence provided from RCTs or quasi‐RCTs undertaken to assess the effects of self‐management education programmes compared with receipt of information only, no treatment, usual care, waiting list control or alternative interventions that are not considered self‐management education programmes.

Several trials that were potentially eligible for inclusion in this review were not included because they involved mixed populations with chronic disease or unspecified 'arthritis' (Barlow 2000; Ehrlich‐Jones 2001; Goeppinger 1989; Laforest 2008; Laforest 2008a; Lindroth 1989; Lorig 1985; Lorig 1999a; Lorig 1999b; Lorig 2005; Nour 2006; Solomon 2002). These trials failed to meet our prespecified criterion that a high proportion of included participants must have OA (90% or greater), or that outcomes for people with OA must be reported separately. Although this is an arbitrary threshold, the rationale for excluding studies based on this criterion was the presumed limited applicability of the results of these studies to people with OA.

This review was also limited to studies that compared self‐management education programmes versus no self‐management or an alternative intervention. We did not assess the superiority of one type of self‐management programme compared with another; this resulted in exclusion of other randomised trials (Coleman 2010; Hoogeboom 2010; Lorig 1998; Martire 2003a; Martire 2008; Murphy 2010).

The RCTs included in this review were generally of short duration, and limited outcome assessments were available at longer time points (i.e. longer than a year after completion of the programme); however, the lack of short‐term effects of self‐management programmes shown in the analysis of this review would indicate that any longer‐term effects are unlikely.

We expected that apart from pain, function, quality of life and global OA score, self‐management or self‐efficacy would be a main efficacy outcome in studies included in this review, as improvement in this outcome should be a major goal of a self‐management education programme. However, only 13 of 29 (45%) studies included in our review specified this as an outcome. Similarly, another main efficacy outcome, positive and active engagement in life, was not reported in any trials that compared self‐management versus attention control or an alternative intervention.

In addition, several main outcomes in the five comparisons that we undertook were reported by only one study (Analysis 1.1; Analysis 1.3; Analysis 1.5; Analysis 3.2; Analysis 3.6; Analysis 3.10; Analysis 3.12; Analysis 3.13; Analysis 3.14; Analysis 2.11; Analysis 4.3; Analysis 4.5). Although two studies were assessed as being at high risk for reporting bias because they purportedly measured but failed to report a primary outcome for this review (Cronan 1997; McKnight 2010), and an additional seven trials were assessed as having unclear risk of reporting bias, it is unlikely that this would have appreciably altered our results, as the direction of bias is likely to have been towards the null.

Only studies conducted by Keefe et al (Keefe 1990; Keefe 1996; Keefe 2004) used the CSQ (Coping Strategies Questionnaire) (see Analysis 2.12; Analysis 4.9). Although the validity of this questionnaire has been extensively tested for the French (Irachabal 2008) and German (Verra 2006) versions, validation studies for the original (American) version of the CSQ are scarce. The study performed by Rosentiel and Keefe (Rosentiel 1983) considered that the CSQ had good internal reliability; however, other studies have questioned its construct validity (Robinson 1997; Steward 2001 ), and Geisser 1994 concluded that individual CSQ subscales may have greater utility in terms of examining coping, appraisal and pain adjustment compared with the composite scores.

When the characteristics of study populations were assessed using the PROGRESS‐Plus framework, a predominance of older, Caucasian, educated females was seen in the included studies. This potentially raises questions regarding the applicability of results from this review to other groups. In investigating whether this combination of factors moderate outcome, we performed subgroup analyses that yielded conflicting results. Self‐management programmes appeared more beneficial for Caucasian, educated female participants with respect to self‐management of OA and self‐reported function, but for self‐reported pain, self‐management programmes appeared more beneficial in trials that did not primarily include this subgroup. Our data contrast with the findings of a systematic review of RCTs of self‐management interventions for chronic musculoskeletal pain performed to identify predictors, moderators and mediators of outcome (Miles 2011). Upon examining 16 RCTs involving 4,047 participants that included appropriate analyses of moderators and/or mediators, Miles 2011 found only tentative evidence to support age and gender as moderators of outcomes resulting from self‐management interventions because the data were insufficient. However, those review authors found strong evidence that self‐efficacy, depression, pain catastrophising and physical activity are all important influences on participant outcome, irrespective of treatment, although evidence suggested that pain catastrophising and physical activity can mediate outcomes resulting from self‐management.

Many studies in our review did not provide enough information to permit assessment of all items in the PROGRESS‐Plus framework. On average, information was available for 5.6 of the nine items. This lack of information means that we were unable to fully assess whether the results of this review are applicable to all individuals with OA. As pointed out by Furler 2011, disadvantaged people might be less able to access support for self‐management from healthcare professionals, and the quality of services may be lower, potentially increasing rather than decreasing health inequities. In addition to the PROGRESS‐Plus framework, we assessed health literacy in each study population, as it is likely to be a key determinant of a person’s ability to optimally manage his or her health and to ensure equitable access to and use of services. Only 4 of 29 (14%) trials provided any information on the health literacy of their study populations.

Quality of the evidence

For the main comparison, self‐management education programmes versus attention control (five studies, N = 937), the overall quality of evidence was graded as low to moderate (summary of findings Table for the main comparison). Evidence was downgraded because of methodological limitations of the studies, including lack of participant blinding, inadequate randomisation or concealment of allocation and greater numbers of withdrawals in the intervention group.

The overall quality of evidence for self‐management education programmes compared with usual care was low to moderate (16 studies, N = 3,738) (summary of findings Table 2). Evidence was downgraded, as participants and study personnel were not blind to group allocations, and some studies had inadequate randomisation or concealment of allocation.

The quality of evidence comparing self‐management education programmes versus the provision of information alone (4 studies, N = 340) was graded as low to moderate using the GRADE approach (summary of findings Table 3). Evidence was downgraded because of lack of participant blinding in all studies and because of additional methodological issues, including inadequate randomisation and unbalanced losses to follow‐up across groups.

For studies comparing self‐management education programmes versus alternative interventions (7 studies, N = 919), the overall quality of evidence was moderate (summary of findings Table 4). Evidence was downgraded because of lack of blinding of participants and study personnel and unclear randomisation and concealment of allocation in some studies.

Although the evidence was downgraded, we are not convinced that further well‐designed trials would substantially change the estimate of the effects, or the direction of the effects. Confounding from biases across studies would have likely favoured self‐management education programmes; thus it is unlikely that correcting for these biases would overturn the direction of the results; it may, in fact, drive some of the small but clinically unimportant improvements seen with self‐management over usual care towards the null (i.e. no significant differences between groups).

As well as grading the overall quality of the evidence, which relies on an assessment of risk of bias of individual trials, a novel aspect of our review included an attempt to assess the quality of the self‐management interventions that were being evaluated in the trials on the basis of the eight domains described in the heiQ (Table 4). At least some evidence indicates that elements of skill and technique acquisition were addressed in 94% of the self‐management education programmes, health‐directed activity was addressed in 85% and self‐monitoring and insight in 79%. However, social integration and support were addressed in only 12% of the self‐management education programmes. On average, interventions resulted in evidence of delivery of about four of the eight components (range two to seven), and limited evidence suggested that they were delivered in a high‐quality fashion. It is important to note that although we used inclusion of a component as a proxy for quality, we were unable to ascertain how well the included components were delivered. Furthermore, it could be argued that although we based our assessment on the heiQ, key therapeutic components of self‐management interventions remain unknown.

Potential biases in the review process

We believe that all relevant published studies were identified for inclusion in this review. A thorough search strategy was devised, and all major databases were searched for relevant studies with no language restrictions applied. Two review authors assessed the trials for inclusion in the review, and a third review author adjudicated any discussions or discrepancies.

Apart from the risk of bias of the included trials, the biggest limitation of the review process was that the self‐management education programmes in the trials differed in mode, personnel, delivery method and duration. Moreover, the trials varied in their outcome measures. For several outcomes, such as pain and function, we elected to pool different measures that may not necessarily be measuring the exact same concept. All post hoc decisions regarding choice of outcome data for inclusion in analyses of this review were recorded in the Notes section of Characteristics of included studies.

Agreements and disagreements with other studies or reviews

Previous systematic reviews that have investigated the effects of self‐management interventions for OA have reported broadly similar findings, although their inclusion criteria have all varied in some way from ours (e.g. inclusion of trials of interventions that provided only information, inclusion of mixed populations of arthritis or musculoskeletal pain in general, inclusion of trials in older participants only, inclusion of trials that included a combination of exercise and self‐management education) (Chodosh 2005; Devos‐Comby 2006; Du 2011; Smith 2009; Walsh 2006; Warsi 2003).

Chodosh 2005 included 14 studies for OA in a meta‐analysis that assessed chronic disease self‐management programmes in older adults (age criterion for inclusion not specified). Although minimal overlap of included trials with our review was noted (only four of their included trials were included in our review, and an additional three studies included in their review were excluded from ours because they included a mixed study population, and data for people with OA were not reported separately), they drew similar conclusions—pooled effects of self‐management interventions were statistically significant but clinically trivial for pain and function outcomes. Warsi 2003 included 15 trials that investigated self‐management education in people with OA or mixed populations; these review authors found similarly small benefits in terms of pain and function and noted that the overall high dropout rate of 19% raises concerns about the validity of these findings.

Devos‐Comby 2006 included 16 studies investigating exercise and/or self‐management interventions for participants with knee OA. Compared with control interventions, these review authors found a significant but modest benefit of self‐management interventions in terms of psychological but not physical well‐being. In contrast, they found that exercise regimens led to improvement in physical health (by self‐report and direct measures) and in overall impact of OA, while perceived psychological health remained unchanged. Walsh 2006 included 10 randomised controlled trials in a systematic review of combined exercise and self‐management programmes for people with OA knee or hip. Although most trials reported significant benefits in terms of pain (seven of 10 trials) and function (eight of 10 trials), these review authors identified many methodological weaknesses, and clinical heterogeneity precluded meta‐analysis. Smith 2009 included 13 trials in a systematic review of self‐management education and/or exercise interventions for knee OA and concluded that no evidence indicated that self‐management reduced pain or improved function or quality of life compared with a wait list or no treatment group, and limited evidence from one trial suggested that self‐management was more effective than standard care in reducing pain and improving function. Similar to Devos‐Comby 2006, these review authors reported that the exercise component appeared to provide a small but significant benefit in terms of reducing pain, improving function and improving aspects of quality of life. Taken together, these reviews suggest that programmes that include an exercise component, as well as components directed at improving psychological outcomes, may be worthwhile, but lack of high‐quality evidence precludes confident extrapolation of these research findings into clinical practice.

Du 2011 performed a systematic review to determine the effectiveness of self‐management programmes for pain and disability in chronic musculoskeletal pain conditions. Trials were included only if interventions focused primarily on managing pain and minimising disability, and pain and disability were the primary outcomes of interest. These review authors included 16 trials of mixed populations of arthritis (only four of their included trials were included in our review, and an additional four studies included in their review were excluded from ours because they included a mixed study population, and data for people with OA were not reported separately). These review authors found that self‐management programmes resulted in only small to moderate effects in terms of improving pain and disability in the long term (and no improvements in disability were seen in the medium term).

In keeping with the findings of our review, Jüni 2006 devised a league table of selected interventions for osteoarthritis in terms of effect sizes typically found in large‐scale randomised controlled trials (comprising at least 100 participants) and found that formal participant education interventions have a minimal effect on pain (effect size ‐0.10, 95% CI ‐0.19 to 0.01— for an approximate difference of ‐0.25 on a 10‐cm visual analogue pain scale).

Reviews on self‐management education programmes in other chronic conditions have generally found small positive results. One Cochrane review examining self‐management education programmes in chronic obstructive pulmonary disease found that self‐management education was associated with a reduction in hospital admissions without detrimental effects on other outcome parameters (Effing 2007). Another Cochrane review found that education in asthma self‐management, which involves self‐monitoring by peak expiratory flow or by symptoms, coupled with regular medical review and a written action plan, has been found to improve health outcomes in adults with asthma (Gibson 2002).

A Cochrane review of self‐management education programmes in type 2 diabetes found that group‐based training in self‐management strategies is effective in improving fasting blood glucose levels, glycate haemoglobin and diabetes knowledge and in reducing systolic blood pressure levels, body weight and the requirement for diabetes medication (Deakin 2005). Finally, in a review that assessed the effects of self‐management education programmes in chronic conditions, lay‐led self‐management education programmes were found to result in small, short‐term improvements in participants' self‐management, self‐rated health, cognitive symptom management and healthcare use (Foster 2009). It may be that the nature of the disease influences the outcomes that can be achieved through self‐management education programmes. Unlike chronic obstructive pulmonary disease, for asthma and diabetes, in which clear demonstrable complications or deterioration in the condition can occur as a result of poor management, the management of OA is largely concerned with managing the symptoms of the persistent underlying pathological condition and impact on quality of life as measured by subjective tools. Further, other chronic conditions tend to be evaluated on the basis of specific clinical outcomes relevant to the condition using measurement tools or tests that can measure the outcomes with high precision.

We found that data were insufficient to allow review authors to perform a preplanned subgroup analysis to explore whether a relationship could be discerned between any of the component domains addressed in the self‐management education programmes and participant outcomes (Pitt 2011). However, Carnes 2012 performed a systematic review to specifically uncover the evidence for effectiveness of different self‐management course characteristics and components for chronic musculoskeletal pain. Upon review of 46 RCTs involving 8,539 participants, these review authors reported more beneficial effects in group‐delivered courses that included healthcare professional input and slightly more consistent beneficial effects for courses with a psychological component.

Flowchart.
Figures and Tables -
Figure 1

Flowchart.

Navigate to figure in ReviewOpen in new tab

Summary of the risk of bias across all included studies.
Figures and Tables -
Figure 2

Summary of the risk of bias across all included studies.

Navigate to figure in ReviewOpen in new tab

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Figures and Tables -
Figure 3

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Navigate to figure in ReviewOpen in new tab

Funnel plot of comparison: 3 SMP versus Usual care/No treatment/Wait list, outcome: 3.1 Self‐management of OA.
Figures and Tables -
Figure 4

Funnel plot of comparison: 3 SMP versus Usual care/No treatment/Wait list, outcome: 3.1 Self‐management of OA.

Navigate to figure in ReviewOpen in new tab

Funnel plot of comparison: 3 SMP versus Usual care/No treatment/Wait list, outcome: 3.3 Dropouts.
Figures and Tables -
Figure 5

Funnel plot of comparison: 3 SMP versus Usual care/No treatment/Wait list, outcome: 3.3 Dropouts.

Navigate to figure in ReviewOpen in new tab

Funnel plot of comparison: 3 SMP versus Usual care/No treatment/Wait list, outcome: 3.4 Pain.
Figures and Tables -
Figure 6

Funnel plot of comparison: 3 SMP versus Usual care/No treatment/Wait list, outcome: 3.4 Pain.

Navigate to figure in ReviewOpen in new tab

Comparison 1 SMP versus attention control, Outcome 1 Self‐management of OA.
Figures and Tables -
Analysis 1.1

Comparison 1 SMP versus attention control, Outcome 1 Self‐management of OA.

Navigate to figure in ReviewOpen in new tab

Comparison 1 SMP versus attention control, Outcome 2 Pain.
Figures and Tables -
Analysis 1.2

Comparison 1 SMP versus attention control, Outcome 2 Pain.

Navigate to figure in ReviewOpen in new tab

Comparison 1 SMP versus attention control, Outcome 3 Global OA scores.
Figures and Tables -
Analysis 1.3

Comparison 1 SMP versus attention control, Outcome 3 Global OA scores.

Navigate to figure in ReviewOpen in new tab

Comparison 1 SMP versus attention control, Outcome 4 Function—self‐reported.
Figures and Tables -
Analysis 1.4

Comparison 1 SMP versus attention control, Outcome 4 Function—self‐reported.

Navigate to figure in ReviewOpen in new tab

Comparison 1 SMP versus attention control, Outcome 5 Quality of life.
Figures and Tables -
Analysis 1.5

Comparison 1 SMP versus attention control, Outcome 5 Quality of life.

Navigate to figure in ReviewOpen in new tab

Comparison 1 SMP versus attention control, Outcome 6 Withdrawals.
Figures and Tables -
Analysis 1.6

Comparison 1 SMP versus attention control, Outcome 6 Withdrawals.

Navigate to figure in ReviewOpen in new tab

Comparison 1 SMP versus attention control, Outcome 7 Emotional distress.
Figures and Tables -
Analysis 1.7

Comparison 1 SMP versus attention control, Outcome 7 Emotional distress.

Navigate to figure in ReviewOpen in new tab

Comparison 2 SMP versus usual care/no treatment/wait list, Outcome 1 Self‐management of OA.
Figures and Tables -
Analysis 2.1

Comparison 2 SMP versus usual care/no treatment/wait list, Outcome 1 Self‐management of OA.

Navigate to figure in ReviewOpen in new tab

Comparison 2 SMP versus usual care/no treatment/wait list, Outcome 2 Engagement in life.
Figures and Tables -
Analysis 2.2

Comparison 2 SMP versus usual care/no treatment/wait list, Outcome 2 Engagement in life.

Navigate to figure in ReviewOpen in new tab

Comparison 2 SMP versus usual care/no treatment/wait list, Outcome 3 Pain.
Figures and Tables -
Analysis 2.3

Comparison 2 SMP versus usual care/no treatment/wait list, Outcome 3 Pain.

Navigate to figure in ReviewOpen in new tab

Comparison 2 SMP versus usual care/no treatment/wait list, Outcome 4 Global OA scores.
Figures and Tables -
Analysis 2.4

Comparison 2 SMP versus usual care/no treatment/wait list, Outcome 4 Global OA scores.

Navigate to figure in ReviewOpen in new tab

Comparison 2 SMP versus usual care/no treatment/wait list, Outcome 5 Function—self‐reported.
Figures and Tables -
Analysis 2.5

Comparison 2 SMP versus usual care/no treatment/wait list, Outcome 5 Function—self‐reported.

Navigate to figure in ReviewOpen in new tab

Comparison 2 SMP versus usual care/no treatment/wait list, Outcome 6 Function—performance.
Figures and Tables -
Analysis 2.6

Comparison 2 SMP versus usual care/no treatment/wait list, Outcome 6 Function—performance.

Navigate to figure in ReviewOpen in new tab

Comparison 2 SMP versus usual care/no treatment/wait list, Outcome 7 Quality of life.
Figures and Tables -
Analysis 2.7

Comparison 2 SMP versus usual care/no treatment/wait list, Outcome 7 Quality of life.

Navigate to figure in ReviewOpen in new tab

Comparison 2 SMP versus usual care/no treatment/wait list, Outcome 8 Withdrawals.
Figures and Tables -
Analysis 2.8

Comparison 2 SMP versus usual care/no treatment/wait list, Outcome 8 Withdrawals.

Navigate to figure in ReviewOpen in new tab

Comparison 2 SMP versus usual care/no treatment/wait list, Outcome 9 Emotional distress.
Figures and Tables -
Analysis 2.9

Comparison 2 SMP versus usual care/no treatment/wait list, Outcome 9 Emotional distress.

Navigate to figure in ReviewOpen in new tab

Comparison 2 SMP versus usual care/no treatment/wait list, Outcome 10 Health‐directed activity.
Figures and Tables -
Analysis 2.10

Comparison 2 SMP versus usual care/no treatment/wait list, Outcome 10 Health‐directed activity.

Navigate to figure in ReviewOpen in new tab

Comparison 2 SMP versus usual care/no treatment/wait list, Outcome 11 Skill and technique acquisition.
Figures and Tables -
Analysis 2.11

Comparison 2 SMP versus usual care/no treatment/wait list, Outcome 11 Skill and technique acquisition.

Navigate to figure in ReviewOpen in new tab

Comparison 2 SMP versus usual care/no treatment/wait list, Outcome 12 Constructive attitudes and approaches.
Figures and Tables -
Analysis 2.12

Comparison 2 SMP versus usual care/no treatment/wait list, Outcome 12 Constructive attitudes and approaches.

Navigate to figure in ReviewOpen in new tab

Comparison 2 SMP versus usual care/no treatment/wait list, Outcome 13 Social integration and support.
Figures and Tables -
Analysis 2.13

Comparison 2 SMP versus usual care/no treatment/wait list, Outcome 13 Social integration and support.

Navigate to figure in ReviewOpen in new tab

Comparison 2 SMP versus usual care/no treatment/wait list, Outcome 14 Health service navigation.
Figures and Tables -
Analysis 2.14

Comparison 2 SMP versus usual care/no treatment/wait list, Outcome 14 Health service navigation.

Navigate to figure in ReviewOpen in new tab

Comparison 3 SMP versus information only, Outcome 1 Self‐management of OA.
Figures and Tables -
Analysis 3.1

Comparison 3 SMP versus information only, Outcome 1 Self‐management of OA.

Navigate to figure in ReviewOpen in new tab

Comparison 3 SMP versus information only, Outcome 2 Engagement in life.
Figures and Tables -
Analysis 3.2

Comparison 3 SMP versus information only, Outcome 2 Engagement in life.

Navigate to figure in ReviewOpen in new tab

Comparison 3 SMP versus information only, Outcome 3 Pain.
Figures and Tables -
Analysis 3.3

Comparison 3 SMP versus information only, Outcome 3 Pain.

Navigate to figure in ReviewOpen in new tab

Comparison 3 SMP versus information only, Outcome 4 Global OA scores.
Figures and Tables -
Analysis 3.4

Comparison 3 SMP versus information only, Outcome 4 Global OA scores.

Navigate to figure in ReviewOpen in new tab

Comparison 3 SMP versus information only, Outcome 5 Function—self‐reported.
Figures and Tables -
Analysis 3.5

Comparison 3 SMP versus information only, Outcome 5 Function—self‐reported.

Navigate to figure in ReviewOpen in new tab

Comparison 3 SMP versus information only, Outcome 6 Function—performance.
Figures and Tables -
Analysis 3.6

Comparison 3 SMP versus information only, Outcome 6 Function—performance.

Navigate to figure in ReviewOpen in new tab

Comparison 3 SMP versus information only, Outcome 7 Quality of life.
Figures and Tables -
Analysis 3.7

Comparison 3 SMP versus information only, Outcome 7 Quality of life.

Navigate to figure in ReviewOpen in new tab

Comparison 3 SMP versus information only, Outcome 8 Withdrawals.
Figures and Tables -
Analysis 3.8

Comparison 3 SMP versus information only, Outcome 8 Withdrawals.

Navigate to figure in ReviewOpen in new tab

Comparison 3 SMP versus information only, Outcome 9 Emotional distress.
Figures and Tables -
Analysis 3.9

Comparison 3 SMP versus information only, Outcome 9 Emotional distress.

Navigate to figure in ReviewOpen in new tab

Comparison 3 SMP versus information only, Outcome 10 Health‐directed activity.
Figures and Tables -
Analysis 3.10

Comparison 3 SMP versus information only, Outcome 10 Health‐directed activity.

Navigate to figure in ReviewOpen in new tab

Comparison 3 SMP versus information only, Outcome 11 Social integration and support.
Figures and Tables -
Analysis 3.11

Comparison 3 SMP versus information only, Outcome 11 Social integration and support.

Navigate to figure in ReviewOpen in new tab

Comparison 3 SMP versus information only, Outcome 12 Health service navigation.
Figures and Tables -
Analysis 3.12

Comparison 3 SMP versus information only, Outcome 12 Health service navigation.

Navigate to figure in ReviewOpen in new tab

Comparison 3 SMP versus information only, Outcome 13 Skill and technique acquisition.
Figures and Tables -
Analysis 3.13

Comparison 3 SMP versus information only, Outcome 13 Skill and technique acquisition.

Navigate to figure in ReviewOpen in new tab

Comparison 3 SMP versus information only, Outcome 14 Constructive attitudes and approaches.
Figures and Tables -
Analysis 3.14

Comparison 3 SMP versus information only, Outcome 14 Constructive attitudes and approaches.

Navigate to figure in ReviewOpen in new tab

Comparison 4 SMP versus non‐SMP intervention, Outcome 1 Self‐management of OA.
Figures and Tables -
Analysis 4.1

Comparison 4 SMP versus non‐SMP intervention, Outcome 1 Self‐management of OA.

Navigate to figure in ReviewOpen in new tab

Comparison 4 SMP versus non‐SMP intervention, Outcome 2 Pain.
Figures and Tables -
Analysis 4.2

Comparison 4 SMP versus non‐SMP intervention, Outcome 2 Pain.

Navigate to figure in ReviewOpen in new tab

Comparison 4 SMP versus non‐SMP intervention, Outcome 3 Global OA scores.
Figures and Tables -
Analysis 4.3

Comparison 4 SMP versus non‐SMP intervention, Outcome 3 Global OA scores.

Navigate to figure in ReviewOpen in new tab

Comparison 4 SMP versus non‐SMP intervention, Outcome 4 Function—self‐reported.
Figures and Tables -
Analysis 4.4

Comparison 4 SMP versus non‐SMP intervention, Outcome 4 Function—self‐reported.

Navigate to figure in ReviewOpen in new tab

Comparison 4 SMP versus non‐SMP intervention, Outcome 5 Function—performance.
Figures and Tables -
Analysis 4.5

Comparison 4 SMP versus non‐SMP intervention, Outcome 5 Function—performance.

Navigate to figure in ReviewOpen in new tab

Comparison 4 SMP versus non‐SMP intervention, Outcome 6 Quality of life.
Figures and Tables -
Analysis 4.6

Comparison 4 SMP versus non‐SMP intervention, Outcome 6 Quality of life.

Navigate to figure in ReviewOpen in new tab

Comparison 4 SMP versus non‐SMP intervention, Outcome 7 Withdrawals.
Figures and Tables -
Analysis 4.7

Comparison 4 SMP versus non‐SMP intervention, Outcome 7 Withdrawals.

Navigate to figure in ReviewOpen in new tab

Comparison 4 SMP versus non‐SMP intervention, Outcome 8 Emotional distress.
Figures and Tables -
Analysis 4.8

Comparison 4 SMP versus non‐SMP intervention, Outcome 8 Emotional distress.

Navigate to figure in ReviewOpen in new tab

Comparison 4 SMP versus non‐SMP intervention, Outcome 9 Constructive attitudes and approaches.
Figures and Tables -
Analysis 4.9

Comparison 4 SMP versus non‐SMP intervention, Outcome 9 Constructive attitudes and approaches.

Navigate to figure in ReviewOpen in new tab

Comparison 5 SMP versus acupuncture, Outcome 1 Global OA scores.
Figures and Tables -
Analysis 5.1

Comparison 5 SMP versus acupuncture, Outcome 1 Global OA scores.

Navigate to figure in ReviewOpen in new tab

Comparison 5 SMP versus acupuncture, Outcome 2 Pain.
Figures and Tables -
Analysis 5.2

Comparison 5 SMP versus acupuncture, Outcome 2 Pain.

Navigate to figure in ReviewOpen in new tab

Comparison 5 SMP versus acupuncture, Outcome 3 Function self‐reported.
Figures and Tables -
Analysis 5.3

Comparison 5 SMP versus acupuncture, Outcome 3 Function self‐reported.

Navigate to figure in ReviewOpen in new tab

Comparison 5 SMP versus acupuncture, Outcome 4 Function performance.
Figures and Tables -
Analysis 5.4

Comparison 5 SMP versus acupuncture, Outcome 4 Function performance.

Navigate to figure in ReviewOpen in new tab

Comparison 5 SMP versus acupuncture, Outcome 5 Withdrawals.
Figures and Tables -
Analysis 5.5

Comparison 5 SMP versus acupuncture, Outcome 5 Withdrawals.

Navigate to figure in ReviewOpen in new tab

Comparison 6 Subgroup analysis, Outcome 1 Self‐management in OA.
Figures and Tables -
Analysis 6.1

Comparison 6 Subgroup analysis, Outcome 1 Self‐management in OA.

Navigate to figure in ReviewOpen in new tab

Comparison 6 Subgroup analysis, Outcome 2 Function self‐reported.
Figures and Tables -
Analysis 6.2

Comparison 6 Subgroup analysis, Outcome 2 Function self‐reported.

Navigate to figure in ReviewOpen in new tab

Comparison 6 Subgroup analysis, Outcome 3 Pain.
Figures and Tables -
Analysis 6.3

Comparison 6 Subgroup analysis, Outcome 3 Pain.

Navigate to figure in ReviewOpen in new tab

Comparison 6 Subgroup analysis, Outcome 4 Withdrawals.
Figures and Tables -
Analysis 6.4

Comparison 6 Subgroup analysis, Outcome 4 Withdrawals.

Navigate to figure in ReviewOpen in new tab
Summary of findings for the main comparison. SMP compared to Attention control for osteoarthritis

SMP compared with attention control for osteoarthritis

Patient or population: patients with osteoarthritis
Settings: primary care, or outpatient
Intervention: SMP
Comparison: attention control

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No of participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Attention control

SMP

Self‐management of OA
Arthritis self‐efficacy scale (ASES). Scale from 1 to 10, higher better
Follow‐up: 12 months

Mean self‐management of osteoarthritis in the control groups was
5.8 points

Mean self‐management of osteoarthritis in the intervention groups was
0.4 points higher
(0.4 lower to 1.2 higher)

344
(one study)

⊕⊕⊝⊝
low1,2

MD 0.4 (‐0.39 to 1.19)

Absolute mean improvement 4% (4% worse to 12% improved) Relative improvement 7% (7% worse to 21% improved)3

Positive and active engagement in life —not measured

See comment

See comment

Not estimable

‐

See comment

No studies measured this outcome

Pain
Multiple tools4. Scale from: 0 to 10, 0 = no pain
Follow‐up: six to 12 months

Mean pain ranged across control groups from
5.67 to 6.19 points

Mean pain in the intervention groups was
0.8 points lower
(0.3 to 0.14 lower)

575
(three studies)

⊕⊕⊝⊝
low1,2

SMD ‐0.26 (‐0.44 to ‐0.09)

Absolute reduction in pain 8% (3% to 14% reduction). Relative reduction in pain 13% (5% to 22% reduction). NNTB = 8 (5 to 23)5

Global OA scores
AIMS2 (average of physical, affect, and pain subscales). Scale from 0 to 10, lower better
Follow‐up: nine months

Mean global osteoarthritis symptom score in the control group was
4.22 points

Mean global osteoarthritis symptom score in the intervention group was
0.14 points lower
(0.54 lower to 0.26 higher)

251
(one study)

⊕⊕⊝⊝
low2,6

Absolute reduction 1.4% (5.4% reduction to 2.6% increase). Relative reduction 3% (11% reduction to 5% increase)

Self‐reported function
Multiple tools.7 Lower score better
Follow‐up: 12 months

Mean self‐reported function in the control groups was
1.29 points on 0 to 3 scale8

Mean self‐reported function in the intervention groups was
0.04 points lower
(0.02 lower to 0.10 higher)

574
(three studies)

⊕⊕⊝⊝
low1,2

SMD ‐0.19 (‐0.5 to 0.11)

Absolute improvement in function 4% (2% reduction to 11% improvement). Relative improvement 11% (6% reduction to 30% improvement)8

Quality of life
Quality of well‐being scale. Scale from 0 to 1, higher better
Follow‐up: 12 months

Mean quality of life in the control groups was
0.57 units

Mean quality of life in the intervention groups was
0.01 lower
(0.03 lower to 0.01 higher)

165
(one study)

⊕⊕⊝⊝
low2,9

Absolute mean reduction 1% (95% CI 3% lower to 1% higher) Relative reduction 2% (95% CI 5% lower to 1% higher)

Withdrawals
Follow‐up: six to 12 months

117 per 1,000

130 per 1,000
(91 to 183)

RR 1.11
(0.78 to 1.57)

937
(five studies)

⊕⊕⊕⊝
moderate2

Absolute risk difference 1% increase (95% CI 3% decrease to 5% increase). Relative percentage change 11% increase (95% CI 22% decrease to 57% increase)

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: Confidence interval; RR: Risk ratio.

GRADE Working Group grades of evidence.
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1One large study was conducted in Veteran population that was mostly men (92.7%), limiting applicability of findings.

2Design flaws, including participants were not blind to group allocation in all trials, other trials had unclear randomisation method or concealment of allocation and unbalanced withdrawals across treatment groups, render the evidence susceptible to bias.

3Estimated relative changes based on mean (SD) ASES score in attention control group at baseline 5.8 (2.0) from Allen 2010.

4Pain VAS, pain on walking VAS and pain subscale of the arthritis impact measurement scale (AIMS).
5Estimated using mean (SD) for control group VAS pain on walking at baseline 6.28 (3.18) from Mazzuca 1997, and an assumed minimal clinically important difference of 1.5 points in 10‐point pain scale.
6Approximately half of total study population in trial had rheumatoid arthritis (data not included); data are presented for the OA subgroup; small sample size and wide CIs reduce precision.
7AIMS physical disability subscale, AIMS2 function subscale and HAQ disability subscale.
8Assumed risk from Mazzuca 1997 control group at 12 months mean HAQ disability scale:1.29 (SD 0.70); 0 to 3 scale, lower score better. Absolute risk difference estimated from control group SD at baseline from the same study (SD 0.66); and relative percent change using mean control group HAQ score at baseline (1.13).
9Potential imprecision due to data available only from a single study (n = 165).

Figures and Tables -
Summary of findings for the main comparison. SMP compared to Attention control for osteoarthritis
Navigate to table in Review
Summary of findings 2. SMP compared with usual care for osteoarthritis

SMP compared with usual care

Patient or population: patients with osteoarthritis
Settings: community, outpatient, primary care
Intervention: SMP
Comparison: usual care or no treatment or wait list control

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No of participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Usual care/No treatment/Wait list

SMP

Self‐management of OA
Multiple tools1. Scale from 1 to 10, higher better
Follow‐up: three to 21 months

Mean self‐management of osteoarthritis in the control groups, using ASES 1 to 10‐point scale (10 is better), was
3.7 points2

Mean self‐management of osteoarthritis in the intervention groups was
0.13 points higher
(0.02 to 0.23 higher)

1,706
(11 studies)

⊕⊕⊕⊝
moderate3

SMD 0.16 (0.03 to 0.29)

Absolute mean improvement 1.3% (0.2% to 2.3% improvement). Relative improvement 3.5% (0.65% to 6.3% improvement). NNTB 13 (7 to 69)2

Positive and active engagement in life
Multiple tools4
Follow‐up: six to 12 months

Mean positive and active engagement in life in the control groups, based on SF‐36 subscale for role emotional, 0 to 100 scale (100 best), was
57 points5

Mean positive and active engagement in life in the intervention groups was

0.4 points higher
(8 lower to 8.4 higher)

357
(three studies)

⊕⊕⊕⊝
moderate3

SMD 0.01 (‐0.2 to 0.21)

Absolute mean improvement 0.4% (8% worsening to 8.8% improvement). Relative improvement 0.5% (10% worsening to 10% improvement)

Pain
Multiple tools6
Follow‐up: three to 21 months

Mean pain in the control groups, based on 0 to 10 VAS scale (0 is no pain), was
3.5 points7

Mean pain in the intervention groups was
0.5 points lower
(0.23 to 0.1 lower)

2,083
(14 studies)

⊕⊕⊕⊝
moderate3

SMD ‐0.19 (‐0.28 to ‐0.10)

Absolute mean reduction 5.5% (8% to 3% reduction). Relative reduction 16% (23% to 8% reduction). NNTB 11 (7 to 21)

Global OA scores
Multiple tools8
Follow‐up: six to 21 months

Mean global osteoarthritis scores in the control groups, based on 0 to 96 point WOMAC scale (lower is better), was
35 points9

Mean global osteoarthritis score in the intervention groups was
5.0 points lower
(7.6 to 2.7 lower)9

1,957
(seven studies)

⊕⊕⊕⊝
moderate3

SMD ‐0.25 (‐0.37 to ‐0.13)

Absolute mean improvement 5% (3% to 8% improvement). Relative improvement 13% (7% to 19% improvement). NNTB 10 (7 to 19)9

Function—Self‐reported
Scale from 0 to 68. Lower score is better
Follow‐up: six to 21 months

Mean function self‐reported in the control groups, based on 0 to 68 WOMAC subscale (lower is better), was
25 points10

Mean function self‐reported in the intervention groups was
2.6 points lower
(3.9 to 1.3 lower)10

2,254
(13 studies)

⊕⊕⊕⊝
moderate3

SMD ‐0.18 (‐0.27 to ‐0.09)

Absolute improvement 4% (2% to 6% improvement). Relative improvement 10% (5% to 15% improvement). NNTB 14 (9 to 27)10

Quality of life
Multiple tools11
Follow‐up: six to 21 months

Mean quality of life in the control groups, based on ‐0.11 to 1.0 EQ‐5D scale (higher score is better), was
0.66 points12

Mean quality of life in the intervention groups was
0.006 points higher
(0.03 lower to 0.04 higher)

1,383
(eight studies)

⊕⊕⊕⊝
moderate3

SMD 0.02 (‐0.09 to 0.13)

Absolute improvement 0.6% (2.7% worsening to 3.9% improvement). Relative improvement 1% (4.5% worsening to 6.5% improvement)

Withdrawals
Losses to follow‐up
Follow‐up: three to 21 months

172 per 1,000

171 per 1,000
(128 to 229)

RR 0.99
(0.74 to 1.33)

3,738
(16 studies)

⊕⊕⊝⊝
low3,13

Absolute risk difference 0% (3% lower to 4% higher). Relative difference 10% lower (26% lower to 33% higher)

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: Confidence interval; RR: Risk ratio.

GRADE Working Group grades of evidence.
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1Arthritis self‐efficacy scale (ASES), ASES subscale for pain, arthritis helplessness index (AHI) and the health education impact questionnaire (heiQ).
2Self‐management measured using ASES 1 to 10 scale; 10 is best score, taken from Heuts 2005: mean (SD) baseline ASES score in control group was 3.7 (0.8), and mean (SD) ASES final score in control group was 3.7 (0.9); for number needed to treat for an additional beneficial outcome (NNTB) calculation, the minimal clinically important difference not known, assumed as 0.5.
3Participants and study personnel were not blind to group allocations; other issues included unclear randomisation and concealment of allocation; thus trials were at risk of selection, performance and detection biases. One trial, Victor 2005, had inconsistent results compared with the other 10 trials, possibly related to a high risk of bias in that trial, but we do not believe it was significant enough to downgrade the evidence further.
4Quality of life short form 36 (SF‐36) subscale for role emotional and heiQ subscale for positive and active engagement in life.

5Positive and active engagement in life calculated from Victor 2005, using SF‐36, 0 to 100 scale (100 is highest score): mean (SD) baseline score in control group was x (y); and mean final score in control group was 57 points.
6Arthritis impact measurement scale (AIMS), visual analogue scale (VAS) and Western Ontario McMaster Universities Arthritis Index (WOMAC).

7Pain calculated from Heuts 2005, using VAS 0 to 10 scale (0 is no pain): Mean (SD) baseline hip pain score in control group was 3.5 (2.9); and mean final score in control group was 3.5 (2.7).
8Western Ontario McMasters University Arthritis Index (WOMAC), Arthritis Impact Measurement Scale (AIMS2) and self‐rated global health questionnaires.
9Global disease scores taken from Hurley 2007, using WOMAC 0 to 96 point scale (lower score better): Mean (SD) baseline score in the control group was 38.4 (19.82); and mean final score in control group was 35 points. For number needed to treat for an additional beneficial outcome (NNTB) calculation, the minimal clinically important difference not known, assumed as 0.5.

10Self‐reported function based on Hurley 2007, using WOMAC function 0 to 68 point scale (lower score better): Mean (SD) baseline score in the control group was 27.2 (14.6); and mean final score in the control group was 25 points. For number needed to treat for an additional beneficial outcome (NNTB) calculation, the minimal clinically important difference not known, assumed as 0.5.
11Quality of life short form 36 (SF‐36), quality of life, quality of well‐being scale and EQ‐5D.

12Quality of life taken based on Hurley EQ‐5D (0 to 1 scale; higher score better): Control group mean (SD) at baseline was 0.6 (0.3) and at follow‐up was 0.66 (0.3).
13Inconsistency across studies regarding whether greater number of withdrawals in the self‐management group or control group.

Figures and Tables -
Summary of findings 2. SMP compared with usual care for osteoarthritis
Navigate to table in Review
Summary of findings 3. SMP compared with information only for osteoarthritis

SMP compared with information only for osteoarthritis

Patient or population: patients with osteoarthritis
Settings: community, outpatient, primary care
Intervention: SMP
Comparison: information only

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No of participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Information only

SMP

Self‐management of OA
Multiple tools1. Scale from 5 to 35 points (higher better)
Follow‐up: 12 months

Mean self‐management of osteoarthritis in the control groups, based on 5 to 35 point ASES self‐efficacy pain scale, was
19.2 points2

Mean self‐management of osteoarthritis in the intervention groups was
1.3 points higher
(0.26 lower to 2.82 higher)

760
(three studies)

⊕⊕⊕⊝
moderate3

SMD 0.20 (‐0.04 to 0.44)

Absolute mean improvement 4% (1% reduction to 9% improvement). Relative improvement 7% (1% reduction to 15% improvement)2

Positive and active engagement in life
heiQ subscale for positive and active engagement in life. Scale from 1 to 6 (higher better)
Follow‐up: 12 months

Mean positive and active engagement in life in the control group, on 1 to 6 point scale (higher better), was
4.76 points4

Mean positive and active engagement in life in the intervention group was
0.2 points lower
(0.59 lower to 0.18 higher)

93
(one study)

⊕⊕⊝⊝
low3,5

Absolute mean worsening 3% (10% worse to 3% improved). Relative mean worsening 4% (12% worse to 4% improved)4

Pain
WOMAC subscale for pain . Scale from 0 to 20 (lower better)
Follow‐up: 12 months

Mean pain in the control group, based on 0 to 20 WOMAC pain subscale (lower is better), was
8.5 points2

Mean pain in the intervention groups was
0.3 points lower
(0.8 lower to 0.3 higher)

751
(three studies)

⊕⊕⊕⊝
moderate3

SMD ‐0.07 (‐0.21 to 0.08)

Absolute mean reduction in pain 1.3% (4.0% reduction to 1.5% increase). Relative mean reduction 3% (9% reduction to 3% increase)

Global OA scores
Multiple tools6. Scale from 0 to 96 (lower score better)
Follow‐up: 12 months

Mean global OA scores in the control group, based on 0 to 96 total WOMAC score (lower better), was
41.1 points2

Mean global OA score in the intervention group was
0.8 points lower
(3.7 lower to 2.1 higher)

751
(three studies)

⊕⊕⊕⊝
moderate3

SMD ‐0.06 (‐0.28 to 0.16)

Absolute mean improvement 0.8% (2% worse to 4% improved). Relative mean improvement 2% (5% worse to 9% improvement)2

Self‐reported function
Multiple tools7. Scale from 0 to 68 (lower better)
Follow‐up: six to 12 months

Mean self‐reported function in the control group, based on 0 to 68 WOMAC function scale (lower better), was
28.9 points2

Mean self‐reported function in the intervention groups was
1.1 points lower
(2.7 lower to 0.6 higher)

854
(four studies)

⊕⊕⊕⊝
moderate3

SMD ‐0.09 (‐0.22 to 0.05)

Absolute mean improvement 2% (4% improved to 1% worse). Relative mean improvement 4% (10% improved to 2% worse)2

Quality of life
Multiple tools8. Scale from 0 to 100 (higher better)
Follow‐up: 12 months

Mean quality of life in the control group, based on 0 to 100 point scale (higher better), was
55.9 points2

Mean quality of life in the intervention group was
0.5 points higher
(1 lower to 2 higher)

648
(two studies)

⊕⊕⊕⊝
moderate3

SMD 0.05 (‐0.1 to 0.21)

Absolute mean improvement 0.5% (1% worsening to 2% improvement). Relative mean improvement 1% (2% worsening to 4% improvement)2

Withdrawals
Losses to follow‐up
Follow‐up: six to 12 months

243 per 1,000

389 per 1,000
(182 to 827)

RR 1.6
(0.75 to 3.4)

1,251
(four studies)

⊕⊕⊝⊝
low3,9

Absolute difference 6% higher withdrawals (8% lower to 19% higher). Relative increase 60% (25% decrease to 240% increase)

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: Confidence interval; RR: Risk ratio.

GRADE Working Group grades of evidence.
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1ASES subscale for pain and heiQ subscale for self‐monitoring and insight.
2Control group baseline and final values taken from Buszewicz 2006, used to estimate the mean difference between groups and absolute and relative changes: mean (SD) self‐management score, based on arthritis self‐efficacy pain 5 to 35 point scale (higher score better), in the control group at baseline was 19.2 (6.4), and at follow‐up was 18.8 (6.5) points; mean (SD) pain score, based on 0 to 20 point WOMAC pain subscale (lower better) in the control group at baseline, was 8.7 (3.7) points, and at follow‐up was 8.5 (3.9) points; mean (SD) global OA score, based on WOMAC 0 to 96 point scale (lower better), in the control group at baseline was 41.6 (13.32) and at follow‐up was 41.4 points; mean (SD) function, based on WOMAC 0 to 68 (lower better), in the control group at baseline was 29.1 (12.7) and at follow‐up was 28.9 points; mean (SD) quality of life, based on SF‐36 mental component score 0 to 100 (higher better), in the control group at baseline was 50.6 (10.6) and at follow‐up was 55.9 points.
3Design flaws, including participants were not blind to group allocation in all trials, some trials had unclear randomisation method or concealment of allocation and unbalanced withdrawals across treatment groups; thus the evidence is susceptible to selection, performance, detection or attrition biases.

4Baseline and final value control group heiQ scores (1 to 6 point scale, higher better) from Ackerman 2012: Mean (SD) at baseline was 4.8 (0.8); mean at follow‐up was 4.76 points.
5Findings based on a single study.
6WOMAC and the hip and knee multi‐attribute priority tool (MAPT).
7Function subscales of WOMAC and the Dutch AIMS‐SF.
8Mental health component of the short form 36 (SF‐36) and assessment of quality of life (AQoL).

9Inconsistency across studies regarding whether greater number of withdrawals in the self‐management group or the control group.

Figures and Tables -
Summary of findings 3. SMP compared with information only for osteoarthritis
Navigate to table in Review
Summary of findings 4. SMP compared with non‐SMP intervention for osteoarthritis

SMP compared with non‐SMP intervention for osteoarthritis

Patient or population: patients with osteoarthritis
Settings: community, outpatient or physiotherapy clinic, age care facility
Intervention: SMP
Comparison: non‐SMP intervention

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No of participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Non‐SMP intervention

SMP

Self‐management of OA
Arthritis self‐efficacy scale. Scale from 30 to 300.
Follow‐up: one to 12 months

Mean self‐management of osteoarthritis in the control groups, based on 30 to 300 ASES scale (higher score better), was
220.46 points1

Mean self‐management of osteoarthritis in the intervention groups was
12 points higher
(0 to 24 higher)1

175
(three studies)

⊕⊕⊕⊝
moderate2

SMD 0.33 (0 to 0.66)

Absolute mean improvement 4% (0% to 9% improvement). Relative improvement 5.7% (0% to 11.3% improvement)

Positive and active engagement in life

Not measured

See comment

See comment

Not estimable

0
(0)

See comment

No studies measured this outcome

Pain
Multiple tools3. Scale from 0 to 20. Lower score is better.
Follow‐up: one to 12 months

Mean pain in the control groups, based on 0 to 20 WOMAC pain subscale (lower better), was

4.2 points4

Mean pain in the intervention groups was
0.3 points lower
(1.2 lower to 0.5 higher)4

321
(five studies)

⊕⊕⊕⊝
moderate2

SMD ‐0.09 (‐0.36 to 0.17)

Absolute mean reduction 1.4% (‐5.8 to 2.7%). Relative reduction 6% (‐20% 9.5%)4

Global OA scores
WOMAC. Scale from 0 to 240. Lower score is better.
Follow‐up: 12 weeks

Mean global osteoarthritis scores in the control group, based on 0 to 240 WOMAC scale (lower better), was
66.8 points5

Mean global osteoarthritis score in the intervention group was
11.6 points higher
(5.6 lower to 28.7 higher)

98
(one study)

⊕⊕⊕⊝
moderate2

SMD 0.27 (‐0.13 to 0.67)

Absolute mean worsening 4.8% (12% worsening to 2.3% improvement). Relative worsening 12.7% (31.4% worsening to 6.1% improvement)5

Function—Self‐reported
Multiple tools3. Scale from 0 to 68. Lower score is better.
Follow‐up: one to 12 months

Mean function self‐reported in the control groups, based on 0 to 68 point WOMAC scale (lower better), was
12.2 points4

Mean function self‐reported in the intervention groups was
0.04 standard deviations higher
(0.34 lower to 0.42 higher)

216
(three studies)

⊕⊕⊕⊝
moderate2

SMD 0.04 (‐0.34 to 0.42)

Absolute worsening 0.6% (6.4% worsening to 5.2% improvement). Relative worsening 2.6% (27.4% worsening to 22.2% improvement)

Quality of life
Multiple tools6. Scale from 0 to 1. Higher score is better.
Follow‐up: 12 to 36 months

Mean quality of life in the control groups was
0.734

Mean quality of life in the intervention groups was
0.06 standard deviations lower
(0.49 lower to 0.36 higher)

226
(two studies)

⊕⊕⊕⊝
moderate2

SMD ‐0.06 (95% CI ‐0.49 to 0.36)

Absolute worsening 0.54% (4.4% worsening to 3.2% improvement). Relative worsening 0.7% (5.8% worsening to 4.3% improvement)4

Withdrawals
Losses to follow‐up.
Follow‐up: one to 36 months

243 per 1,000

209 per 1,000
(168 to 265)

RR 0.86
(0.69 to 1.09)

919
(seven studies)

⊕⊕⊕⊝
moderate2

Absolute difference of 2% fewer withdrawals (7% fewer to 12% more). Relative percentage change of 14% fewer withdrawals (31% fewer to 9% more)

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: Confidence interval; RR: Risk ratio.

GRADE Working Group grades of evidence.
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1Baseline and final values for comparison group taken from Keefe 2004, ASES 30 to 300 self‐efficacy scale (higher better): Control group baseline mean (SD) was 215 (36.95); control group follow‐up score was 220.46.
2Trials had design flaws, making results susceptible to bias, including participants and study personnel were not blind to group allocations; unclear randomisation and concealment of allocation; and selective reporting.
3Arthritis impact measurement scale (AIMS) and Western Ontario McMasters University Arthritis Index (WOMAC).
4Control group baseline and final values taken from Jessep 2009. WOMAC 0 to 20 point pain subscale (lower better): Control group mean (SD) pain at baseline was 5.7 (3.2) and at 12‐month follow‐up was 4.2 (4.0). Control group WOMAC 0 to 68 point function subscale (lower better): Control group mean (SD) at baseline was 15.9 (10.4) and at follow‐up was 12.2. Control group quality of life 0 to 1 point EQ‐5D scale (higher is better): Control group mean (SD) at baseline was 0.76 (0.09) and at follow up was 0.73.
5Baseline and final control group means from Maurer 1999: Control group mean at baseline was 91.5 points on 0 to 240 WOMAC scale (0 is better) and at 12 weeks was 66.8 points; control group SD taken from Wolfe 1999 (42.9).
6Quality of well‐being scale and EQ‐5D.

Figures and Tables -
Summary of findings 4. SMP compared with non‐SMP intervention for osteoarthritis
Navigate to table in Review
Table 1. PROGRESS‐Plus

Study

P

R

O

Gr

E

S

S

A

D

HL

Allen 2010

US

54% white,

46% non‐white,

2% Hispanic

38.3% employed

92.7% male,

7.3% female

33.3% high school education or less

25.7% inadequate income

69.3% married

60.1 (10.4)

32.3% fair or poor health

‐

Berman 2004

US

69.1% white,

28.6% African American,

2.3% other

‐

35.9% male,

64.1% female

31.1% no college, 68.9% some college

‐

‐

65.5 (8.6)

75.1% moderate to lesser pain, 24.9% severe to extreme pain

‐

Blixen 2004

US

72% white,

28% African American

‐

62.5% male,

37.5% female

6% grade 7 to 9,

3% grade 10 and 11, 22% high school, 50% 1 to 4 year college, 9.5% college graduate, 9.5% professional/graduate school

22% < $10,000/y, 13% $10,000 to $19,999/y, 19% $20,000 to $29,999/y, 17.5% $30,000 to $39,999/y, 13.5% $40,000 to $49,999/y, 17.5% > $50,000/y

56.5% married, 25% divorced,

13% single,

25% widowed

70.8 (6.1)

16% joint replacement surgery

Domain social support for health

Buszewicz 2006

UK

> 99% white,

< 1% black Caribbean,

< 1% black African

‐

37% male,

63% female

72.5% no higher education,

27.5% higher education

‐

81% house owner, 1% staying with friends or family, 18.5% rented accommodation

68.6 (8.4)

‐

‐

Calfas 1992

US

97.5% white,

2.5% non‐white

82.5% retired, 5% full‐time employed, 12.5% part‐time employed

27.5% male,

72.5% female

‐

‐

70% married, 20% divorced, 10% widowed,

5% other

67.0 (7.3)

‐

‐

Cronan 1997

US

92.3% white,

2.5% African American,

2.2% Hispanic,

1.7% Asian,

1.4% other

‐

36.3% male, 64.0% female

31.2% high school, 37.7% some college, 28.7% college degree

27.2% < $20,000, 43.2% $20,000 to $40,000, 19.7% $40,000 to $60,000, 10.0% > $60,000

‐

69.2 (5.6)

69.9% other medical conditions present

‐

Crotty 2009

Australia

‐

11.6% employed, 5.4% home duties, 82.7% retired

39.5% male,

60.5% female

3.5% none to some primary school, 15.0% primary school, 30.6% high school to year 8, 27.2% high school to year 12, 17.4% TAFE/trade, 6.2% university to above

‐

48.3% live alone

67.5 (10.8)

32.9% on waiting list for hip replacement

Domains social support for health, navigating health system, actively managing my health

Hansson 2010

Sweden

‐

‐

‐

‐

‐

‐

62.5 (9.4)

‐

‐

Heuts 2005

The Netherlands

‐

40.5% paying job, 35.5% no paying job

40.5% male,

59.5% female

25.5% low,

30.5% middle, 20% high

‐

‐

51.6 (5.1)

‐

‐

Hopman‐Rock 2000

The Netherlands

‐

‐

17% male,

83% female

21.5% primary, 49.5% secondary, 23.5% college/university

‐

67.5% living together/married, 27% living alone

65.3 (5.5)

2.5 (1.6) other chronic conditions

‐

Hughes 2004

US

72.2% white,

22.1% African American, 2.6% Hispanic,

2.1% Asian‐Pacific Islander, 1.1% other

‐

16.8% male, 83.3% female

10.5% less than high school, 20.0% high school, 69.6% more than high school

33.1% income less than $20,000

‐

73.4

22.4% ARA class I, 64.4% ARA class II, 13.3% ARA class III

‐

Hurley 2007

UK

‐

‐

29.7% male, 70.3% female

‐

‐

‐

67 (range 50 to 91)

‐

‐

Jessep 2009

UK

‐

‐

30.5% male, 69.5% female

‐

‐

‐

66.5 (range 51 to 81)

‐

‐

Keefe 1990

US

‐

‐

28.3% male, 71.7% female

‐

‐

‐

63.9 (11.5)

‐

‐

Keefe 1996

US

‐

‐

40% male,

60% female

‐

‐

‐

62.6

‐

‐

Keefe 2004

US

‐

‐

46.9% male, 53.4% female

‐

‐

‐

59.5 (11.1)

‐

‐

Lorig 2008

US

92.3% non‐Hispanic white

‐

9.9% male, 90.2% female

15.7 (3.1) years of education

‐

68.3% married

52.4 (11.6)

‐

Domain ability to actively engage with healthcare

providers

Maisiak 1996

US

85% white

‐

7.7% male, 92.3% female

12.1 years of schooling

‐

‐

60.4

48% see a specialist

‐

Martire 2007

US

‐

‐

27.3% male, 72.7% female

‐

‐

‐

68.6 (7.6)

‐

‐

Maurer 1999

US

‐

‐

58.5% male, 41.5% female

‐

‐

‐

65.4 (8.6)

‐

‐

Mazzuca 1997

US

69% African American

53% unemployed, 15% employed, 31.5% retired

15% male, 85% female

9.7 (3.1) years of education

96.5% annual income ≤ $20,000

73.5% living alone

62.4 (11.6)

1.6 (1.1) comorbid conditions

‐

Mazzuca 2004

US

70.5% white

‐

28.5% male, 71.5% female

87.5% 12 or more years of education

‐

62% married

61.8 (12.2)

‐

‐

McKnight 2010

US

91.7% white

‐

23.4% male, 77.0% female

63.0% college educated

‐

‐

52.6 (7.2)

‐

‐

Murphy 2008

US

91% white

‐

11% male, 89% female

66.5% some college to advanced degree

‐

22% married

75.3 (7.1)

4.5 (2.1) painful or stiff joints

‐

Nunez 2006

Spain

‐

84% retired or housewives, 13% permanently disabled, 3% active (sick leave)

24.5% male, 75.5% female

‐

‐

67.5% have family, 30.5% alone, 3% living with carer

71.1 (6.7)

33% prior prostheses

‐

Ackerman 2012

Australia

68.5% Australian‐born

26% paid employment, 62% retired, 8.5% not working because of medical condition, 3.5% unemployed

40% male, 60% female

12% primary school or less, 46% years 7 to 10, 15.5% years 11 and 12, 14.5% trade/technical education, 12% university

‐

63.5% married or living with partner

65.1 (10.9)

‐

Domains social support for health, navigating

health system, actively managing

my health

Victor 2005

UK

64% non‐white ethnic group

30% employed, 33.5% professional or managerial job

28% male, 72% female

37% higher education

66% home owner

35.5% living alone, 51% married

63.1 (11.1)

60% OA in both knees, 64% OA in other joints

‐

Wetzels 2005

The Netherlands

‐

‐

24% male, 76% female

52% primary or lower secondary, 48% upper secondary or further

‐

‐

74.5 (6.4)

‐

‐

Yip 2007

Hong Kong

100% Asian

26.4% housewife, 8.8% professional and administration, 64.9% service provider and non‐professional

16.0% male, 84.1% female

87.4% Form 3 level or below, 12.7% above Form 3

‐

69.8% married and living together, 30.3% single

64.8 (10.0)

10.9% one joint OA, 48.9% two joints OA, 23.8% three joints OA, 16.5% four or more joints OA

‐

P = Place of residence.
R = Race, ethnicity, culture.
O = Occupation.
Gr = Gender.
E = Education.
S = Socioeconomic status.
S = Social capital.
A = Age (in years).
D = Disability.
HL = Health literacy.

If not stated differently, data are ‘mean (SD)’.

y = year; IQR = interquartile range, ‐ = no information available.

No information was available on 'Religion' and 'Sexual orientation'; therefore these domains are not included in this table.

Figures and Tables -
Table 1. PROGRESS‐Plus
Navigate to table in Review
Table 2. Additional characteristics of included studies

Study

No

Location of OA

BMI

Duration of OA

Allen 2010

523

80% knee, 15% hip, 5% knee and hip

31.8 (6.7)

16.1 (12.2) years

Berman 2004

570

100% knee

‐

50.4% < 5 years,

20.7% 6 to 10 years,

28.7% > 10 years

Blixen 2004

32

‐

‐

8.3 (IQR 2 to 24) years

Buszewicz 2006

812

‐

‐

‐

Calfas 1992

40

‐

‐

10% < 1 year,

5% 1 to 5 years,

92.5% > 5 years

Cronan 1997

363

‐

‐

7.0 (5.5) years since diagnosis

Crotty 2009

152

32.9% hip

Height 167.2 (10.1) cm

Weight 85.8 (21.0) kg

‐

Hansson 2010

114

4.5% hip, 34% knee, 32% hand, 29% more locations

35% BMI 20 to 25,

38.5% BMI 25 to 30,

27% BMI > 30

‐

Heuts 2005

297

‐

28.1 (5.0)

‐

Hopman‐Rock 2000

120

‐

27.6 (4.3)

3% < 1 year,

20.5% 1 to 3 years,

33.5% 3 to 10 years,

18% 10 to 20 years,

17% > 20 years

Hughes 2004

215

‐

‐

‐

Hurley 2007

431

100% knee

30.2 (range 18 to 51)

6 (IQR 3 to 13) years

Jessep 2009

64

100% knee

29.5 (range 1 to 47)

12.5 (range 0.5 to 55) years

Keefe 1990

99

100% knee

24.2 (23.6)% above ideal weight

‐

Keefe 1996

88

100% knee

‐

‐

Keefe 2004

72

100% knee

‐

‐

Lorig 2008

551

‐

‐

‐

Maisiak 1996

405

‐

‐

16.0 years

Martire 2007

242

Hip and knee

‐

15.1 (10.9) years

Maurer 1999

113

100% knee

Weight 187.1 (33.9) lb

11.4 (10.5) years

Mazzuca 1997

211

100% knee

‐

14.0 (15.9) years

Mazzuca 2004

186

100% knee

‐

‐

McKnight 2010

273

100% knee

27.7 (4.2)

‐

Murphy 2008

54

67% knee, 11% hip, 22% hip and knee

30.1 (5.7)

‐

Nunez 2006

100

100% knee

‐

11.9 (10.6) months

Ackerman 2012

120

31% hip, 62.5% knee, 6.5% hip and knee

24.5 (IQR 25 to 35)

‐

Victor 2005

193

100% knee

‐

55% > 3 years

Wetzels 2005

104

53.8% knee, 20.1% hip, 26% hip and knee

‐

‐

Yip 2007

182

100% knee

‐

8.1 (6.8) years

Mean (SD), unless indicated otherwise.
No = Number of participants randomly assigned (total of all groups).
y = year(s).
‐ = no information available.
IQR = interquartile range.
BMI in kg/m2.

Figures and Tables -
Table 2. Additional characteristics of included studies
Navigate to table in Review
Table 3. Summary of comparisons

Study

Country

Setting

Intervention (N)

Mode/Personnel/Delivery method/Duration

Attention control (N)

Mode/Personnel/Delivery method/Duration

Allen 2010

US

Primary care

Self‐management intervention (174)

M: individual

P: health professionals

De: telephone

Du: 12 calls in 12 months

Health education (175)

M: individual

P: health professionals

De: telephone

Du: 12 calls in 12 months

Calfas 1992

US

Outpatients

Cognitive‐behaviour modification (20)

M: group

P: trained facilitator

De: face‐to‐face

Du: 10 weekly sessions

Traditional education intervention (20)

M: group

P: health specialists

De: face‐to‐face

Du: several lectures

Keefe 1990

US

Outpatients

Pain coping skills training (32)

M: group

P: nurse and psychologist

De: face‐to‐face

Du: 10 sessions in 10 weeks (1.5 hours)

and three phone calls

Arthritis education (36)

M: group

P: nurse and psychologist

De: lectures and telephone

Du: 10 sessions in 10 weeks (1.5 hours) and three phone calls

Maisiak 1996

US

Primary care and outpatients

Treatment counselling (135)

M: individual

P: trained counsellors

De: telephone

Du: 11 sessions in nine months

Symptom monitoring (135)

M: individual

P: college students trained for two hours

De: telephone

Du: 11 sessions in nine months

Mazzuca 1997

US

Primary care

Education (105)

M: individual

P: arthritis nurse educator

De: face‐to‐face/telephone

Du: one interview/two calls in one month

Attention control (106)

M: group and individual

P: not specified

De: audiovisual presentation and telephone

Du: 20‐minute presentation, two phone calls (five to 10 minutes)

Study

Setting

Intervention (N)

Information only (N)

Buszewicz 2006

UK

Primary care

Self‐management programme (406)

M: group

P: trained volunteer

De: face‐to‐face

Du: six sessions in six weeks

Education (406)

M: individual

P: none

De: education booklet

Du: ‐

Hughes 2004

US

General population

Fit & Strong (115)

M: group

P: physical therapist

De: face‐to‐face

Du: 24 sessions in eight weeks

Education (100)

M: individual

P: none

De: Arthritis Helpbook and handouts

Du: ‐

Ackerman 2012

Australia

Outpatients

Arthritis self‐management programme (58)

M: group

P: peer leader and health professional

De: face‐to‐face

Du: six sessions in six weeks

Education only (62)

M: individual

P: none

De: Arthritis Helpbook

Du: ‐

Wetzels 2005

The Netherlands

Primary care

Self‐management intervention (51)

M: individual

P: nurse

De: face‐to‐face/telephone

Du: one session/one call

Education (53)

M: individual

P: none

De: educational booklet

Du: ‐

Study

Setting

Intervention (N)

Usual care/Waiting list/No treatment (N)

Allen 2010

US

Primary care

Self‐management intervention (174)

M: individual

P: health professionals

De: telephone

Du: 12 calls in 12 months

Usual care (174)

 

Blixen 2004

US

Outpatients

Telephone health education strategy (16)

M: individual

P: advanced practice nurse

De: telephone

Du: six sessions in six weeks

Usual care (16)

 

Cronan 1997

US

General population

Education group (97)

M: group

P: professional health educator

De: face‐to‐face

Du: 10 weekly sessions and 10 monthly sessions

No treatment (90)

 

Cronan 1997

US

General population

Combination education + social support (89)

M: group

P: professional health educator (first hour)

De: face‐to‐face

Du: 10 weekly sessions and 10 monthly sessions

No treatment (90)

Crotty 2009

Australia

Outpatients

Patient education (75)

M: individual

P: peer support volunteer/research nurse

De: telephone/face‐to‐face

Du: six calls in six months/one interview

Usual care (77)

Hansson 2010

Sweden

Primary care

PEPOA (61)

M: group

P: health professionals

De: face‐to‐face

Du: five sessions in five weeks

Usual care (53)

Heuts 2005

The Netherlands

Primary care

Self‐management programme (149)

M: group

P: trained physiotherapists

De: face‐to‐face

Du: six sessions

Usual care (148)

Hopman‐Rock 2000

The Netherlands

General population

Health educational and exercise programme (60)

M: group

P: peer educator and health professionals

De: face‐to‐face

Du: six sessions in six weeks

No treatment (60)

Hurley 2007

UK

Primary care

Individual rehabilitation (146)

M: individual

P: physiotherapist

De: face‐to‐face

Du: 12 sessions in six weeks

Usual care (140)

Hurley 2007

UK

Primary care

Group rehabilitation (132)

M: group

P: physiotherapist

De: face‐to‐face

Du: 12 sessions in six weeks

Usual care (140)

Keefe 1990

US

Outpatients

Pain coping skills training (32)

M: group

P: nurse and psychologist

De: face‐to‐face

Du: 10 sessions in 10 weeks (1.5 hours)

and three phone calls

Usual care (31)

Keefe 2004

US

Outpatients and general population

Spouse‐assisted coping skills training (18)

M: group

P: trained psychologist

De: face‐to‐face

Du: 12 sessions in 12 weeks

Usual care (18)

Keefe 2004

US

Outpatients and general population

Spouse‐assisted coping skills training + exercise (20)

M: group

P: trained psychologist/exercise trainer

De: face‐to‐face

Du: 38 sessions in 12 weeks

Usual care (18)

Lorig 2008

US

General population

Internet‐based arthritis self‐management programme (433)

M: individual

P: peer moderators

De: Internet

Du: 18 sessions in six weeks (at least)

Usual care (422)

Maisiak 1996

US

Primary care and outpatients

Treatment counselling (135)

M: individual

P: trained counsellors

De: telephone

Du: 11 sessions in nine months

Usual care (135)

Martire 2007

US

Outpatients

Patient‐oriented education and support (89)

M: group

P: trained facilitator

De: face‐to‐face

Du: six sessions in six weeks

Usual care (54)

Martire 2007

US

Outpatients

Couple‐oriented education and support (99)

M: group

P: trained facilitator

De: face‐to‐face

Du: six sessions in six weeks

Usual care (54)

Mazzuca 2004

US

Primary care

Nurse‐directed intervention (111)

M: individual

P: arthritis nurse educator

De: face‐to‐face/telephone

Du: one interview/five to nine calls inthree to five months

Waiting list (75)

Nunez 2006

US

Outpatients

Self‐management programme (51)

M: group/individual

P: trained health educator

De: face‐to‐face

Du: four sessions in three months

Usual care (49)

Victor 2005

UK

Primary care

Self‐management (120)

M: group

P: research nurse

De: face‐to‐face

Du: four sessions

Waiting list (73)

Yip 2007

China

Outpatients

Arthritis self‐management programme (88)

M: group

P: nurses

De: face‐to‐face

Du: six sessions in six weeks

Usual care (94)

Study

Setting

Intervention (N)

Alternate intervention (N)

Berman 2004

US

General population

Education (189)

M: group

P: patient education specialist

De: face‐to‐face

Du: six sessions in 12 weeks

True/sham acupuncture (190 + 191)

M: individual

P: acupuncturists

De: face‐to‐face

Du: 23 treatments in 26 weeks

Cronan 1997

US

General population

Education group (97)

M: group

P: professional health educator

De: face‐to‐face

Du: 10 weekly sessions and 10 monthly sessions (two hours)

Social support group (87)

M: group

P: none

De: face‐to‐face

Du: 10 weekly sessions and 10 monthly sessions (two hours)

Cronan 1997

US

General population

Combination education + social support (89)

M: group

P: professional health educator (first hour)

De: face‐to‐face

Du: 10 weekly sessions and 10 monthly sessions (two hours)

Social support group (87)

M: group

P: none

De: face‐to‐face

Du: 10 weekly sessions and 10 monthly sessions (two hours)

Jessep 2009

UK

Outpatients

ESCAPE‐knee pain (29)

M: group

P: physiotherapist

De: face‐to‐face

Du: 10 sessions in five weeks

Physiotherapy (35)

M: not specified

P: physiotherapist

De: face‐to‐face

Du: maximum of 10 sessions

Keefe 1996

US

Not specified

Coping skills training (29)

M: group

P: nurse and psychologist

De: face‐to‐face

Du: 10 sessions in 10 weeks (two hours)

Arthritis education—spousal support (29)

M: group

P: nurse and psychologist

De: discussions and educational material

Du: one time a week (two hours) for 10 weeks

Keefe 1996

US

Not specified

Spouse‐assisted coping skills training (30)

M: group

P: nurse and psychologist

De: face‐to‐face

Du: 10 sessions in 10 weeks (two hours)

Arthritis education—spousal support (29)

M: group

P: nurse and psychologist

De: discussions and educational material

Du: one time a week (two hours) for 10 weeks

Keefe 2004

US

Outpatients and general population

Spouse‐assisted coping skills training (18)

M: group

P: trained psychologist

De: face‐to‐face

Du: 12 sessions in 12 weeks

Exercise (16)

M: group

P: exercise physiologists (BA level or above)

De: face‐to‐face

Du: three times a week (one hour) for 12 weeks

Keefe 2004

US

Outpatients and general population

Spouse‐assisted coping skills training + exercise (20)

M: group

P: trained psychologist/exercise trainer

De: face‐to‐face

Du: 38 sessions in 12 weeks

Exercise (16)

M: group

P: exercise physiologists (BA level or above)

De: face‐to‐face

Du: three times a week (one hour) for 12 weeks

Maurer 1999

US

Outpatients

Education (56)

M: group

P: healthcare professionals

De: face‐to‐face

Du: four sessions in eight weeks

Exercise (57)

M: not specified

P: not specified

De: face‐to‐face

Du: three times a week for eight weeks

McKnight 2010

US

General population

Self‐management group (87)

M: group/individual

P: healthcare professionals

De: face‐to‐face/telephone

Du: 12 sessions in 12 weeks/staggering calls

Strength training (91)

M: not specified

P: expert physical trainers

De: face‐to‐face

Du: three weekly sessions for nine months (phase 1), then contact every two weeks in first six weeks, then monthly for a total of 15 months (phase 2)

McKnight 2010

US

General population

Combination group (95)

M: group/individual

P: healthcare professionals and physiotherapists

De: face‐to‐face/telephone

Du: 48 sessions in 12 weeks/staggering calls

Strength training (91)

M: not specified

P: expert physical trainers

De: face‐to‐face, telephone

Du: three weekly sessions for nine months (phase 1), then contact every two weeks in first six weeks, then monthly for a total of 15 months (phase 2)

Murphy 2008

US

Senior housing facilities

Exercise + activity strategy training (28)

M: group + one individual session

P: occupational therapists

De: face‐to‐face

Du: eight sessions (1.5 hours) in four weeks

Exercise + education (26)

M: group

P: health education interventionists

De: educational materials

Du: two sessions per week (1.5 hours) for four weeks

M = Mode, P = Personnel, De = Delivery method, Du = Duration.
Figures and Tables -
Table 3. Summary of comparisons
Navigate to table in Review
Table 4. heiQ‐components addressed in interventions

Intervention

Health‐directed activity

Positive and active engagement in life

Emotional distress

Self‐monitoring and insight

Constructive attitudes and approaches

Skill and technique acquisition

Social integration and support

Health service navigation

TOTAL + (%)

29/34

85%

9/34

26%

13/34

38%

27/34

79%

15/34

44%

32/34

94%

4/34

12%

11/34

32%

Allen 2010

+

‐

‐

+

+

+

‐

+

Berman 2004

+

U

+

‐

+

+

‐

+

Blixen 2004

+

‐

+

+

‐

+

‐

+

Buszewicz 2006

+

+

+

+

+

+

U

+

Calfas 1992

+

+

+

+

+

+

‐

‐

Cronan 1997 (education)

+

‐

‐

+

+

+

‐

+

Cronan 1997 (combination)

+

‐

‐

+

+

+

‐

+

Crotty 2009

‐

‐

‐

+

+

‐

‐

‐

Hansson 2010

+

‐

‐

+

‐

+

‐

‐

Heuts 2005

+

‐

+

+

‐

+

U

+

Hopman‐Rock 2000

+

‐

‐

+

+

+

‐

‐

Hughes 2004

+

‐

‐

‐

‐

+

‐

‐

Hurley 2007 (Indiv Rehab)

+

+

+

+

+

+

‐

‐

Hurley 2007 (Group Rehab)

+

+

+

+

+

+

‐

‐

Jessep 2009

+

+

+

+

+

+

‐

‐

Keefe 2004

‐

‐

‐

+

+

+

+

‐

Keefe 1996 (CST)

‐

‐

‐

+

+

+

‐

‐

Keefe 1996 (SA‐CST)

‐

‐

‐

+

+

+

+

‐

Keefe 1990

‐

‐

‐

+

+

+

‐

‐

Lorig 2008

+

+

+

+

+

+

‐

+

Maisiak 1996

+

‐

‐

+

‐

+

‐

+

Martire 2007 (PES)

+

+

+

+

+

+

‐

‐

Martire 2007 (CES)

+

+

+

+

+

+

+

‐

Maurer 1999

+

‐

‐

+

‐

+

‐

‐

Mazzuca 1997

+

‐

‐

‐

‐

+

‐

‐

Mazzuca 2004

+

‐

‐

‐

‐

+

‐

‐

McKnight 2010 (SMP)

+

‐

‐

+

+

+

‐

‐

McKnight 2010 (combination)

+

‐

‐

+

+

+

‐

‐

Murphy 2008

+

‐

‐

‐

+

+

‐

‐

Nunez 2006

+

‐

‐

‐

‐

+

‐

‐

Ackerman 2012

+

‐

+

+

+

+

‐

+

Victor 2005

+

‐

‐

‐

+

+

+

‐

Wetzels 2005

+

+

‐

+

+

‐

‐

‐

Yip 2007

+

‐

+

+

+

+

‐

+

+ = heiQ component addressed, ‐ = heiQ component not addressed, U = Unclear whether heiQ component was addressed

Figures and Tables -
Table 4. heiQ‐components addressed in interventions
Navigate to table in Review
Comparison 1. SMP versus attention control

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

1 Self‐management of OA Show forest plot

1

Mean Difference (IV, Random, 95% CI)

Totals not selected

2 Pain Show forest plot

3

Std. Mean Difference (Random, 95% CI)

Subtotals only

2.1 Short term

1

Std. Mean Difference (Random, 95% CI)

‐0.62 [‐1.11, ‐0.13]

2.2 Intermediate term

3

Std. Mean Difference (Random, 95% CI)

‐0.26 [‐0.44, ‐0.09]

3 Global OA scores Show forest plot

1

Mean Difference (IV, Random, 95% CI)

Totals not selected

4 Function—self‐reported Show forest plot

3

Std. Mean Difference (Random, 95% CI)

Subtotals only

4.1 Short term

1

Std. Mean Difference (Random, 95% CI)

‐0.13 [‐0.49, 0.23]

4.2 Intermediate term

3

Std. Mean Difference (Random, 95% CI)

‐0.19 [‐0.50, 0.11]

5 Quality of life Show forest plot

1

Mean Difference (IV, Random, 95% CI)

Totals not selected

6 Withdrawals Show forest plot

5

937

Risk Ratio (M‐H, Random, 95% CI)

1.11 [0.78, 1.57]

7 Emotional distress Show forest plot

2

Std. Mean Difference (IV, Random, 95% CI)

Subtotals only

7.1 Short term

1

68

Std. Mean Difference (IV, Random, 95% CI)

‐0.37 [‐0.85, 0.11]

7.2 Intermediate term

2

409

Std. Mean Difference (IV, Random, 95% CI)

0.02 [‐0.18, 0.21]
Figures and Tables -
Comparison 1. SMP versus attention control
Navigate to table in Review
Comparison 2. SMP versus usual care/no treatment/wait list

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

1 Self‐management of OA Show forest plot

11

Std. Mean Difference (Random, 95% CI)

Subtotals only

1.1 Short term

5

721

Std. Mean Difference (Random, 95% CI)

0.22 [‐0.00, 0.45]

1.2 Intermediate term

10

1647

Std. Mean Difference (Random, 95% CI)

0.14 [0.00, 0.27]

1.3 Long term

1

195

Std. Mean Difference (Random, 95% CI)

0.23 [‐0.05, 0.51]

2 Engagement in life Show forest plot

3

Std. Mean Difference (IV, Random, 95% CI)

Subtotals only

2.1 Short term

1

143

Std. Mean Difference (IV, Random, 95% CI)

‐0.23 [‐0.57, 0.11]

2.2 Intermediate term

3

357

Std. Mean Difference (IV, Random, 95% CI)

0.01 [‐0.20, 0.21]

3 Pain Show forest plot

14

Std. Mean Difference (Random, 95% CI)

Subtotals only

3.1 Short term

6

Std. Mean Difference (Random, 95% CI)

‐0.26 [‐0.41, ‐0.10]

3.2 Intermediate term

13

Std. Mean Difference (Random, 95% CI)

‐0.17 [‐0.26, ‐0.08]

3.3 Long term

1

Std. Mean Difference (Random, 95% CI)

‐0.18 [‐0.45, 0.09]

4 Global OA scores Show forest plot

7

Std. Mean Difference (Random, 95% CI)

Subtotals only

4.1 Short term

2

Std. Mean Difference (Random, 95% CI)

‐0.34 [‐0.59, ‐0.09]

4.2 Intermediate term

7

Std. Mean Difference (Random, 95% CI)

‐0.28 [‐0.39, ‐0.17]

4.3 Long term

1

Std. Mean Difference (Random, 95% CI)

‐0.29 [‐0.56, ‐0.02]

5 Function—self‐reported Show forest plot

13

Std. Mean Difference (Random, 95% CI)

Subtotals only

5.1 Short term

5

Std. Mean Difference (Random, 95% CI)

‐0.01 [‐0.19, 0.18]

5.2 Intermediate term

13

Std. Mean Difference (Random, 95% CI)

‐0.16 [‐0.25, ‐0.08]

5.3 Long term

1

Std. Mean Difference (Random, 95% CI)

‐0.27 [‐0.55, 0.01]

6 Function—performance Show forest plot

2

Std. Mean Difference (Random, 95% CI)

Subtotals only

6.1 Short term

1

Std. Mean Difference (Random, 95% CI)

0.33 [‐0.07, 0.73]

6.2 Intermediate term

2

Std. Mean Difference (Random, 95% CI)

0.06 [‐0.24, 0.36]

7 Quality of life Show forest plot

8

Std. Mean Difference (Random, 95% CI)

Subtotals only

7.1 Short term

2

Std. Mean Difference (Random, 95% CI)

0.14 [‐0.47, 0.75]

7.2 Intermediate term

8

Std. Mean Difference (Random, 95% CI)

0.03 [‐0.08, 0.14]

7.3 Long term

2

Std. Mean Difference (Random, 95% CI)

0.10 [‐0.10, 0.31]

8 Withdrawals Show forest plot

16

3738

Risk Ratio (M‐H, Random, 95% CI)

0.99 [0.74, 1.33]

9 Emotional distress Show forest plot

9

Std. Mean Difference (Random, 95% CI)

Subtotals only

9.1 Short term

3

Std. Mean Difference (Random, 95% CI)

0.01 [‐0.44, 0.45]

9.2 Intermediate term

8

Std. Mean Difference (Random, 95% CI)

0.11 [‐0.06, 0.28]

10 Health‐directed activity Show forest plot

3

Std. Mean Difference (Random, 95% CI)

Subtotals only

10.1 Short term

1

Std. Mean Difference (Random, 95% CI)

0.67 [0.37, 0.97]

10.2 Intermediate term

3

Std. Mean Difference (Random, 95% CI)

0.25 [0.05, 0.46]

11 Skill and technique acquisition Show forest plot

1

Mean Difference (IV, Random, 95% CI)

Totals not selected

12 Constructive attitudes and approaches Show forest plot

2

Std. Mean Difference (IV, Random, 95% CI)

Totals not selected

12.1 Short term

1

Std. Mean Difference (IV, Random, 95% CI)

0.0 [0.0, 0.0]

12.2 Intermediate term

1

Std. Mean Difference (IV, Random, 95% CI)

0.0 [0.0, 0.0]

13 Social integration and support Show forest plot

3

Std. Mean Difference (IV, Random, 95% CI)

Subtotals only

13.1 Short term

1

143

Std. Mean Difference (IV, Random, 95% CI)

‐0.19 [‐0.52, 0.15]

13.2 Intermediate term

3

357

Std. Mean Difference (IV, Random, 95% CI)

‐0.08 [‐0.30, 0.14]

14 Health service navigation Show forest plot

2

Std. Mean Difference (Random, 95% CI)

0.15 [‐0.03, 0.34]
Figures and Tables -
Comparison 2. SMP versus usual care/no treatment/wait list
Navigate to table in Review
Comparison 3. SMP versus information only

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

1 Self‐management of OA Show forest plot

3

Std. Mean Difference (IV, Random, 95% CI)

Subtotals only

1.1 Short term

1

90

Std. Mean Difference (IV, Random, 95% CI)

0.06 [‐0.36, 0.47]

1.2 Intermediate term

3

760

Std. Mean Difference (IV, Random, 95% CI)

0.20 [‐0.04, 0.44]

2 Engagement in life Show forest plot

1

Mean Difference (IV, Random, 95% CI)

Totals not selected

2.1 Short term

1

Mean Difference (IV, Random, 95% CI)

0.0 [0.0, 0.0]

2.2 Intermediate term

1

Mean Difference (IV, Random, 95% CI)

0.0 [0.0, 0.0]

3 Pain Show forest plot

3

751

Std. Mean Difference (IV, Random, 95% CI)

‐0.07 [‐0.21, 0.08]

4 Global OA scores Show forest plot

3

Std. Mean Difference (IV, Random, 95% CI)

Subtotals only

4.1 Short term

1

89

Std. Mean Difference (IV, Random, 95% CI)

0.09 [‐0.33, 0.50]

4.2 Intermediate term

3

751

Std. Mean Difference (IV, Random, 95% CI)

‐0.06 [‐0.28, 0.16]

5 Function—self‐reported Show forest plot

4

854

Std. Mean Difference (IV, Random, 95% CI)

‐0.09 [‐0.22, 0.05]

6 Function—performance Show forest plot

1

Mean Difference (IV, Random, 95% CI)

Totals not selected

7 Quality of life Show forest plot

2

648

Std. Mean Difference (IV, Random, 95% CI)

0.05 [‐0.10, 0.21]

8 Withdrawals Show forest plot

4

1251

Risk Ratio (M‐H, Random, 95% CI)

1.60 [0.75, 3.40]

9 Emotional distress Show forest plot

3

775

Std. Mean Difference (IV, Random, 95% CI)

‐0.00 [‐0.30, 0.30]

10 Health‐directed activity Show forest plot

1

Mean Difference (IV, Random, 95% CI)

Totals not selected

10.1 Short term

1

Mean Difference (IV, Random, 95% CI)

0.0 [0.0, 0.0]

10.2 Intermediate term

1

Mean Difference (IV, Random, 95% CI)

0.0 [0.0, 0.0]

11 Social integration and support Show forest plot

2

Std. Mean Difference (IV, Random, 95% CI)

Subtotals only

11.1 Short term

1

90

Std. Mean Difference (IV, Random, 95% CI)

‐0.02 [‐0.44, 0.40]

11.2 Intermediate term

2

181

Std. Mean Difference (IV, Random, 95% CI)

‐0.02 [‐0.39, 0.35]

12 Health service navigation Show forest plot

1

Mean Difference (IV, Random, 95% CI)

Totals not selected

12.1 Short term

1

Mean Difference (IV, Random, 95% CI)

0.0 [0.0, 0.0]

12.2 Intermediate term

1

Mean Difference (IV, Random, 95% CI)

0.0 [0.0, 0.0]

13 Skill and technique acquisition Show forest plot

1

Mean Difference (IV, Random, 95% CI)

Totals not selected

13.1 Short term

1

Mean Difference (IV, Random, 95% CI)

0.0 [0.0, 0.0]

13.2 Intermediate term

1

Mean Difference (IV, Random, 95% CI)

0.0 [0.0, 0.0]

14 Constructive attitudes and approaches Show forest plot

1

Mean Difference (IV, Random, 95% CI)

Totals not selected

14.1 Short term

1

Mean Difference (IV, Random, 95% CI)

0.0 [0.0, 0.0]

14.2 Intermediate term

1

Mean Difference (IV, Random, 95% CI)

0.0 [0.0, 0.0]
Figures and Tables -
Comparison 3. SMP versus information only
Navigate to table in Review
Comparison 4. SMP versus non‐SMP intervention

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

1 Self‐management of OA Show forest plot

3

Std. Mean Difference (IV, Random, 95% CI)

Subtotals only

1.1 Short term

3

186

Std. Mean Difference (IV, Random, 95% CI)

0.42 [‐0.05, 0.89]

1.2 Intermediate term

1

70

Std. Mean Difference (IV, Random, 95% CI)

0.54 [0.05, 1.04]

2 Pain Show forest plot

5

Std. Mean Difference (Random, 95% CI)

Subtotals only

2.1 Short term

5

Std. Mean Difference (Random, 95% CI)

0.03 [‐0.29, 0.36]

2.2 Intermediate term

2

Std. Mean Difference (Random, 95% CI)

‐0.18 [‐0.56, 0.19]

3 Global OA scores Show forest plot

1

Std. Mean Difference (Random, 95% CI)

Totals not selected

3.1 Short term

1

Std. Mean Difference (Random, 95% CI)

0.0 [0.0, 0.0]

4 Function—self‐reported Show forest plot

3

Std. Mean Difference (Random, 95% CI)

Subtotals only

4.1 Short term

3

Std. Mean Difference (Random, 95% CI)

0.23 [‐0.03, 0.48]

4.2 Intermediate term

2

Std. Mean Difference (Random, 95% CI)

‐0.17 [‐0.54, 0.20]

5 Function—performance Show forest plot

1

Std. Mean Difference (Random, 95% CI)

Totals not selected

6 Quality of life Show forest plot

2

Std. Mean Difference (IV, Random, 95% CI)

Subtotals only

6.1 Short term

1

57

Std. Mean Difference (IV, Random, 95% CI)

0.24 [‐0.28, 0.76]

6.2 Intermediate term

2

243

Std. Mean Difference (IV, Random, 95% CI)

‐0.01 [‐0.28, 0.26]

6.3 Long term

1

178

Std. Mean Difference (IV, Random, 95% CI)

‐0.23 [‐0.55, 0.10]

7 Withdrawals Show forest plot

7

919

Risk Ratio (M‐H, Random, 95% CI)

0.86 [0.69, 1.09]

8 Emotional distress Show forest plot

3

Std. Mean Difference (IV, Random, 95% CI)

Subtotals only

8.1 Short term

3

192

Std. Mean Difference (IV, Random, 95% CI)

0.13 [‐0.28, 0.55]

8.2 Intermediate term

2

118

Std. Mean Difference (IV, Random, 95% CI)

0.18 [‐0.19, 0.55]

9 Constructive attitudes and approaches Show forest plot

2

Std. Mean Difference (IV, Random, 95% CI)

Subtotals only

9.1 Short term

2

135

Std. Mean Difference (IV, Random, 95% CI)

0.92 [0.49, 1.34]

9.2 Intermediate term

1

70

Std. Mean Difference (IV, Random, 95% CI)

0.62 [0.12, 1.12]
Figures and Tables -
Comparison 4. SMP versus non‐SMP intervention
Navigate to table in Review
Comparison 5. SMP versus acupuncture

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

1 Global OA scores Show forest plot

1

Std. Mean Difference (Random, 95% CI)

Totals not selected

1.1 Short term

1

Std. Mean Difference (Random, 95% CI)

0.0 [0.0, 0.0]

1.2 Intermediate term

1

Std. Mean Difference (Random, 95% CI)

0.0 [0.0, 0.0]

2 Pain Show forest plot

1

Std. Mean Difference (Random, 95% CI)

Totals not selected

2.1 Short term

1

Std. Mean Difference (Random, 95% CI)

0.0 [0.0, 0.0]

2.2 Intermediate term

1

Std. Mean Difference (Random, 95% CI)

0.0 [0.0, 0.0]

3 Function self‐reported Show forest plot

1

Std. Mean Difference (Random, 95% CI)

Totals not selected

3.1 Short term

1

Std. Mean Difference (Random, 95% CI)

0.0 [0.0, 0.0]

3.2 Intermediate term

1

Std. Mean Difference (Random, 95% CI)

0.0 [0.0, 0.0]

4 Function performance Show forest plot

1

Std. Mean Difference (Random, 95% CI)

Totals not selected

5 Withdrawals Show forest plot

1

Risk Ratio (M‐H, Random, 95% CI)

Totals not selected
Figures and Tables -
Comparison 5. SMP versus acupuncture
Navigate to table in Review
Comparison 6. Subgroup analysis

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

1 Self‐management in OA Show forest plot

7

Std. Mean Difference (Random, 95% CI)

0.14 [‐0.08, 0.36]

1.1 White, educated, female

3

Std. Mean Difference (Random, 95% CI)

0.29 [0.07, 0.50]

1.2 NOT white, educated, female

4

Std. Mean Difference (Random, 95% CI)

0.03 [‐0.29, 0.36]

2 Function self‐reported Show forest plot

8

Std. Mean Difference (Random, 95% CI)

‐0.12 [‐0.23, ‐0.01]

2.1 White, educated, female

3

Std. Mean Difference (Random, 95% CI)

‐0.20 [‐0.37, ‐0.02]

2.2 NOT white, educated, female

5

Std. Mean Difference (Random, 95% CI)

‐0.06 [‐0.21, 0.08]

3 Pain Show forest plot

9

Std. Mean Difference (Random, 95% CI)

‐0.17 [‐0.28, ‐0.06]

3.1 White, educated, female

4

Std. Mean Difference (Random, 95% CI)

‐0.11 [‐0.30, 0.07]

3.2 NOT white, educated, female

5

Std. Mean Difference (Random, 95% CI)

‐0.20 [‐0.35, ‐0.05]

4 Withdrawals Show forest plot

12

3095

Risk Ratio (M‐H, Random, 95% CI)

1.04 [0.81, 1.34]

4.1 White, educated, female

7

2129

Risk Ratio (M‐H, Random, 95% CI)

1.05 [0.76, 1.45]

4.2 NOT white, educated, female

5

966

Risk Ratio (M‐H, Random, 95% CI)

1.04 [0.63, 1.69]
Figures and Tables -
Comparison 6. Subgroup analysis
Navigate to table in Review