Intervention Review

You have free access to this content

Self-management education programmes for osteoarthritis

  1. Féline PB Kroon1,†,
  2. Lennart RA van der Burg1,‡,
  3. Rachelle Buchbinder2,*,
  4. Richard H Osborne3,
  5. Renea V Johnston2,
  6. Veronica Pitt4

Editorial Group: Cochrane Musculoskeletal Group

Published Online: 15 JAN 2014

Assessed as up-to-date: 17 JAN 2013

DOI: 10.1002/14651858.CD008963.pub2


How to Cite

Kroon FPB, van der Burg LRA, Buchbinder R, Osborne RH, Johnston RV, Pitt V. Self-management education programmes for osteoarthritis. Cochrane Database of Systematic Reviews 2014, Issue 1. Art. No.: CD008963. DOI: 10.1002/14651858.CD008963.pub2.

Author Information

  1. 1

    Leiden University Medical Center, Department of Rheumatology, Leiden, Netherlands

  2. 2

    Department of Epidemiology and Preventive Medicine, School of Public Health and Preventive Medicine, Monash University, Monash Department of Clinical Epidemiology, Cabrini Hospital, Malvern, Victoria, Australia

  3. 3

    Deakin University, Director, Public Health Innovation, Deakin, Burwood, Victoria, Australia

  4. 4

    National Trauma Research Institute, The Alfred Hospital, Monash University, Melbourne, Victoria, Australia

  1. Joint first author

  2. Joint first author

*Rachelle Buchbinder, Monash Department of Clinical Epidemiology, Cabrini Hospital, Department of Epidemiology and Preventive Medicine, School of Public Health and Preventive Medicine, Monash University, Suite 41, Cabrini Medical Centre, 183 Wattletree Road, Malvern, Victoria, 3144, Australia. rachelle.buchbinder@monash.edu.

Publication History

  1. Publication Status: New
  2. Published Online: 15 JAN 2014

SEARCH

 
Characteristics of included studies [ordered by study ID]
Ackerman 2012

MethodsStudy design: RCT (blocks four to six, stratified by site), multi-centre, two arms, non-blinded

Country in which trial was carried out:Australia

Method of recruitment of participants:referral by orthopaedic surgeons or rheumatologists

Setting:outpatients (secondary and tertiary care)

Was the sample size justified with a priori calculation of effect size/power?Yes

Length of follow-up:12 months

Dropouts:14 (24%) dropped out from the intervention group (11 were unable to be allocated to a course or had their course cancelled, one did not attend the scheduled course, one was scheduled for joint replacement, one died); two (3%) dropped out from the control group (no reasons provided)


ParticipantsInclusion criteria

  • Hip or knee OA diagnosis from radiology reports or able to be classified according to ACR criteria
  • Aged 18 years or over
  • Referred to an orthopaedic surgeon or rheumatologist
  • Sufficient English language skills and vision to self-complete questionnaires
  • A reasonable expectation of attending six sessions of the ASMP


Exclusion criteria

  • Cognitive dysfunction
  • Previous participation in an ASMP or similar education programme
  • Placement on an orthopaedic waiting list for joint replacement surgery
  • Scheduled joint replacement surgery


Baseline characteristics

Baseline characteristics were similar among all treatment groups

Intervention group: ASMP (N = 58 randomly assigned, 39 analysed after six weeks, 40 analysed after three months, 38 analysed after 12 months of follow-up (FU))

Location of OA: 36% hip, 57% knee, 7% hip and knee

BMI (mean (IQR)): 30 (24 to 35) kg/m2

PROGRESS-Plus

Place of residence: Australia

Race, ethnicity and culture: 69% Australian-born

Occupation: 29% paid employment, 59% retired, 7% not working because of OA or medical condition, 5% unemployed

Sex: 38% male, 62% female

Education: 12% primary school or less, 47% years seven to 10, 14% years 11 and 12, 11% trade/technical education, 16% university

Social capital: 62% married or living with partner

Age (mean (SD)), years: 63.5 (10.8)

Control group: ASMP booklet (N = 62 randomly assigned, 51 analysed after six weeks, 55 analysed after three months, 56 analysed after 12 months of FU)

Location of OA: 26% hip, 68% knee, 6% hip and knee

BMI (mean (IQR)): 29 (26 to 35) kg/m2

PROGRESS-Plus

Place of residence: Australia

Race, ethnicity and culture: 68% Australian-born

Occupation: 23% paid employment, 65% retired, 10% not working because of OA or medical condition, 2% unemployed

Sex: 42% male, 58% female

Education: 12% primary school or less, 45% years seven to 10, 17% years 11 and 12, 18% trade/technical education, 8% university

Social capital: 65% married or living with partner

Age (mean (SD)), years: 66.6 (10.9)


InterventionsIntervention: ASMP

Description: The Stanford ASMP covers management of pain and fatigue, physical activity, managing emotions, health-related problem solving and communication with doctors. Participants also received a copy of the arthritis self-help book

Intended audience: people with OA of the knee or hip

Mode: group sessions

Personnel: one peer leader, one healthcare professional

Delivery method: face-to-face

Language: English

Format: standard format

Location: in community-based and hospital locations

Duration: one session per week, lasting 2.5 hours, for six weeks

Additional treatment during trial: unclear

Were the following heiQ components addressed?

Health-directed behaviour: yes

Positive and active engagement in life: no

Emotional well-being: yes

Self-monitoring and insight: yes

Constructive attitudes and approaches: yes

Skill and technique acquisition: yes

Social integration and support: No

Health service navigation: yes 

Comparator

Type: education only

Description: Participants were mailed a copy of the arthritis self-help book. No advice was given regarding use of the book

Additional treatment during trial: unclear


OutcomesOutcome assessed at: baseline, six weeks and three and 12 months of follow-up

Primary outcomes of study

  • Quality of life (assessment of quality of life (AQoL), -0.04 to 1.00, higher score is better)


Secondary outcomes of study

  • Health-directed activities (heiQ health-directed activities subscale, 1 to 6, higher score is better)
  • Positive and active engagement in life (heiQ positive and active engagement in life subscale, 1 to 6, higher score is better)
  • Skill and technique acquisition (heiQ skill and technique acquisition subscale, 1 to 6, higher score is better)
  • Constructive attitudes and approaches (heiQ constructive attitudes and approaches subscale, 1 to 6, higher score is better)
  • Self-management (heiQ self-monitoring and insight subscale, 1 to 6, higher score is better)
  • Health service navigation (heiQ health service navigation subscale, 1 to 6, higher score is better)
  • Social integration and support (heiQ social integration and support subscale, 1 to 6, higher score is better)
  • Emotional distress (heiQ emotional distress subscale, 1 to 6, higher score is better)
  • Global OA scores (WOMAC subscales pain 0 to 20, stiffness 0 to 8, function 0 to 68, lower scores are better)
  • Emotional distress (Kessler Psychological Distress Scale (K10), 10 to 50, lower score is better)
  • Global OA scores (disease severity and need for surgery on the hip and knee multi-attribute priority tool (MAPT), 0 to 100, lower score is better)


NotesWe extracted the following outcomes at six weeks (short term) and at 12 months (intermediate term) for the analyses in this review: self-management (heiQ subscale self-monitoring and insight), engagement in life (heiQ subscale positive and active engagement in life), pain (WOMAC subscale pain), global OA scores (MAPT), function self-reported (WOMAC subscale function), quality of life (AQoL), emotional distress (K10), health-directed activity (heiQ subscale health-directed activity), social integration and support (heiQ subscale social integration and support), health service navigation (heiQ subscale health service navigation), skill and technique acquisition (heiQ subscale skill and technique acquisition), constructive attitudes and approaches (heiQ subscale constructive attitudes and approaches) and dropouts (proportion of missing participants)

The study was funded by the National Health and Medical Research Council of Australia (Project Grant number 400210); Dr Ackerman was supported in part by an Australian National Health and Medical Research Council Public Health (Australia) Training Fellowship (#520004); Prof Buchbinder was supported in part by Australian National Health and Medical Research Council Practitioner Fellowships (#334010 and #606429); Prof Osborne was supported in part by an Australian National Health and Medical Research Council Population Health Career Development Award (#400391)

The author (I Ackerman) provided the unpublished manuscript and additional information about the trial on request

The prior calculated sample size was not reached; therefore, a type II error has possibly occurred


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low riskQuote: “For each site, group allocation was assigned using a computer-generated random list in permuted blocks of 4 or 6”

Comment: This method of random sequence generation has low risk of introducing bias

Allocation concealment (selection bias)Low riskQuote: “Group allocation was concealed using opaque sealed envelopes, with individual envelopes opened at the coordinating centre by a research assistant not associated with the study and verified by an independent observer”

Comment: This method of allocation concealment has low risk of introducing bias

Blinding of participants and personnel (performance bias)
All outcomes
High riskQuote: “Participants and investigators were not blinded”

Comment: Blinding of participants to the intervention was not feasible, with high risk of biasing results. Although personnel who conducted the intervention were not blinded either, risk of bias remains low, as the control intervention did not come in contact with study personnel

Blinding of outcome assessment (detection bias)
All outcomes
High riskComment: No information provided on blinding of outcome assessment; however, most outcomes are subjective, and participants are not blinded to group allocation introducing a risk of detection bias

Incomplete outcome data (attrition bias)
All outcomes
High riskQuote: “Statistical analysis was performed using all randomised participants who provided at least one post-baseline assessment. Intervention group participants who did not receive the allocated intervention were not included in post-baseline analyses”

Comment: The number of dropouts in both groups differs substantially (3% in the control vs 24% in the intervention group), with a reasonably high number of dropouts in the intervention group. An intention-to-treat analysis was performed

Selective reporting (reporting bias)Low riskComment: All outcomes listed in methods or in the protocol have been reported in the results

Other biasHigh riskQuote: “Of the intervention group (n = 58), 44 (76%) participants received the intervention as allocated. Of those who commenced the ASMP, only 21 participants (47%) attended all sessions. The median (IQR) number of sessions attended was 5 (4-6)”

Comment: Participants who commenced the ASMP had quite high compliance. However, a large number of the intervention group did not receive the allocated intervention (24%), thereby introducing a risk of bias

Allen 2010

MethodsStudy design:RCT (block randomisation, stratified by race), single-centre, three arms, outcome assessment blinded

Country in which trial was carried out:US

Method of recruitment of participants:DVAMC's electronic medical record system.

Setting: primary care (Veterans Affairs Medical Center)

Was the sample size justified with a priori calculation of effect size/power?yes

Length of follow-up: 12 months

Dropouts: 28 (16%) dropped out from the self-management education programme (seven excluded, nine lost to follow-up, 12 withdrew), 17 (10%) dropped out from health education (four excluded, nine lost to follow-up, four withdrew) and 17 (10%) dropped out from usual care (six excluded, six lost to follow-up, five withdrew)


ParticipantsCriteria for defining the condition being treated (OA)

  • Physician diagnosis of hip or knee
  • Radiographic evidence of knee/hip OA


Inclusion criteria

  • A physician's diagnosis of hip or knee OA with radiographic evidence
  • Enrolled in primary care at the Durham VAMC
  • Current and persistent joint symptoms
  • A physician visit during the study period


Exclusion criteria

  • Diagnosis of other systemic rheumatic disease (e.g. RA, FM)
  • Hospitalised for cardiovascular disease (e.g. stroke, MI)
  • Diagnosis of metastatic cancer in past three months
  • Active diagnosis of psychosis or diagnosis of dementia
  • Any other serious health condition that limits participation
  • On waiting list for arthroplasty
  • Resident in a nursing home
  • Severely impaired hearing/speech
  • No access to a telephone
  • Participation in another interventional study


Baseline characteristics

Baseline characteristics were similar in all treatment groups

Intervention group: osteoarthritis self-management Intervention (N = 174 randomly assigned, 172 analysed)

Location of OA: 82% knee, 12% hip, 6% knee and hip

BMI (mean (SD)): 32.0 (7.0) kg/m2

Duration of OA (mean (SD)): 16.5 (12.7) years

PROGRESS-Plus

Place of residence: US

Race, ethnicity and culture: 55% white, 45% non-white, 1% Hispanic

Occupation: 38% employed

Sex: 91% male, 9% female

Education: 33% high school education or less

Socioeconomic status: 28% inadequate income

Social capital: 72% married

Age (mean (SD)), years: 60.3 (10.3)

Disability: 30% fair or poor health

Control group: health education intervention (N = 175 randomly assigned, 172 analysed)

Location of OA: 79% knee, 16% hip, 5% knee and hip

BMI (mean (SD)): 31.6 (6.5) kg/m2

Duration of OA (mean (SD)): 15.8 (12.0) years

PROGRESS-Plus

Place of residence: US

Race, ethnicity and culture: 53% white, 47% non-white, 3% Hispanic

Occupation: 38% employed

Sex: 93% male, 7% female

Education: 34% high school education or less

Socioeconomic status: 27% inadequate income

Social capital: 65% married

Age (mean (SD)), years: 60.3 (10.8)

Disability: 37% fair or poor health

Control group: usual care (N = 174 randomly assigned, 171 analysed)

Location of OA: 79% knee, 17% hip, 4% knee and hip

BMI (mean (SD)): 31.8 (6.5) kg/m2

Duration of OA (mean (SD)): 15.9 (11.9) years

PROGRESS-Plus

Place of residence: US

Race, ethnicity and culture: 54% white, 46% non-white, 2% Hispanic

Occupation: 39% employed

Sex: 94% male, 6% female

Education: 33% high school education or less

Socioeconomic status: 22% inadequate income

Social capital: 71% married

Age (mean (SD)), years: 59.7 (10.1)

Disability: 30% fair or poor health

Note:Several participants were excluded from analysis because they did not meet eligibility criteria after subsequent medical record review (two in the intervention group, three in the health education group and three in the usual care group)


InterventionsIntervention: osteoarthritis self-management intervention

Description: The intervention included two main components: providing education related to managing OA symptoms, and helping participants develop goals and action plans related to OA management. Participants were asked to identify and write down one or more goals related to their OA symptoms and management, as well as weekly action plans for achieving these goals

Intended audience: people with OA of the knee and/or hip

Mode: individual

Personnel: health professionals

Delivery method: telephone, written, audio and video material

Language: English

Format: tailored to individual’s needs

Location: home

Duration: monthly telephone calls for 12 months, mean duration of calls 9.0 minutes

Additional treatment during trial: usual care for OA (incl analgesic and anti-inflammatory medication)

Were the following heiQ components addressed?

Health-directed behaviour: yes

Positive and active engagement in life: no

Emotional well-being: no

Self-monitoring and insight: yes

Constructive attitudes and approaches: yes

Skill and technique acquisition: yes

Social integration and support: no

Health service navigation: yes

Comparator: health education intervention

Type: attention control

Description: Participants received written and audio materials regarding common health problems, as well as related screening recommendations. The health educator called participants to review key points from the modules, to assess whether participants were being appropriately screened and to make suggestions for screening as needed

Frequency: monthly telephone calls for 12 months, mean duration of calls 4.9 minutes

Additional treatment during trial: usual care for OA (incl analgesic and anti-inflammatory medication)

Comparator: usual care

Type: usual care

Description: Participants received their usual care for OA


OutcomesOutcome assessed at: baseline and 12 months of follow-up

Primary outcomes of study

  • Pain (Arthritis Impact Measurement Scales-2 (AIMS-2) subscale pain, 0 to 10, lower score is better)


Secondary outcomes of study

  • Function self-reported (AIMS-2 physical function subscale, 0 to 10, lower score is better)
  • Emotional distress (AIMS-2 affect subscale, 0 to 10, lower score is better)
  • Self-management (Arthritis Self-Efficacy Scale (ASES), 1 to 10, higher score is better)
  • Pain (VAS, 0 to 10, lower score is better)


NotesWe extracted the following outcomes at 12 months (intermediate term) for the analyses in this review: self-management (ASES), pain (VAS), function self-reported (AIMS-2), emotional distress (AIMS-2) and dropouts (proportion of missing participants)

Funding was provided by the US Department of Veterans Affairs Health Services Research and Development Service

Author (K Allen) provided additional information about the trial on request

The study sample consisted mainly of male veterans, which is not a representative sample of general primary care

Participants were reimbursed $10 after baseline assessment and $10 after follow-up assessment


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low riskQuote: “Randomisation is computer generated, stratified by race (white vs. non-white) using block sized of 12 (…).”

Comment: correct method of randomisation with low risk of bias

Allocation concealment (selection bias)Low riskQuote: “Randomisation was (…) maintained separately from participant enrolment (…)."

Comment: The randomisation sequence was adequately concealed during participant enrolment

Blinding of participants and personnel (performance bias)
All outcomes
High riskQuote: “(…) study coordinator informs participants of their group assignment (…)."

Comment: Participants were informed about their allocated treatment before the start of the intervention. Personnel were not adequately blinded either and administered both interventions

Blinding of outcome assessment (detection bias)
All outcomes
High riskQuote: “(…) research assistants conducting the assessments are blinded to participants’ group"

Comment: Personnel assessing study outcomes were blinded; however, most outcomes are subjective and participants are not blinded to group allocation

Incomplete outcome data (attrition bias)
All outcomes
High riskQuote: “(…) we fit linear mixed models by using an intention-to-treat approach”

Comment: An intention-to-treat-analysis without imputation techniques was used. The number of withdrawals is more than twofold higher in the SMP intervention group compared with the health education or usual care group (15.1% vs 8.1% vs 8.2%). Reasons for dropout and postintervention exclusions are not reported

Selective reporting (reporting bias)Unclear riskComment: The authors did not report the self-management behaviours, analgesic/anti-inflammatory medication use and intervention adherence/intensity that were mentioned in the previously published protocol. All prespecified outcomes that we considered important for the review were reported

Other biasHigh riskQuote: “We assumed a common baseline value among treatment groups”

Comment: Incorrect assumption that baseline values are similar after randomisation in all treatment groups

Comment: The same health educator conducts the telephone calls in the SMP and the health education group. Authors attempted to minimise the risk of contamination by using standardised scripts

Berman 2004

MethodsStudy design:RCT, multi-centre, three arms, outcome assessment and two treatment arms blinded

Country in which trial was carried out:US

Method of recruitment of participants:through print and radio advertisements

Setting:general population

Was the sample size justified with a priori calculation of effect size/power?yes

Length of follow-up:26 weeks (6.5 months)

Dropouts: 50 (26%) in the true acupuncture group dropped out (17 were disqualified for medical reasons, 33 withdrew), 52 (27%) in the sham acupuncture group dropped out (27 were disqualified for medical reasons, 25 withdrew) and 99 (52%) dropped out from the educational control group (29 were disqualified for medical reasons, 70 withdrew)


ParticipantsInclusion criteria

  • OA of the knee with radiographic evidence of at least one osteophyte at the tibiofemoral joint (Kellgren-Lawrence grade 2 or higher)
  • Aged 50 years or older
  • Moderate or greater clinically significant knee pain on most days during the past month
  • Willingness to be randomly assigned


Exclusion criteria

  • Presence of serious medical conditions that precluded participation in the study
  • Bleeding disorders that might contraindicate acupuncture
  • Intra-articular corticosteroid or hyaluronate injections during the past six months
  • Knee surgeries during the past six months
  • Concomitant use of topical capsaicin cream during the past six months
  • Previous experience with acupuncture
  • Any planned events (including total knee replacement) that would interfere with participation in the study during the following 26 weeks


Baseline characteristics

Baseline characteristics were similar among all treatment groups

Intervention group: education group (N = 189 randomly assigned, 174 analysed at baseline, 124 analysed after four weeks of FU, 125 analysed after eight weeks of FU, 113 analysed after 14 weeks of FU, 108 analysed after 26 weeks of FU)

Location of OA: 100% knee

Duration of OA: 44.3% less than five years, 24.3% six to 10 years, 31.4% longer than 10 years

PROGRESS-Plus

Place of residence: US, around Baltimore or Towson (both in Maryland) or around New York City (New York)

Race, ethnicity and culture: 66.7% white, 31.7% African American, 1.6% other

Sex: 32.8% male, 67.2% female

Education: 35.1% no college, 64.9% some college

Age (mean (SD)), years: 65.1 (8.8)

Disability: 74.7% one target knee, 25.3% two target knees; 73.4% moderate or lesser pain when walking on flat surface, 26.6% severe or extreme pain when walking on flat surface

Control group: true acupuncture (N = 190 randomly assigned, 186 analysed at baseline, 173 analysed after four weeks of FU, 169 analysed after eight weeks of FU, 158 analysed after 14 weeks of FU, 142 analysed after 26 weeks of FU)

Location of OA: 100% knee

Duration of OA: 53.8% less than five years, 19.9% six to 10 years, 25.8% longer than 10 years

PROGRESS-Plus

Place of residence: US, around Baltimore or Towson (both in Maryland) or around New York City (New York)

Race, ethnicity and culture: 70.0% white, 27.4% African American, 2.6% other

Sex: 36.8% male, 63.2% female

Education: 32.8% no college, 67.2% some college

Age (mean (SD)), years: 65.2 (8.4)

Disability: 75.0% one target knee, 25.0% two target knees; 76.5% moderate or lesser pain when walking on flat surface, 23.5% severe or extreme pain when walking on flat surface

Control group: sham acupuncture (N = 191 randomly assigned, 183 analysed at baseline, 163 analysed after four weeks of FU, 161 analysed after eight weeks of FU, 157 analysed after 14 weeks of FU, 141 analysed after 26 weeks of FU)

Location of OA: 100% knee

Duration of OA: 53.0% less than five years, 18.0% six to 10 years, 29.0% longer than 10 years

PROGRESS-Plus

Place of residence: US, around Baltimore or Towson (both in Maryland) or around New York City (New York)

Race, ethnicity and culture: 70.7% white, 26.7% African American, 2.6% other

Sex: 38.2% male, 61.8% female

Education: 25.4% no college, 74.6% some college

Age (mean (SD)), years: 66.2 (8.7)

Disability: 71.1% one target knee, 28.9% two target knees; 75.5% moderate or lesser pain when walking on flat surface, 24.5% severe or extreme pain when walking on flat surface


InterventionsIntervention: education

Description: The education-attention control intervention consisted of six two-hour group sessions based on the Arthritis Self-Management Program and taught by an experienced Arthritis Foundation–trained patient education specialist. In addition, educational materials were mailed to the education group periodically in an attempt to equalise the amount of experimental contact in all groups. Many topics were taught to the education control group, including types of arthritis, various treatments for arthritis, self-management of arthritis, creating an action plan to manage arthritis, body mechanics for home and office, products that make mobility and daily routine activities easier and safer, physical fitness and flexibility exercises for arthritis, pain management, depression and acceptance

Intended audience: people with arthritis of the knee

Mode: group sessions (mean number of participants: seven)

Personnel: trained patient education specialist

Delivery method: face-to-face

Language: English

Format: standard format

Location: -

Duration: six sessions, each of two hours, delivered every other week for a total duration of 12 weeks

Additional treatment during trial: unclear

Were the following heiQ components addressed?

Health-directed behaviour: yes

Positive and active engagement in life: unclear

Emotional well-being: yes

Self-monitoring and insight: no

Constructive attitudes and approaches: yes

Skill and technique acquisition: yes

Social integration and support: no

Health service navigation: yes

Comparator: true acupuncture

Type: alternate intervention

Description: The true acupuncture group underwent 26 weeks of gradually tapering treatment. The acupuncture point selection was based on Traditional Chinese Medicine meridian theory to treat knee joint pain

Frequency: eight weeks of two treatments per week, two weeks of one treatment per week, four weeks of one treatment every other week, 12 weeks of one treatment per month

Additional treatment during trial: unclear

Comparator: sham acupuncture

Type: attention control

Description: For the sham treatment, a combined insertion and non-insertion procedure was modified. The sham acupuncture procedure was given on the same schedule as in the experimental group and used the same active needle placements, except that actual insertion did not occur at the nine critical points

Frequency: eight weeks of two treatments per week, two weeks of one treatment per week, four weeks of one treatment every other week, 12 weeks of one treatment per month

Additional treatment during trial: unclear


OutcomesOutcome assessed at: baseline and four, eight, 14 and 26 weeks of follow-up

Primary outcomes of study

  • Pain (WOMAC pain subscale, 0 to 20, lower score is better)
  • Function self-reported (WOMAC function subscale, 0 to 68, lower score is better)


Secondary outcomes of study

  • Quality of life (SF-36 physical component score, 0 to 10, higher is better)
  • How does OA affect you? (patient global assessment score)
  • Functional performance (six-minute walk test, as many feet as possible, higher is better)
  • Adverse events
  • Which treatment participants were believed to receive


NotesWe extracted the following outcomes at week four (short term) and week 26 (intermediate term) for the analyses in this review: pain (WOMAC), function self-reported (WOMAC), quality of life (SF-36), functional performance (six-minute walk test) and dropouts (proportion of missing participants)

Funding was provided by the National Centre for Complementary and Alternative Medicine (National Institutes of Health Cooperative Agreement I01 AT-00171), with advice and encouragement by the National Institute of Arthritis and Musculoskeletal and Skin Diseases

Author (B Berman) sent additional information about the trial on request

Data analysis: Change scores were combined with end point scores using generic inverse variance


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low riskQuote: “(…) each cohort at each site was randomly assigned to 1 of 3 groups by a computer-generated process using randomly selected blocks of 3, 6 and 9”

Comment: This method of random sequence generation has low risk of selection bias

Allocation concealment (selection bias)Low riskQuote: “We assured allocation concealment by using disguised letter codes that were generated and sent to the site coordinators by a central statistical core”

Comment: Allocation was sufficiently concealed to minimise risk of bias

Blinding of participants and personnel (performance bias)
All outcomes
High riskQuote: “(…) the participants themselves (in the true acupuncture and sham acupuncture groups) (…) were blinded to group assignment”

Comment: Although the acupuncture groups were blinded, it was not possible to blind the education group, which may have introduced a risk of bias. It was not possible to blind personnel to treatment allocation either

Blinding of outcome assessment (detection bias)
All outcomes
High riskQuote: “The research assistants who collected assessments from participants, (…) and the statistician were blinded to group assignment”

Comment: Outcome assessors were blinded to treatment allocation; however, most outcomes are subjective, and participants in the control group are not blinded to group allocation

Incomplete outcome data (attrition bias)
All outcomes
High riskQuote: “At 26 weeks, 43% of the participants in the education group and 25% in each of the true and sham acupuncture groups were not available for analysis”

Quote: “Thus, we present only the results from the analyses that used all available data”

Comment: Dropout rate was very high and differed among treatment groups, introducing high risk of attrition bias. Also, dropouts from the education and true acupuncture group reported significantly more pain at baseline than was reported by completers. No intention-to-treat analysis was presented

Selective reporting (reporting bias)Low riskComment: All outcomes listed in the Methods section are reported in the Results

Other biasLow riskComment: No other potential sources of bias were identified

Blixen 2004

Methods


ParticipantsCriteria for defining the condition being treated (OA)

  • Documented diagnosis of OA in arthritis/rheumatology clinic


Inclusion criteria

  • Documented diagnosis of OA in an arthritis/rheumatology clinic
  • Aged 60 years or older
  • Visit to department of rheumatology in previous six months


Exclusion criteria: none

Baseline characteristics

Income and marital status differed significantly between groups (P < 0.05) at baseline

Intervention group: telephone health education strategy (N = 16 randomly assigned, 15 analysed)

Duration of OA (mean (IQR)): 10 (2 to 30) years

PROGRESS-Plus

Place of residence: US, urban

Race, ethnicity and culture: 69% Caucasian, 31% African American

Sex: 56% male, 44% female

Education: 6% grade 7 to 9, 25% high school, 44% 1 to 4 years college, 6% college graduate, 19% professional/graduate school

Socioeconomic status: 19% < $10,000 per year, 7% $10,000 to 19,999 per year, 7% $20,000 to 29,999 per year, 29% $30,000 to 39,999 per year, 14% $40,000 to 49,999 per year, 29% > $50,000 per year

Social capital: 75% married, 25% divorced

Age (mean (SD)), years: 71.7 (6.3)

Disability: 13% had joint replacement surgery, mean (SD) number of comorbid conditions 2.2 (1.0)

Control group: usual care (N = 16 randomly assigned, 15 analysed)

Duration of OA (mean (IQR)): 6.5 (2 to 18) years

PROGRESS-Plus

Place of residence: US, urban

Race, ethnicity and culture: 75% Caucasian, 25% African American

Sex: 69% male, 31% female

Education: 6% grade 7 to 9, 6% grade 10 to 11, 19% high school, 56% 1 to 4 years college, 13% college graduate, 0% professional/graduate school

Socioeconomic status: 25% < $10,000 per year, 19% $10,000 to 19,999 per year, 31% $20,000 to 29,999 per year, 6% $30,000 to 39,999 per year, 13% $40,000 to 49,999 per year, 6% > $50,000 per year

Social capital: 38% married, 13% single, 25% divorced, 25% widowed

Age (mean (SD)), years: 69.9 (5.9)

Disability: 19% had joint replacement surgery, mean (SD) number of comorbid conditions 2.3 (1.2)


InterventionsIntervention: telephone health education strategy

Description: The intervention was a telephone health education strategy, which involved six weekly mailings of OA self-management modules (adapted from The Arthritis Helpbook: A Tested Self-Management Program for Coping With Arthritis and Fibromyalgia). The modules covered (1) pathology, (2) OA medications, (3) interrelationship between emotional and physical components of pain and the importance of relaxation techniques, (4) depression, (5) importance of regular exercise and (6) weight management. Participants also received an audiotape that explained how to perform various relaxation techniques. The content of each module was reinforced by six weekly 45-minute telephone calls conducted by an advanced practice nurse. The nurse reviewed the materials, answered questions and helped participants set self-management goals and learn new skills

Intended audience: people with OA

Mode: individual

Personnel: advanced practice nurse

Delivery method: phone, audiotape, mailed modules

Language: English

Format: tailored to individual needs

Location: home

Duration: six weekly sessions of 45 minutes, total duration of six weeks

Additional treatment during trial: unclear

Were the following heiQ components addressed?

Health-directed behaviour: yes

Positive and active engagement in life: no

Emotional well-being: yes

Self-monitoring and insight: yes

Constructive attitudes and approaches: no

Skill and technique acquisition: yes

Social integration and support: no

Health service navigation: yes

Comparator: usual care

Type: usual care

Description: usual care for their OA with their respective rheumatologist

Additional treatment during trial: unclear


Outcomes


Notes


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Unclear riskQuote: “(…) and was randomly assigned to either to control or the intervention group”

Comment: The randomisation method is not described in the article

Allocation concealment (selection bias)Unclear riskComment: Concealment of allocation is not described in the article

Blinding of participants and personnel (performance bias)
All outcomes
High riskComment: The types of control and intervention groups make blinding of participants and personnel impossible, which may have introduced a risk of performance bias. Rheumatologists who cared for the control group, however, are unlikely to be influenced by randomisation of a participant to a control group

Blinding of outcome assessment (detection bias)
All outcomes
High riskQuote: “To minimize bias (…) did not participate in the patient self-management programme and who were unaware of treatment assignment”

Comment: Outcome assessors were blinded from participants’ allocated groups; however, most outcomes are subjective and participants are not blinded to group allocation

Incomplete outcome data (attrition bias)
All outcomes
Low riskQuote: “All 32 subjects completed the baseline and three-month telephone interviews. Two of the subjects (one from each group) had protracted periods in hospital at the six-month follow-up and we were unable to interview them”

Comment: Although no intention-to-treat-analysis was carried out, risk of bias is low because of the low percentage of dropouts, which was equal in the two groups and was explained by the reasons provided

Selective reporting (reporting bias)Low riskComment: All outcomes listed in the Methods section are described in the Results section

Other biasHigh riskComment: Because of the small sample size, a type II error is possible

Buszewicz 2006

MethodsStudy design:RCT, multi-centre, two arms, non-blinded

Country in which trial was carried out: United Kingdom (UK)

Method of recruitment of participants:GP identified potential participants from practice attendees over a six-week period, and computerised records of the practice were searched for relevant OA clinical terms and prescriptions for regular NSAIDs or analgesics

Setting: primary care

Was the sample size justified with a priori calculation of effect size/power? yes

Length of follow-up:12 months

Dropouts: 112 (27%) dropped out from the intervention group (five died, 35 withdrew, 72 did not respond), and 81 (20%) dropped out from the control group (two died, 23 withdrew, 56 did not respond)


ParticipantsInclusion criteria

  • Clinical diagnosis of knee and/or hip OA by GP
  • Aged 50 years or older
  • Diagnosis of OA for at least a year
  • Associated pain and/or functional disability during past month


Exclusion criteria

  • Recommendation for surgery for arthritis
  • Poor mobility
  • Poor understanding of English
  • Associated neurological signs
  • Cognitive impairment


Baseline characteristics

Baseline characteristics were similar in all treatment groups

Intervention group: self-management programme (N = 406 randomly assigned, 406 included in analysis)

PROGRESS-Plus

Place of residence: UK

Race, ethnicity and culture: 100% white, 0% Black African, < 1% Black Caribbean

Sex: 37% male, 63% female

Education: 72% no higher education, 28% higher education

Social capital: 83% house owner/occupier, < 1% staying with family or friends, 17% rented accommodation

Age (mean (SD)), years: 68.4 (8.2)

Control group: education (N = 406 randomly assigned, 406 included in analysis)

PROGRESS-Plus

Place of residence: UK

Race, ethnicity and culture:  99% white, < 1% Black African, 1% Black Caribbean

Sex: 37% male, 63% female

Education: 73% no higher education, 27% higher education

Social capital: 79% house owner/occupier, 1% staying with family or friends, 20% rented accommodation

Age (mean (SD)), years: 68.7 (8.6)


InterventionsIntervention: self-management programme

Description: The intervention consisted of group sessions provided by the Arthritis Care organisation. Components of the group sessions included information about arthritis, information about principles of self-help, exercises (breathing, physical exercise, muscle relaxation, distraction techniques, communication, guided imagery, dealing with fatigue), healthy eating, strategies for accessing resources, etc

Intended audience: people with arthritis

Mode: group sessions (12 to 18 participants)

Personnel: a trained volunteer from the voluntary organisation Arthritis Care

Delivery method: face-to-face

Language: English

Format: standard format

Location: -

Duration: six weekly sessions of 2½ hours each

Additional treatment during trial: unclear

Were the following heiQ components addressed?

Health-directed behaviour: yes

Positive and active engagement in life: yes

Emotional well-being: yes

Self-monitoring and insight: yes

Constructive attitudes and approaches: yes

Skill and technique acquisition: yes

Social integration and support: unclear

Health service navigation: yes

Comparator: education

Type: information only

Description: -

Frequency: one single education booklet was provided at the start of the trial

Additional treatment during trial: unclear


OutcomesOutcomes assessed at: baseline and four and 12 months of follow-up

Primary outcomes of study

  • Quality of life (SF-36 subscale mental and physical health, 0 to 100, higher score is better)


Secondary outcomes of study

  • Pain (WOMAC, 0 to 20, lower score is better)
  • Stiffness (WOMAC, 0 to 8, lower score is better)
  • Self-reported function (WOMAC, 0 to 68, lower score is better)
  • Emotional distress (Hospital Anxiety and Depression Scale (HADS), 0 to 21, lower score is better)
  • Self-management (ASES pain and other symptoms subscales, 5 to 35 and 6 to 42, higher score is better)


NotesWe extracted the following outcomes at 12 months (intermediate term) for the analyses in this review: self-management (ASES), pain (WOMAC), function self-reported (WOMAC), emotional distress (HADS subscale depression), quality of life (EQ-5D utility score) and dropouts (proportion of missing participants)

Funding was provided by the Medical Research Council (reference G9900306)

Author (M Buszewicz) sent end of treatment mean and SD scores for the intervention and control groups at all follow-up times on request

Data analysis: For the outcome self-management in OA, we could choose between ASES subscale pain and ASES subscale other and chose ASES subscale pain, as we judged that pain was a more measurable aspect of self-management. For the outcome quality of life, we could choose between SF-36 mental health subscale and EQ-5D utility score and chose EQ-5D utility score, as we judged this to be a better measure for quality of life. For the outcome global OA scores, we added the subscales of WOMAC (pain, stiffness and physical function) provided by the author to get the WOMAC total score; we estimated the SD with the formula provided in Table 7.7a in the Cochrane Handbook for Systematic Reviews of Interventions and the SDs from the subscales (this method was chosen in close consultation with a biostatistician)


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low riskQuote: “We used an independent centralised computerised system to randomise the participants. Practices were stratified by area and we used minimisation to balance for differences in age and sex”

Comment: adequate sequence generation resulting in low risk of selection bias

Allocation concealment (selection bias)Low riskQuote: “Research nurses faxed details of consenting participants to the trial manager, who passed the information on to the randomisation centre that allocated participants to their experimental group”

Comment: Because the randomisation centre allocated participants to their groups, allocation seemed to have remained concealed and the risk of selection bias low

Blinding of participants and personnel (performance bias)
All outcomes
High riskComment: Study participants were not blinded during the trial; therefore outcomes in this trial may have been biased

Comment: Although the volunteers who give the intervention (SMP) are not blinded to the allocation, this probably did not result in a higher risk of performance bias because they did not have contact with the control arm

Blinding of outcome assessment (detection bias)
All outcomes
High riskQuote: “Outcomes were assessed through questionnaires completed by the participants (…).”

Comment: Outcomes were self-reported by participants who were aware of their allocated treatment. Therefore detection bias was not sufficiently accounted for in this trial

Incomplete outcome data (attrition bias)
All outcomes
High riskQuote: “Questionnaire response rates were 95% at baseline, 80% at four months, and 76% at 12 months (…)”

Quote: “Analysis was based on intention to treat”

Comment: Both ITT and per-protocol analyses were carried out. Although correct imputing techniques were used, the response rates per time point were low, which means that a lot of data had to be imputed (24%), thus introducing a risk of attrition bias. Also, the numbers of withdrawals and non-responders are extensive in both treatment groups and are greater in the intervention arm (27% vs 20%)

Selective reporting (reporting bias)Low riskComment: All prespecified outcomes were reported in the article

Other biasHigh riskQuote: “In the intervention group more than half (56%; 219/392) attended four or more sessions, 9% (37/392) attended only one or two sessions, and 29% (115/392) attended none”

Comment: Overall adherence in this trial was low, which results in high risk for biased results

Calfas 1992

MethodsStudy design:RCT, single centre, two arms, outcome assessment blinded

Country in which trial was carried out:US

Method of recruitment of participants:Individuals were recruited by small newspaper ads, physician referral and public service announcements

Setting:outpatient

Was the sample size justified with a priori calculation of effect size/power?no justification of sample size provided

Length of follow-up: 12 months

Dropouts:Five (25%) dropped out from the intervention group (four refused or dropped out, one had transportation problems or died or fell ill); five (25%) dropped out from the control group (four refused or dropped out, one had transportation problems or died or fell ill)


ParticipantsInclusion criteria

  • Diagnosis of OA verified by each participant's physician according to the ARA criteria for OA
  • Individuals had to report at least some functional impairment (i.e. answered at least one of the functional questions as “modest” or above)


Exclusion criteria: none

Baseline characteristics

Baseline characteristics were similar among all treatment groups

Intervention group: cognitive-behaviour modification (N = 20 randomly assigned, 17 analysed post-test and after two months of FU, 15 analysed after six and 12 months of FU)

Duration of OA: 85% longer than five years, 5% one to five years, 10% less than one year

PROGRESS-Plus

Place of residence: San Diego, urban

Race, ethnicity and culture: 100% Caucasian

Occupation: 85% retired, 5% full-time employment, 10% part-time employment

Sex: 25% male, 75% female

Social capital: 70% married, 20% divorced, 10% widowed

Age (mean (SD)), years: 66.7 (8.1)

Control group: traditional education intervention (N = 20 randomly assigned, 18 analysed post-test, 16 analysed after two months of FU, 15 analysed after six and 12 months of FU)

Duration of OA: 100% longer than five years

PROGRESS-Plus

Place of residence: San Diego, urban

Race, ethnicity and culture: 95% Caucasian, 5% non-Caucasian

Occupation: 80% retired, 5% full-time employment, 15% part-time employment

Sex: 30% male, 70% female

Social capital: 70% married, 15% divorced, 10% widowed, 5% other

Age (mean (SD)), years: 67.3 (6.5)


InterventionsIntervention: cognitive-behaviour modification

Description: Participants were introduced to the rationale for a cognitive approach to pain management. Goals of the cognitive intervention were to (a) teach participants to reconceptualise their pain; (b) teach them to monitor their thoughts or beliefs, feelings and actions, and to recognise the relationships among them; (c) strengthen participants' belief that they can cope effectively; and (d) help participants make behavioural changes to improve their functioning. Some of the techniques used included cognitive reappraisal, imagery, relaxation training, mental distraction and goal setting. Each session was videotaped. Each session began with group members reporting on goals and homework from the previous session. Most of the sessions were spent on one topic. The facilitator used examples from the participants to illustrate points, emphasising group participation. At the end of each session, group members were given a homework assignment and set goals that were shared with the group

Intended audience: people with OA

Mode: group sessions

Personnel: trained facilitators

Delivery method: face-to-face

Language: English

Format: tailored to individual’s needs

Location: Arthritis Centre at San Diego State University

Duration: one session per week, for 10 weeks

Additional treatment during trial: unclear

Were the following heiQ components addressed?

Health-directed behaviour: yes

Positive and active engagement in life: yes

Emotional well-being: yes

Self-monitoring and insight: yes

Constructive attitudes and approaches: yes

Skill and technique acquisition: yes

Social integration and support: no

Health service navigation: no

Comparator: traditional education intervention

Type: attention control

Description: The education control group consisted of a series of didactic lectures given by healthcare professionals. Topics for the series included rheumatology, pharmacology, nutrition, joint replacement surgery, fibromyalgia, occupational therapy, physical therapy and recreational therapy, and a presentation given by the Arthritis Foundation. Specific behavioural instructions were not given during any of the lectures

Additional treatment during trial: unclear


OutcomesOutcome assessed at: baseline, post-test and two, six and 12 months of follow-up

Primary outcomes of study

  • Quality of life (Quality of Well-Being scale (QWB))
  • Physical health, social and psychological well-being on the AIMS
  • Emotional distress (Beck Depression Inventory (BDI))


Secondary outcomes of study

  • Social support (Social Support Questionnaire (SSQ))


NotesOnly outcomes on dropouts could be extracted because presentation of data was insufficient

Funding was provided by a grant from the Arthritis Foundation and by a grant from the National Institute of Arthritis and Musculoskeletal and Skin Diseases (AR 400770-01A1)


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Unclear riskQuote: “After screening and recruitment, each subject was randomly assigned to one of two groups (…)”

Comment: Authors do not describe the method used for randomisation

Allocation concealment (selection bias)Unclear riskComment: no information provided on allocation concealment

Blinding of participants and personnel (performance bias)
All outcomes
High riskComment: Participants and personnel were not blinded during the study; therefore results are possibly biased

Blinding of outcome assessment (detection bias)
All outcomes
High riskQuote: “Subjects were assessed by interviewers who were blind to treatment condition”

Comment: Outcome assessors were blinded at each outcome assessment; however, most outcomes are subjective, and participants are not blinded to group allocation

Incomplete outcome data (attrition bias)
All outcomes
High riskQuote: “At the 6- and 12-month follow-up, 75% of the subjects in each group completed testing”

Quote: “Seventy-five percent of the pretest scores of those who dropped out or had missing data were significantly different from the scores of those who stayed in the study”

Quote: “Tests are based on unequal n’s because some cases were lost to follow-up”

Comment: Dropout rates in both groups are equal but substantial. A difference in pretest scores between dropouts and participants who completed follow-up introduces a high risk of bias. No intention-to-treat analysis was performed

Selective reporting (reporting bias)Unclear riskOnly outcomes on dropouts could be extracted because data presentation was insufficient.

Comment: In Methods, the authors list social support using the Social Support Questionnaire (SSQ), but this is not reported in the Results section. No separate group data are presented in Table 3

Other biasLow riskComment: No other potential sources of bias were identified

Cronan 1997

MethodsStudy design:RCT, single-centre, four arms, non-blinded

Country in which trial was carried out:US

Method of recruitment of participants:Letters explaining the study and inviting people to participate were sent to 3,000 randomly selected persons from a large HMO membership list of 50,450 people who were 60 years of age or older

Setting:general population

Was the sample size justified with a priori calculation of effect size/power?No justification was provided for the sample size

Length of follow-up:three years

Dropouts

  • After one year: 24 (27.6%) dropped out from the social support group, 11 (11.3%) dropped out from the education group, 16 (18.0%) dropped out from the combination group, 13 (14.4%) dropped out from the control group


  • After two years: 32 (36.8%) dropped out from the social support group, 17 (17.5%) dropped out from the education group, 26 (29.2%) dropped out from the combination group, 19 (21.1%) dropped out from the control group


  • After three years: 35 (40%) dropped out from the social support group, 27 (28%) dropped out from the education group, 33 (37%) dropped out from the combination group, 24 (27%) dropped out from the control group


No reasons were provided for dropout


ParticipantsCriteria for defining the condition being treated (OA)

  • Self-reported symptoms of OA (in 90.3%, the diagnosis was confirmed in the medical records)


Inclusion criteria

  • Self-reported symptoms of OA (confirmed in medical records in 90.3%)
  • Aged 60 or older
  • Willingness to attend 10 weekly and 10 monthly sessions


Exclusion criteria: none

Baseline characteristics

Baseline characteristics were similar among all treatment groups

Intervention group: education group (N = 97 randomly assigned, 86 analysed after one year of FU, 80 analysed after two years of FU, 70 analysed after three years of FU)

Duration of OA (mean (SD)): 7.48 (5.93) years since diagnosis (N = 83)

PROGRESS-Plus

Place of residence: San Diego County

Race, ethnicity and culture: 94.8% Caucasian, 3.1% African American, 1.0% Hispanic, 1.0% Asian, 0% other

Sex: 29.9% male, 70.1% female

Education: 29.9% high school, 43.3% some college, 26.8% college degree

Socioeconomic status: income 26.2% under $20,000, 41.7% $20,000 to $40,000, 21.4% $40,000 to $60,000, 10.8% over $60,000

Age (mean (SD)), years: 69.68 (5.79)

Disability: 78.4% had other medical conditions, 48.5% had another serious comorbidity

Intervention group: combination group (N = 89 randomly assigned, 73 analysed after one year of FU, 63 analysed after two years of FU, 56 analysed after three years of FU)

Duration of OA (mean (SD)): 6.56 (5.09) years since diagnosis (N = 75)

PROGRESS-Plus

Place of residence: San Diego County

Race, ethnicity and culture: 92.1% Caucasian, 2.2% African American, 2.2% Hispanic, 2.2% Asian, 1.1% other

Sex: 34.8% male, 65.2% female

Education: 34.8% high school, 40.5% some college, 24.7% college degree

Socioeconomic status: income 31.8% under $20,000, 43.5% $20,000 to $40,000, 17.6% $40,000 to $60,000, 7.1% over $60,000

Age (mean (SD)), years: 68.77 (5.70)

Disability: 66.3% had other medical conditions, 47.2% had another serious comorbidity

Control group: social support group (N = 87 randomly assigned, 63 analysed after one year of FU, 55 analysed after two years of FU, 52 analysed after three years of FU)

Duration of OA (mean (SD)): 6.47 (5.55) years since diagnosis (N = 75)

PROGRESS-Plus

Place of residence: San Diego County

Race, ethnicity and culture: 92.0% Caucasian, 3.5% African American, 1.1% Hispanic, 3.4% Asian, 0% other

Sex: 43.7% male, 56.3% female

Education: 31.0% high school, 36.8% some college, 32.2% college degree

Socioeconomic status: income 20.0% under $20,000, 44.0% $20,000 to $40,000, 25.4% $40,000 to $60,000, 10.7% over $60,000

Age (mean (SD)), years: 69.14 (5.54)

Disability: 73.6% had other medical conditions, 50.6% had another serious comorbidity

Control group: no treatment (N = 90 randomly assigned, 77 analysed after one year of FU, 71 analysed after two years of FU, 66 analysed after three years of FU)

Duration of OA (mean (SD)): 7.34 (5.19) years since diagnosis (N = 79)

PROGRESS-Plus

Place of residence: San Diego County

Race, ethnicity and culture: 90.1% Caucasian, 1.1% African American, 4.4% Hispanic, 0% Asian, 4.4% other

Sex: 35.6% male, 64.4% female

Education: 28.9% high school, 30.0% some college, 31.1% college degree

Socioeconomic status: income 30.8% under $20,000, 43.6% $20,000 to $40,000, 14.1% $40,000 to $60,000, 11.5% over $60,000

Age (mean (SD)), years: 69.21 (5.51)

Disability: 61.1% had other medical conditions, 43.3% had another serious comorbidity


InterventionsIntervention: education group

Description: The education intervention was provided in a presentation format by professional health educators. Many presentations included active involvement of the participants. Presentations contained information about preventive health behaviours and self-management strategies, in addition to information about when to see a healthcare provider for ailments related to OA. The presentations emphasised appropriate healthcare usage, which is not always less healthcare usage. Participants were taught to recognise signs that indicate the need for quick medical attention to avoid future problems, in addition to learning to eliminate unnecessary healthcare utilisation by developing self-management skills. All materials were written using a self-help, consumer-centred approach. The materials encouraged people to play an important role in the decision-making process involving their health. In addition, materials on psychology, coping strategies, diet and exercise were presented. All materials were based on programmes developed by Lorig et al.

Intended audience: people with OA

Mode: group

Personnel: professional health educators

Delivery method: face-to-face

Language: English

Format: standard format

Location: central site (median 10 miles from participant’s home (range 2 to 75 miles))

Duration: 10 weekly sessions followed by 10 monthly sessions; each session took two hours

Additional treatment during trial: unclear

Were the following heiQ components addressed?

Health-directed behaviour: yes

Positive and active engagement in life: no

Emotional well-being: no

Self-monitoring and insight: yes

Constructive attitudes and approaches: yes

Skill and technique acquisition: yes

Social integration and support: no

Health service navigation: yes

Intervention: combination group

Description: The combined intervention included both educational classes and social support, with the first hour dedicated to education and the second to social support. During the second hour, no staff members were present (see social support group)

Intended audience: people with OA

Mode: group

Personnel: professional health educators (first hour only)

Delivery method: face-to-face

Language: English

Format: standard format and tailored to individual needs

Location: central site (median 10 miles from participant’s home (range 2 to 75 miles))

Duration: 10 weekly sessions followed by 10 monthly sessions; each session took two hours

Additional treatment during trial: unclear

Were the following heiQ components addressed?

Health-directed behaviour: yes

Positive and active engagement in life: no

Emotional well-being: no

Self-monitoring and insight: yes

Constructive attitudes and approaches: yes

Skill and technique acquisition: yes

Social integration and support: no

Health service navigation: yes

Comparator: social support group

Type: alternate intervention

Description: The social support intervention involved group discussions prompted by weekly task assignments aimed at promoting empathy and sharing coping techniques between group members. Group members were told that support groups can be effective in helping people deal with their arthritis. Tasks ranged from “elect a chair and secretary of the group and discuss what these people should do” to “divide the group into sets of 2 – 4 people, each of whom will agree to contact the others on a daily basis; work out a way a time that each member of a set can contact other members as near to every day as possible.” Other topics include problem solving, the role of emotion, accessing community resources and developing a group project. No staff members were present during these meetings

Frequency: 10 weekly sessions followed by 10 monthly sessions; each session took two hours

Additional treatment during trial: unclear

Comparator: control group

Type: no treatment

Description: Participants assigned to the control group were told that they would not need to attend regular meetings, that they would be contacted for periodic assessments and that they would receive quarterly newsletters. The quarterly newsletters focused on events in the area, offered helpful hints for issues not related to health and provided updates on the status of the study. Control group participants were also told that they would be invited to a presentation of the results at the end of the study and that they would be invited to participate in any continuing social support groups. In addition, they were told that their participation was vital in determining whether these programmes would be effective in helping people with osteoarthritis

Additional treatment during trial: unclear


OutcomesOutcome assessed at: baseline and one, two and three years of follow-up

Primary outcomes of study

  • Cohesiveness (sociometric questionnaire)
  • Helplessness (arthritis helplessness index (AHI))
  • Knowledge of OA (true/false questionnaire)
  • Self-management (ASES, 10 to 100, higher score is better)
  • Healthcare utilisation and costs (data from HMO database)
  • Quality of life (quality of well-being scale (QWB), 0 to 1.0, higher score is better)
  • Intervention evaluation (four open-ended questions)
  • Utilisation rates (participants’ medical records)
  • Comorbidity (10 yes/no questions from AIMS)


NotesWe extracted the following outcomes at 12 months (intermediate term) and at three years (long term) for the analyses in this review: quality of life (QWB) and dropouts (proportion of missing participants)

Funding was provided by several grants (NIH Grant AH-40423, NIH Grant 5P60 AR-40770, NIAMS Grant AR-33489)

Data analysis: Intervention groups were judged to be similar enough to combine by pooling data


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Unclear riskQuote: “After the interview, participants were randomly assigned to one of three health intervention groups or to a control group”

Comment: No information was provided on the method of randomisation

Allocation concealment (selection bias)Unclear riskComment: No information was provided on allocation concealment

Blinding of participants and personnel (performance bias)
All outcomes
High riskComment: Participants could not be blinded to their treatment allocation because of the nature of the intervention, thus introducing a risk of bias. Although personnel could not be blinded to treatment allocation either, the risk of bias seems to remain low because personnel were involved in only one intervention group (health education intervention)

Blinding of outcome assessment (detection bias)
All outcomes
High riskComment: No information was provided on the blinding of outcome assessors; however, most outcomes are subjective, and participants are not blinded to group allocation, thus introducing a risk of detection bias

Incomplete outcome data (attrition bias)
All outcomes
High riskQuote: “Attrition from the study after the intervention period was more likely to occur in the social support group than in the education group at the 1-year and 2-year assessment periods”

Quote: “Participants were allowed to skip items that they felt uncomfortable answering, resulting in variability in sample sizes between measures”

Comment: After three years, 35 participants (40%) dropped out from the social support group, 27 (28%) dropped out from the education group, 33 (37%) dropped out from the combination group and 24 (27%) dropped out from the control group. Dropout rates are high and differ among treatment groups. This, combined with the fact that participants were allowed to skip items they did not want to answer, results in a very high risk of attrition bias

Selective reporting (reporting bias)High riskComment: Only quality of life and dropouts could be extracted from this trial

Other biasHigh riskQuote: “Participants were present at an average of 6.59 of the 10 weekly meetings and 4.32 of the 10 monthly meetings, for a total of 10.9 of the 20 intervention sessions”

Comment: Attendance at meetings is on average very low

Crotty 2009

MethodsStudy design: RCT, single-centre, two arms, non-blinded

Country in which trial was carried out: Australia

Method of recruitment of participants:Participants who had an initial consultation with an orthopaedic surgeon concerning a potential hip or knee replacement and who were then added to the waiting list for joint replacement surgery were invited to take part in the study

Setting:outpatients added to waiting list for joint replacement surgery

Was the sample size justified with a priori calculation of effect size/power?yes

Length of follow-up: six months

Dropouts:no information provided


ParticipantsInclusion criteria

  • Mini-Mental Score ≥ 24
  • Able to read and speak English
  • Living in the southern region of Adelaide
  • Not classified as requiring urgent surgery
  • No significant frailty or illness that precluded completion of the protocol


Exclusion criteria:none

Baseline characteristics

Baseline characteristics were similar among all treatment groups

Intervention group: patient education self-management (N = 75 randomly assigned, 75 analysed)

Location of OA: 33.3% hip

Height (mean (SD)): 168.0 (10.3) cm

Weight (mean (SD)): 89.0 (25.4) kg

PROGRESS-Plus

Place of residence: urban (Adelaide)

Occupation: 12.4% working full-time or part-time, 4.0% home duties, 82.7% retired

Sex: 40% male, 60% female

Education: 5.5% none or some primary school, 13.7% primary school, 30.1% high school to year eight, 23.3% high school to year 12, 21.3% TAFE/trade, 5.5% university or above

Social capital: 42.7% lives alone

Age (mean (SD)), years: 68.1 (10.6)

Disability: two (one to three) medical conditions (mean (IQR)); 33.3% on waiting list for hip replacement

Control group: usual care (N = 77 randomly assigned, 77 analysed)

Location of OA: 32.5% hip

Height (mean (SD)): 166.4 (9.9) cm

Weight (mean (SD)): 82.6 (16.4) kg

PROGRESS-Plus

Place of residence: urban (Adelaide)

Occupation: 10.7% working full-time or part-time, 6.7% home duties, 82.7% retired

Sex: 39% male, 61% female

Education: 1.4% none or some primary school, 16.2% primary school, 31.1% high school to year eight, 31.1% high school to year 12, 13.5% TAFE/trade, 6.8% university or above

Social capital: 53.9% lives alone

Age (mean (SD)), years: 67 (11.0)

Disability: two (one to three) medical conditions (mean (IQR)); 32.5% on waiting list for hip replacement


InterventionsIntervention: patient education self-management

Description: Flinders University Chronic Disease Self-Management Model, which incorporates assessment of self-management knowledge, behaviours, attitudes, strengths and barriers, was used as a basis for the following interventions:

  • Face-to-face interview using the "Partners in Health"-scale, which identifies strengths of and barriers to self-management
  • Problems and goals interview that elicits participants' main life problem and medium-term goals; these are rated
  • Completion of a self-management action plan


Monthly telephone calls were made to check health status, self-management strategies reinforcement and progress with goal attainment monitoring. Encouraging to participate in self-management programme "Moving Towards Wellness" and in two joint replacement–specific education modules run by Arthritis SA

Intended audience: people with knee or hip OA on the waiting list for arthroplasty

Mode: individual

Personnel: peer support volunteers

Delivery method: face-to-face, by phone

Language: English

Format: tailored to individual needs

Location: -

Duration: telephone calls conducted monthly, 2.5 hours weekly for six weeks ("Moving Towards Wellness")/2.5 hours weekly for two weeks (joint replacement education programme)

Additional treatment during trial: unclear

Were the following heiQ components addressed?

Health-directed behaviour: no

Positive and active engagement in life: no

Emotional well-being: no

Self-monitoring and insight: yes

Constructive attitudes and approaches: yes

Skill and technique acquisition: no

Social integration and support: no

Health service navigation: no

Comparator: usual care

Type: usual care

Description: management by primary care physician; participant self-initiates appointments. Participant has access to a generic chronic disease self-management course, “Moving Towards Wellness” (2.5 hours weekly for six weeks), as advertised to the community

Additional treatment during trial: unclear


OutcomesOutcome assessed at: baseline and six months of follow-up

Primary outcomes of study

  • Health-directed activities (heiQ health-directed activities subscale, 1 to 6, higher score is better)


Secondary outcomes of study

  • Quality of life (AQoL, -0.04 to 1.00, higher score is better)
  • Beliefs about medication (beliefs about medication questionnaire (BMQ))
  • Pain (WOMAC, 0 to 20, lower score is better)
  • Stiffness (WOMAC, 0 to 8, lower score is better)
  • Function self-reported (WOMAC, 0 to 68, lower score is better)
  • Emotional distress (CES-D, 0 to 60, lower score is better)
  • Positive and active engagement in life (heiQ positive and active engagement in life subscale, 1 to 6, higher score is better)
  • Skill and technique acquisition (heiQ skill and technique acquisition subscale, 1 to 6, higher score is better)
  • Constructive attitudes and approaches (heiQ constructive attitudes and approaches subscale, 1 to 6, higher score is better)
  • Self-management (heiQ self-monitoring and insight subscale, 1 to 6, higher score is better)
  • Health service navigation (heiQ health service navigation subscale, 1 to 6, higher score is better)
  • Social integration and support (heiQ social integration and support subscale, 1 to 6, higher score is better)
  • Emotional distress (heiQ emotional distress subscale, 1 to 6, higher score is better)


NotesWe extracted the following outcomes at six months (intermediate term) for the analyses in this review: health-directed activities (heiQ health-directed activities subscale), quality of life (AQoL), pain (WOMAC), function self-reported (WOMAC), emotional distress (CES-D), positive and active engagement in life (heiQ positive and active engagement in life subscale), skill and technique acquisition (heiQ skill and technique acquisition subscale), constructive attitudes and approaches (heiQ constructive attitudes and approaches subscale), self-management (heiQ self-monitoring and insight subscale), health service navigation (heiQ health service navigation subscale), social integration and support (heiQ social integration and support subscale) and dropouts (proportion of missing participants)

The project was funded by the Commonwealth Department of Health and Ageing as part of its Better Arthritis Care budget initiative. Funding was administered through the Royal Australian College of Physicians, which was contracted to manage the Arthritis and Musculoskeletal Conditions Quality Improvement Program (AMQuIP)

Data analysis: For all outcomes, 95% confidence intervals were converted into SD


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low riskQuote: “A statistician external to the study generated the randomisation sequence using the random number generator in Microsoft Excel…”

Comment: Appropriate method of randomisation with low risk of bias

Allocation concealment (selection bias)Low riskQuote: “…sequentially numbered, opaque, sealed envelopes containing group allocation for participants”

Comment: Allocation was concealed during recruitment and randomisation

Blinding of participants and personnel (performance bias)
All outcomes
High riskComment: In this type of intervention and control, it is impossible to blind participants to their treatment group, thus introducing a risk of bias. The nurses conducting the intervention treatment, although also not blinded, have no influence on the treatment given to the control group (usual care by their own physician), leading to low risk of introducing bias

Blinding of outcome assessment (detection bias)
All outcomes
High riskQuote: "Six months after randomisation all participants were mailed a set of the same questionnaires that they completed at baseline and asked to complete and return them in the included reply paid envelope"

Comment: Outcome assessment was self-reported, and as participants were not blinded to treatment allocation, a possible risk of bias is present

Incomplete outcome data (attrition bias)
All outcomes
Low riskQuote: “Data were analysed by intention to treat according to the random allocation”

Comment: No information provided on loss to follow-up or dropouts. Appropriate statistical analyses were performed though, with the number of participants analysed at baseline equal to the number analysed at follow-up

Selective reporting (reporting bias)Low riskComment: All outcomes listed in Methods-section are reported in the Results

Other biasLow riskComment: No other potential sources of bias were identified

Hansson 2010

Methods


ParticipantsInclusion criteria

  • Clinical diagnosis of OA in the knee, hip or hand
  • Pain, stiffness and limitation of movement in the affected joint


Exclusion criteria

  • Inability to speak and understand Swedish


Baseline characteristics

Baseline characteristics were similar among all treatment groups 

Intervention group: PEPOA (N = 61 randomly assigned, 61 analysed)

Location of OA: 5% hip, 34% knee, 34% hand, 26% more locations

BMI: 36% BMI 20 to 25, 39% BMI 25 to 30, 26% BMI > 30

PROGRESS-Plus

Place of residence: Sweden

Age (mean (SD)), years: 62 (9.43)

Control group: usual care (N = 53 randomly assigned, 53 analysed)

Location of OA: 4% hip, 34% knee, 30% hand, 32% more locations

BMI: 34% BMI 20 to 25, 38% BMI 25 to 30, 28% BMI > 30

PROGRESS-Plus

Place of residence: Sweden

Age (mean (SD)), years: 63 (9.51)


InterventionsIntervention: PEPOA

Description: PEPOA consists of five group sessions. The first session included information about anatomy, physiology of pain, coping with pain and brainstorming about what participants find hard to do. Next sessions were about exercise and physical activity (plus practical demonstration of home-training exercises and orthopaedic aids), current research in OA, medication, appropriate diet, ergonomics and practical instructions about equipment and technical aids and surgery. In the fourth session, feedback on the brainstorming of the first session was provided

Intended audience: people with OA

Mode: group sessions (eight to 10 participants)

Personnel: healthcare professionals (different each session)

Delivery method: face-to-face

Language: Swedish

Format: standard format

Location: -

Duration: five group sessions, three hours for each session, once a week, for five weeks

Additional treatment during trial: unclear

Were the following heiQ components addressed?

Health-directed behaviour: yes

Positive and active engagement in life: no

Emotional well-being: no

Self-monitoring and insight: yes

Constructive attitudes and approaches: no

Skill and technique acquisition: yes

Social integration and support: no

Health service navigation: no

Comparator: usual care

Type: usual care

Description: Control group continued living as usual

Additional treatment during trial: unclear


Outcomes


Notes


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low riskQuote: “Then an independent person randomised the patient to either the intervention group or the control group, using a random number list and sealed envelopes”

Comment: A random number list is an appropriate randomisation method

Allocation concealment (selection bias)Unclear riskQuote: “Then an independent person randomised the patient to either the intervention group or the control group, using a random number list and sealed envelopes”

Comment: It is unclear whether the sealed envelopes were opaque and thus appropriately concealed

Blinding of participants and personnel (performance bias)
All outcomes
High riskQuote: “As in all randomised controlled trials concerning rehabilitation, it is not possible to use a double-blind design, since the patients always know whether they are in the intervention group or the control group”

Comment: Participants were not blinded during the trial for their allocation, possibly introducing a risk of bias. Although personnel also were not blinded to treatment allocation, it is unlikely that this has influenced the results 

Blinding of outcome assessment (detection bias)
All outcomes
High riskQuote: “The present study has a single-blind design, with measurements at baseline and after six months performed by either a physiotherapist or an occupational therapist, who did not know whether the patient had been in the intervention group or the control group”

Comment: Outcome assessors were blinded at the six-month follow-up measurement; however, most outcomes are subjective, and participants are not blinded to group allocation

Incomplete outcome data (attrition bias)
All outcomes
Low riskQuote: “Data were analysed on an intention-to-treat basis, with the dropouts included and the last observation carried forward”

Comment: Although the numbers of dropouts between intervention and control groups are not equal (respectively, 16% and 8%), six of the 10 dropouts in the intervention group dropped out before receiving the intervention. Furthermore, reasons for dropping out were similar between groups. Appropriate methods were used to correct for incomplete outcome data

Selective reporting (reporting bias)Low riskComment: All outcomes listed in Methods section were reported in Results

Other biasLow riskComment: No other potential sources of bias were identified

Heuts 2005

Methods


ParticipantsInclusion criteria

  • OA conforms to ICHPP-2 criteria
  • Aged 40 to 60 years


Exclusion criteria

  • A diagnosis of rheumatoid arthritis, ankylosing spondylitis or gout


Baseline characteristics

Baseline characteristics were similar in all treatment groups

Intervention group: self-management programme (N = 149 randomly assigned, 132 analysed)

BMI (mean (SD)): 28.0 (4.8) kg/m2

PROGRESS-Plus

Place of residence: The Netherlands

Occupation: 42% paying job, 33% no paying job (data from N = 99)

Sex: 41% male, 59% female

Education: 24% low, 33% middle, 18% high (data from N = 99)

Age (mean (SD)), years: 51.0 (5.0)

Control group: care-as-usual (N = 148 randomly assigned, 141 analysed)

BMI: 28.3 (5.2) kg/m2

PROGRESS-Plus

Place of residence: The Netherlands

Occupation: 39% paying job, 38% no paying job (data from N = 108)

Sex: 40% male, 60% female

Education: 27% low, 28% middle, 22% high (data from N = 108)

Age (mean (SD)), years: 52.2 (5.1)

Note: After randomisation, before the start of the intervention, 24 participants withdrew for practical reasons (17 from the intervention group, seven from the control group); some were not able to participate in the intervention schedule, and some were disappointed about the result of the randomisation


InterventionsIntervention: self-management programme

Description: People were taught how to take initiative in their personal health and functioning. Participants learned to use adequate goal setting in combination with self-incentives as motivators to optimise their level of activity. Rational use of prescribed medication and other treatments was discussed. Self-relaxation training was given for pain control, as well as for improvement of overall well-being. Problem solving was part of the programme for empowering the participant in handling daily hassles. Self-diagnostic skills were taught for monitoring and interpreting changes in one’s health status. Participants also received information about community resources and were trained to optimise use of healthcare services

Intended audience: persons with chronic complaints ascribed to OA

Mode: group sessions

Personnel: trained physiotherapists

Delivery method: face-to-face, written and audiovisual material

Language: Dutch

Format: standard format

Location: physiotherapists’ clinic

Duration: six sessions of two hours each

Additional treatment during trial: unclear

Were the following heiQ components addressed?

Health-directed behaviour: yes

Positive and active engagement in life: no

Emotional well-being: yes

Self-monitoring and insight: yes

Constructive attitudes and approaches: no

Skill and technique acquisition: yes

Social integration and support: unclear

Health service navigation: yes

Comparator: care-as-usual

Type: usual care

Description: What was prescribed by a family physician or a consulted specialist remained unchanged

Additional treatment during trial: unclear


Outcomes


Notes


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low riskQuote: “A computer generated randomization scheme was prepared (…)”

Comment: Appropriate method for generating a random sequence was used

Allocation concealment (selection bias)Low riskQuote: “(…) and managed by a secretary, who was not involved in patient selection, treatment and data analysis”

Comment: Allocation was concealed from study personnel during selection

Blinding of participants and personnel (performance bias)
All outcomes
High riskComment: Participants could not be blinded from their treatment, thus participant-reported outcomes are at high risk of performance bias

Comment: No blinding of personnel (physiotherapists) does not seem to introduce a risk of performance bias because the comparator is usual care

Blinding of outcome assessment (detection bias)
All outcomes
High riskQuote: “(…) assessments were performed by an independent research assistant who was blinded to treatment assignment and was not involved in the treatment (…)”

Comment: Outcome was assessed by blinded personnel who were not involved in treatment; however, most outcomes are subjective, and participants are not blinded to group allocation

Incomplete outcome data (attrition bias)
All outcomes
Low riskQuote: “(…) all participants were analysed in the intervention group they entered by randomization (…) no imputation method was administered for missing questionnaires”

Quote: “Five participants withdrew during the intervention from the self-management program: 3 were not satisfied by the program, one because of knee pain, and one because of the situation at home”

Comment: Participants were analysed in the group to which they were randomly assigned. Although missing data are not treated according to the intention-to-treat principle, the risk of attrition bias remains low, as the dropout rate is low with reasons provided

Selective reporting (reporting bias)Low riskComment: All outcomes mentioned in the Methods section were reported in the Results. No prepublished protocol is available

Other biasLow riskComment: No other potential sources of bias were identified

Hopman-Rock 2000

MethodsStudy design:RCT, single-centre, two arms, outcome assessment blinded

Country in which trial was carried out:The Netherlands

Method of recruitment of participants:Participants were recruited by announcements in newspapers and on television in the area around the research centre

Setting: general population

Was the sample size justified with a priori calculation of effect size/power?yes

Length of follow-up:six months

Dropouts:One (2%) dropped out from the control group (personal reasons), no one dropped out from the intervention group

Note: “During the statistical analyses after the follow-up assessment, subjects without a confirmed diagnosis of OA were excluded (4 in the experimental group, 10 in the control group). (...) The results for the total group (...) led to the same conclusions”


ParticipantsInclusion criteria:

  • Self-reported OA of hip or knee, confirmed by clinical (only for knee) or radiographic ACR criteria (hip and knee, Kellgren-Lawrence score of 2 or higher)
  • Aged 55 to 75 years


Exclusion criteria

  • People on a waiting list for knee or hip replacement


Baseline characteristics

Baseline characteristics were similar among all treatment groups

Intervention group: health educational and exercise programme (N = 60 randomly assigned, 56 analysed)

BMI (mean (SD)): 28.4 (4.8) kg/m2

Duration of OA: 4% less than one year, 25% one to three years, 30% three to 10 years, 18% 10 to 20 years, 18% longer than 20 years

PROGRESS-Plus

Place of residence: around Leiden, The Netherlands

Sex: 22% male, 78% female

Education: 17% primary, 54% secondary, 27% college/university

Social capital: 68% living together/married, 29% living alone

Age (mean (SD)), years: 65.4 (5.3)

Disability: 2.2 (1.5) other chronic conditions (SD)

Control group (N = 60 randomly assigned, 49 analysed)

BMI (mean (SD)): 26.8 (3.5) kg/m2

Duration of OA: 2% less than one year, 16% one to three years, 37% three to 10 years, 18% 10 to 20 years, 16% longer than 20 years

PROGRESS-Plus

Place of residence: around Leiden, The Netherlands

Sex: 12% male, 88% female

Education: 26% primary, 45% secondary, 20% college/university

Social capital: 67% living together/married, 25% living alone

Age (mean (SD)), years: 65.2 (5.7)

Disability: 2.8 (1.7) of other chronic conditions (SD)


InterventionsIntervention: health educational and exercise programme

Description: The first hour of each session was guided by a peer educator, and the following topics were discussed: pathophysiology of OA, lifestyle and physical activity, pain management, the importance of weight reduction and diet, ergonomic aspects and medical aspects of OA (treatments, radiographs). Additionally, questions were answered by an invited occupational therapist and a GP. The course included the use of a pain diary and personal goal planning, as well as interactive methods in the group. The second hour, participants were taught an exercise programme by a physical therapist. Fifteen minutes of each session was spent on education about the balance between rest and activity, preferable types of activity and how to incorporate them in a daily lifestyle, along with practical advice on physical activity, such as the benefits of walking. Participants were encouraged to do the exercises at home at least three times a week

Intended audience: people with hip or knee OA

Mode: group (max 15 participants)

Personnel: first hour, a peer educator and an invited occupational therapist and a GP; second hour, a physical therapist

Delivery method: face-to-face

Language: Dutch

Format: tailored to individual needs and standard format

Location: -

Duration: six weekly sessions of two hours

Additional treatment during trial: unclear

Were the following heiQ components addressed?

Health-directed behaviour: yes

Positive and active engagement in life: no

Emotional well-being: no

Self-monitoring and insight: yes

Constructive attitudes and approaches: yes

Skill and technique acquisition: yes

Social integration and support: no

Health service navigation: no

Comparator

Type: no treatment

Description: control group without intervention. The programme book (together with a gift voucher for $25) was offered to the control group after the follow-up test was finished

Additional treatment during trial: unclear


OutcomesOutcome assessed at: baseline, post-treatment, and six months of follow-up

Primary outcomes of study

  • Pain (pain subscale of IRGL (Dutch AIMS), 6 to 25, lower score is better)
  • Pain (pain intolerance on 10-cm VAS, 0 to 10, lower score is better)
  • Quality of life (VAS, 0 to 100, higher score is better)
  • Quality of life (sum score seven questions, 7 to 39, higher score is better)
  • Functional performance (range of motion with goniometer)
  • Functional performance (isometric muscle strength with dynamometer)
  • Functional performance (20-m walking test, timed up-and-go test, stair walking, reaching for toes in sitting position, lower scores are better)
  • Knowledge about OA (20 statements)
  • Self-management (self-efficacy on five 10-cm VAS, 0 to 100, higher score is better)
  • Function self-reported (mobility subscale of IRGL (Dutch AIMS), 7 to 28)
  • Healthcare utilisation (use of medication, number of visits to GP, number of visits to physical therapist)


NotesWe extracted the following outcomes at post-treatment (short term) and at six months (intermediate term) for the analyses in this review:

self-management (five 10-cm VAS), pain (IRGL subscale pain), function self-reported (IRGL subscale mobility), functional performance (20-m walking test), quality of life (sum score seven questions) and dropouts (proportion of missing participants)

Funding provided by a grant from The Netherlands Health Research and Development Council

Data analysis: For the outcome function—performance—we could choose between the 20-m walking test and the timed up-and-go test and chose for the 20-m walking test, as we judged that more studies reported performance measures similar to the 20-m walking test


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Unclear riskQuote: “Finally, 120 subjects were randomly assigned to one of the 2 conditions (…)”

Comment: no information provided on method of random sequence generation

Allocation concealment (selection bias)Unclear riskComment: no information provided on allocation concealment

Blinding of participants and personnel (performance bias)
All outcomes
High riskComment: With this kind of intervention and control, it is not possible to blind participants from their allocated treatment, possibly introducing a risk of bias. Although personnel were not blinded during the trial, it is unlikely that this has biased the results

Blinding of outcome assessment (detection bias)
All outcomes
High riskQuote: “The personal interview (by a trained interviewer who was blinded) contained (…). The examination was carried out by 3 trained physical therapists who were blinded for the condition”

Comment: Personnel who assessed several outcomes were blinded to treatment allocation; however, most outcomes are subjective, and participants are not blinded to group allocation

Incomplete outcome data (attrition bias)
All outcomes
Low riskComment: Although it is unclear whether an intention-to-treat analysis was performed, risk of bias is low because dropout is minimal (N = 1)

Selective reporting (reporting bias)Low riskComment: All outcomes reported in the Methods section are reported in the article

Other biasLow riskComment: No other potential sources of bias were identified

Hughes 2004

MethodsStudy design: RCT (block-randomisation (30), stratified for ACR functional class), single-centre, two arms, non-blinded

Country in which trial was carried out:US

Method of recruitment of participants: by newsletter, announcements in the local media and presentations to local senior groups

Setting:general population

Was the sample size justified with a priori calculation of effect size/power?no justification of sample size given

Length of follow-up: 12 months

Dropouts:After 12 months, 57 (50%) dropped out from the intervention group (38 unable to contact, eight refused interview, five illness, three moved out of state, two caregiving responsibilities, one disconnected phone), 68 (68%) dropped out from the control group (43 unable to contact, 11 refused interview, six illness, three moved out of state, three caregiving responsibilities, two disconnected phone)


ParticipantsInclusion criteria

  • Clinical criteria of the ACR for knee OA or hip OA


Exclusion criteria

  • Younger than age 60
  • Currently participating in an aerobic exercise programme
  • Uncomplicated hip or knee surgery in the past six months
  • Complicated surgery within the past year
  • Steroid injections in either knee or hip within the previous three months
  • Diagnosed with RA
  • Acutely inflamed or significantly swollen joint
  • Severe limiting cardiovascular disease
  • Active thrombophlebitis
  • Recent pulmonary embolus
  • An acute systemic illness
  • Poorly controlled diabetes
  • Other health conditions that might preclude exercise training


Baseline characteristics

Baseline characteristics were similar among all treatment groups

Intervention group: fit and strong! (N = 115 randomly assigned, 58 analysed)

PROGRESS-Plus

Place of residence: US

Race, ethnicity and culture: 69.4% white-Caucasian, 27.8% African American, 1.9% Hispanic, 0.9% Asian-Pacific Islander, 0% other

Sex: 19.4% male, 80.6% female

Education: 12.1% less than high school, 21.5% high school, 66.4% more than high school

Socioeconomic status: 32.4% income less than $20,000

Age, years: 73.3

Disability: 22.6% ARA class I, 64.5% ARA class II, 12.9% ARA class III; 51.4% hypertension, 46.7% cardiovascular disease, 6.1% asthma, 3.5% emphysema, 14.6% diabetes, 6.1% cancer

Control group: information only (placed on waiting list for Fit and Strong! programme) (N = 100 randomly assigned, 32 analysed)

PROGRESS-Plus

Place of residence: US

Race, ethnicity and culture: 75.0% white-Caucasian, 16.3% African American, 3.3% Hispanic, 3.3% Asian-Pacific Islander, 2.2% other

Sex: 14.1% male, 85.9% female

Education: 8.8% less than high school, 18.5% high school, 72.7% more than high school

Socioeconomic status: 33.7% income less than $20,000

Age, years: 73.4

Disability: 22.2% ARA class I, 64.2% ARA class II, 13.6% ARA class III; 58.3% hypertension, 42.3% cardiovascular disease, 7.0% asthma, 5.0% emphysema, 12.8% diabetes, 2.0% cancer


InterventionsIntervention: Fit and Strong!

Description: The first 60 minutes of the intervention included both resistance training and fitness walking. In resistance training, resistance was progressively increased throughout the programme by adding weight. Fitness walking progressed from maximum duration at baseline to 30 minutes over time. The last 30 minutes included an adapted version of the group discussion-educational component by Kovar and colleagues (1992) to enhance adherence efficacy. Self-efficacy for exercise (confidence in the ability to conduct the exercises in a safe and effective manner) and self-efficacy for exercise adherence (confidence in the ability to adhere to exercise participation over time and in the presence of barriers) were addressed, as was self-efficacy to manage pain and other arthritis-related symptoms. Therapists provided systematic feedback to participants on progress made. The emphasis was on building skills and identifying strategies that will assist the participant in maintaining adherence. Staff used group and individual sessions to inform participants about opportunities for maintaining exercise within the community or in the individual’s home. All graduates are also given a copy of The Arthritis Helpbook, a graduation certificate and tapes of music used during the class at a graduation ceremony at 8 weeks.

Intended audience: people with OA of hip or knee

Mode: group sessions (15 participants)

Personnel: physical therapist

Delivery method: face-to-face

Language: English

Format: tailored to individual needs

Location: senior centres and senior housing residences

Duration: 90-minute sessions, held three times a week, for eight weeks

Additional treatment during trial: unclear

Were the following heiQ components addressed?

Health-directed behaviour: yes

Positive and active engagement in life: no

Emotional well-being: no

Self-monitoring and insight: no

Constructive attitudes and approaches: no

Skill and technique acquisition: yes

Social integration and support: no

Health service navigation: no

Comparator

Type: Information only

Description: Control group participants are given a copy of The Arthritis Helpbook and a list of exercise programmes in the community that they can access. They are also given a variety of self-care materials and handouts at each post-test.

Additional treatment during trial: unclear


OutcomesOutcome assessed at: baseline and two, six and 12 months of follow-up

Primary outcomes of study

  • Self-management (self-efficacy (ASES) subscales exercise (1 to 10), pain management (10 to 100), other symptoms (10 to 100))
  • Self-management (barriers adherence efficacy scale, time adherence self-efficacy)
  • Adherence
  • Functional performance (lower extremity muscle strength, timed sit-to-stand in 60 seconds, six-minute distance walk)
  • Pain (WOMAC-pain, 0 to 20, lower score is better)
  • Function self-reported (WOMAC-stiffness (0 to 8), physical function (0 to 68), lower scores are better)
  • Maintenance of physical activity
  • Pain (geri-AIMS pain, 0 to 10)


NotesWe extracted the following outcomes at 12 months (intermediate term) for the analyses in this review: self-management (ASES subscale pain), pain (WOMAC subscale pain), global OA scores (WOMAC total), function self-reported (WOMAC subscale physical function), functional performance (six-minute distance walk) and dropouts (proportion of missing participants)

The development of Fit and Strong! was made possible by a grant from the Chicago Chapter of the Arthritis Foundation. The research was also supported by funding from the National Institute on Arthritis and Musculoskeletal Disease (Grant AR60692) and by the National Institute on Aging and the Royal Center for Research on Applied Gerontology (Grant AG 15890)

Data analysis: For all outcomes, 95% confidence intervals were converted into SD. For the outcome self-management in OA, we could choose between ASES subscale pain and ASES subscale other and chose for ASES subscale pain, as we judged that pain was a more measurable aspect of self-management. For the outcome global OA scores, we added the subscales of WOMAC (pain, stiffness and physical function) to get the WOMAC total score; we estimated the SD with the formula provided in Table 7.7a in the Cochrane Handbook for Systematic Reviews of Interventions and the SDs from the subscales (this method was chosen in close consultation with a biostatistician)


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Unclear riskQuote: “At that point, we randomised the participant to either the treatment or the control group”

Comment: no information provided on the method of random sequence generation

Allocation concealment (selection bias)Unclear riskComment: no information provided on allocation concealment

Blinding of participants and personnel (performance bias)
All outcomes
High riskQuote: “Third, participants were not blinded regarding their treatment status. It is not possible in an exercise study to blind the instructor to the participants, nor is it possible to blind the exercise participants to the fact that they are receiving a treatment”

Comment: Participants were not blinded, which may have introduced a risk of bias. Although personnel were not blinded either, it is unlikely that this introduced a risk of bias because no intervention staff came in contact with the control group

Blinding of outcome assessment (detection bias)
All outcomes
High riskQuote: “In this comparatively small study, it also was not possible to blind the research staff regarding group assignment because many of the staff also helped to set up the class, assisted the physician to conduct the physical examinations, and other activities”

Comment: Outcome assessors were not blinded, which resulted in a risk of detection bias

Incomplete outcome data (attrition bias)
All outcomes
High riskQuote: “These analyses demonstrated no significant differences between responders and non-responders on any of the outcome measures, (…) in terms of demographic characteristics or level of arthritis severity. In both groups, participants who attrited from post-test measurement had slightly worse scores on the study outcome measures at baseline”

Comment: The dropout rate in this study is very high and is not equally divided between treatment arms (50% in the intervention group vs 68% in the control group). Although not statistically significant, an apparent trend suggested a slightly worse score on study outcome measures at baseline among participants who dropped out. No intention-to-treat analysis was performed, but a completers-only analysis was done

Selective reporting (reporting bias)Low riskComment: All outcomes listed in the Methods were reported in the Results

Other biasLow riskComment: No other potential sources of bias were identified

Hurley 2007

MethodsStudy design: cluster-RCT, multi-centre, three arms, outcome assessment blinded

Country in which trial was carried out:UK

Method of recruitment of participants:Potential participants were identified from databases of primary care practices

Setting:primary care

Was the sample size justified with a priori calculation of effect size/power?yes

Length of follow-up:six months

Dropouts:27 (19%) in the usual primary care group dropped out, 25 (17%) in the individual rehabilitation group dropped out, 24 (18%) in the group rehabilitation group dropped out (five exercise-related events, 36 stopped attending without reason, 15 lost interest, 10 were unable to fulfil time commitment, five had unrelated medical problems, five moved away)


ParticipantsInclusion criteria

  • Knee pain due to OA based on clinical presentation and history (no attempts were made to identify the cause of the pain using investigations not routinely available to primary care physicians (e.g. radiographs))
  • Age 50 years or older
  • Consulted primary care physician for mild, moderate or severe knee pain of > six months’ duration


Exclusion criteria

  • Lower limb arthroplasty
  • Physiotherapy for knee pain in the preceding 12 months
  • Intra-articular injections in the preceding six months
  • Unstable medical conditions
  • Inability/unwillingness to exercise
  • Wheelchair dependence
  • Inability to understand English


Baseline characteristics

Baseline characteristics were similar among all treatment groups

Intervention group: individual rehabilitation (N = 186 randomly assigned, 121 analysed)

Location of OA: 100% knee

BMI (mean (range)): 30.0 (18 to 45) kg/m2

Duration of OA (mean (IQR)): 7 (3 to 15) years

PROGRESS-Plus

Place of residence: South-East London

Sex: 28.8% male, 71.2% female

Age (mean (range)), years: 66 (50 to 91)

Intervention group: group rehabilitation (N = 132 randomly assigned, 108 analysed)

Location of OA: 100% knee

BMI (mean (range)): 30.18 (20 to 50) kg/m2

Duration of OA (mean (IQR)): 5 (2.5 to 11) years

PROGRESS-Plus

Place of residence: South-East London

Sex: 28.8% male, 71.2% female

Age (mean (range)), years: 68 (51 to 84)

Control group (N = 140 randomly assigned, 113 analysed)

Location of OA: 100% knee

BMI (mean (range)): 30.3 (20 to 51) kg/m2

Duration of OA (mean (IQR)): 6 (3 to 15) years

PROGRESS-Plus

Place of residence: South-East London

Sex: 31.4% male, 68.6% female

Age (mean (range)), years: 67 (51 to 89)


InterventionsIntervention (individual/group rehabilitation)

Description: The intervention comprised integrated patient education, with simple self-management and pain coping strategies, delivered in the first 15 to 20 minutes of each rehabilitation session. Sessions were designed to be interactive, including active problem solving when appropriate. This was followed by a 35 to 45-minute individualised progressive exercise programme. The order in which the exercises were performed varied because of the circuitous regimen, and exercise specificity varied between participants and within participants over time, depending on their ability, rate of progression and identified areas of disability. Exercise complexity and intensity were increased through mutual agreement between physiotherapist and participant. The content was similar in individual or group rehabilitation

Intended audience: people with chronic knee pain

Mode: individual/group sessions (˜eight participants)

Personnel: experienced physiotherapist

Delivery method: face-to-face

Language: English

Format: tailored to individual needs

Location: physiotherapy outpatient department

Duration: 12 sessions (twice weekly for six weeks), one hour per session (15 to 20 minutes + 35 to 45 minutes)

Additional treatment during trial: Management of all participants’ knee and coexistent medical problems continued at the primary care physician’s discretion

Were the following heiQ components addressed?

Health-directed behaviour: yes

Positive and active engagement in life: yes

Emotional well-being: yes

Self-monitoring and insight: yes

Constructive attitudes and approaches: yes

Skill and technique acquisition: yes

Social integration and support: no

Health service navigation: no

Comparator

Type: usual care

Description: Primary care physicians were free to prescribe or refer participants (to any services or interventions they considered appropriate), whatever intervention they considered appropriate, and followed up as necessary. Management was typical of other primary care reports

Additional treatment during trial: unclear


OutcomesOutcome assessed at: baseline, post-treatment (or six weeks after recruitment to usual care arm) and six months after completion intervention (or 7.5 months after recruitment to usual care arm)

Primary outcomes of study

  • Function self-reported (WOMAC subscale function, 0 to 68, lower score is better)
  • Costs of intervention


Secondary outcomes of study

  • Pain (WOMAC subscale pain, 0 to 20, lower score is better)
  • Global OA scores (WOMAC-total score, 0 to 96, lower score is better)
  • Functional performance (aggregated functional performance time (AFPT) in seconds, lower score is better)
  • Exercise health beliefs (ExBeliefs questionnaire)
  • Emotional distress (HADS-anxiety (0 to 21), -depression subscales (0 to 21), lower score is better)
  • Quality of life (EQ-5D, higher score is better)
  • Quality of life (MACTAR, higher score is better)
  • Functional performance (quadriceps strength and voluntary activation, higher score is better)


NotesWe extracted the following outcomes at six months (intermediate term) for the analyses in this review: pain (WOMAC subscale pain), global OA scores (WOMAC-total), function self-reported (WOMAC subscale function), quality of life (EQ-5D), emotional distress (HADS subscale depression) and dropouts (proportion of missing participants)

Research funded by Arthritis Research Campaign Research Fellowship

Study author (M Hurley) provided additional information on request

Data analysis: Intervention groups were judged to be similar enough to combine by pooling data


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low riskQuote: “The randomization list was generated at a central location away from the research centre by an author (BR), who was not involved in the execution of the trial”

Comment: This method of random sequence generation has a low risk of bias

Allocation concealment (selection bias)Unclear riskComment: no information on allocation concealment given

Blinding of participants and personnel (performance bias)
All outcomes
High riskComment: Blinding of participants and personnel to treatment allocation is not possible because of the types of intervention and control in this trial

Blinding of outcome assessment (detection bias)
All outcomes
High riskQuote: “Outcome assessors were blinded to a participant’s allocation. Success of blinding was evaluated (…)”

Comment: In ‘discussion’, authors explain that subgroup analysis suggested that unblinding was associated with slightly better outcome, but that this difference was not statistically significant. Most outcomes are self-reported, and participants were not blinded to treatment allocation, so high risk of detection bias

Incomplete outcome data (attrition bias)
All outcomes
Low riskQuote: “There was no evidence of differential attrition”

Quote: “Statistical analysis followed a prespecified protocol, based on intent-to-treat with no interim or post hoc analyses”

Quote: “By 6 months, 76 (18%) participants had withdrawn”

Comment: Loss to follow-up was low and equal among all treatment groups (19%, 17% and 18%). Reasons for loss to follow-up are stated; only 1% withdrew because of exercise-related adverse events

Selective reporting (reporting bias)Low riskAll planned outcomes were reported

Other biasHigh riskQuote: “Of the participants who attended 6-month follow-up, 105 (85%) of 120 Indiv-rehab participants and 59 (55%) of 107 Grp-rehab participants attended >= 10 of the 12 sessions”

Comment: Adherence in the group-rehabilitation arm was lower than in the individual-rehabilitation arm

Quote: "Multilevel modeling was used to adjust for the intracluster correlations (...)"

Comment: The cluster design was accounted for in the statistical analysis

Jessep 2009

MethodsStudy design: RCT, single-centre, two arms, outcome assessment was blinded

Country in which trial was carried out: UK

Method of recruitment of participants: Potential participants were identified from two general practice databases and were sent a letter outlining the study and inviting to participate

Setting: outpatient physiotherapy department

Was the sample size justified with a priori calculation of effect size/power?no (no data available on which to base a power calculation; sample size of 60 was considered convenient and adequate)

Length of follow-up: 12 months

Dropouts:Eight (23%) dropped out from the physiotherapy group (one withdrew, one was diagnosed with spinal stenosis, two developed unrelated health problems, one had heart surgery, one moved away, two stopped attending), eight (28%) dropped out from the self-management group (one withdrew, one developed heart problems, one was diagnosed with polymyalgia rheumatica, one developed hip complications, one had related knee surgery, one moved away, one stopped attending)


ParticipantsInclusion criteria

  • Clinical OA based on clinical presentation and history (mild, moderate or severe non-specific knee pain, lasting > six months with no identifiable recent cause)
  • Aged over 50
  • Presented to GP with chronic knee pain


Exclusion criteria

  • Reporting that knee pain emanated from knee trauma within the past year
  • Lower limb arthroplasty
  • Unstable, coexisting medical or psychological conditions
  • Physiotherapy for knee pain in the previous 12 months
  • Intra-articular injections in the previous six months
  • Unable or unwilling to exercise
  • Unable to walk 100 metres
  • Insufficient command of English to complete the assessment and undertake the intervention
  • Other joint pain that would prevent them from participating in an exercise programme
  • Taking steroids
  • Wheelchair bound


Baseline characteristics

Baseline characteristics were similar in all treatment groups

Intervention group: ESCAPE-Knee pain (N = 29 randomly assigned, 26 analysed post-treatment, 21 analysed after 12 months of FU)

Location of OA: 100% knee

BMI (mean (range)): 30 (20 to 42) kg/m2

Duration of OA (mean (range)): 13 (1 to 30) years

PROGRESS-Plus

Place of residence: UK

Sex: 24% male, 76% female

Age (mean (range)), years: 66 (53 to 81)

Control group: physiotherapy (N = 35 randomly assigned, 31 analysed post-treatment, 27 analysed after 12 months of FU)

Location of OA: 100% knee

BMI (mean (range)): 29 (20 to 47) kg/m2

Duration of OA (mean (range)): 12 (0.5 to 55) years

PROGRESS-Plus

Place of residence: UK

Sex: 37% male, 63% female

Age (mean (range)), years: 67 (51 to 76)


InterventionsIntervention: ESCAPE-Knee pain

Description: The programme aims to change people’s behaviour by challenging inappropriate beliefs regarding their condition and physical activity, encouraging regular exercise and enabling self-management. It exists of informal group discussions that promote shared learning, information and advice about simple coping strategies, problem-solving and planning skills and active participation in a progressive exercise regimen. By the end of the programme, participants have learnt how to utilise physical activity to self-manage their symptoms

Intended audience: people with chronic knee pain (ascribed to OA)

Mode: group sessions (six participants)

Personnel: physiotherapist

Delivery method: face-to-face, written information summary

Language: English

Format: standard format and tailored to individual needs (exercises)

Location: local authority adult education centre

Duration: 10 sessions of one hour, held twice a week (total duration of five weeks)

Additional treatment during trial: unclear

Were the following heiQ components addressed?

Health-directed behaviour: yes

Positive and active engagement in life: yes

Emotional well-being: yes

Self-monitoring and insight: yes

Constructive attitudes and approaches: yes

Skill and technique acquisition: yes

Social integration and support: no

Health service navigation: no

Comparator: physiotherapy

Type: alternate intervention

Description: usual care physiotherapy, with the treatment modalities that the physiotherapist believed were necessary

Frequency: 30 to 45-minute assessment, and after that, up to a maximum of 10 sessions

Additional treatment during trial: unclear


OutcomesOutcome assessed at: baseline, post-treatment and 12 months of follow-up

Primary outcomes of study

  • Function self-reported (WOMAC subscale physical function, 0 to 68, lower score is better)


Secondary outcomes of study

  • Pain (WOMAC subscale pain, 0 to 20, lower score is better)
  • Functional performance (Aggregate Functional Performance Test (AFPT), 0 to infinite, lower score is better)
  • Quality of life (EQ-5D, 0 to 1, higher score is better)
  • Exercise-related health beliefs (0 to 68, lower score is better)
  • Emotional distress (Hospital Anxiety and Depression scale (HADS), 0 to 21, lower score is better)
  • Healthcare utilisation (Client Services Receipt Inventory (CSRI))


NotesWe extracted the following outcomes at post-treatment (short term) and 12 months (intermediate term) for the analyses in this review: pain (WOMAC subscale pain), function self-reported (WOMAC subscale function), quality of life (EQ-5D), emotional distress (HADS subscale depression) and dropouts (proportion of missing participants)

Funding provided by the Physiotherapy Research Foundation Project (number PRF/03/3)


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Unclear riskQuote: “(…) participants were allocated to receive outpatient physiotherapy or ESCAPE-knee pain using a randomisation list generated (…) at a centre away from Sevenoaks Hospital (…)”

Comment: It remains unclear how the random sequence was generated

Allocation concealment (selection bias)Low riskQuote: “(…) and held at a centre away from Sevenoaks Hospital to ensure concealed allocation”

Comment: Allocated treatment was concealed from personnel during inclusion of participants

Blinding of participants and personnel (performance bias)
All outcomes
High riskComment: Participants were not blinded from their allocated treatment, and therefore outcomes might be biased. Physiotherapists treating the participants (both intervention and control groups) were not blinded from the allocated treatment group

Blinding of outcome assessment (detection bias)
All outcomes
High riskQuote: “The assessor was unaware of each participant’s treatment allocation”

Comment: Outcome assessors were blinded to treatment allocation; however, most outcomes are subjective, and participants are not blinded to group allocation

Incomplete outcome data (attrition bias)
All outcomes
High riskQuote: “All clinical and cost data analyses were by intention-to-treat (i.e. participant data were analysed in the groups to which they were randomised) (…)”

Comment: Although dropout rates are equal between groups and the reasons given indicate no withdrawals due to treatment allocation, the rate of dropout is high in both groups. Authors report using an intention-to-treat analysis

Selective reporting (reporting bias)Low riskComment: All prespecified outcomes in the protocol were reported in the Outcomes section

Other biasLow riskComment: No other potential sources of bias were identified

Keefe 1990

MethodsStudy design:RCT, single-centre, three arms, non-blinded

Country in which trial was carried out:US

Method of recruitment of participants:unclear

Setting:outpatients

Was the sample size justified with a priori calculation of effect size/power?no justification provided

Length of follow-up: six months

Dropouts:After six months, three (10%) dropped out from usual care, one (3%) dropped out from arthritis education and two (6%) dropped out from pain coping skills training. No reasons provided


ParticipantsInclusion criteria

  • Diagnosed as having OA of the knee on the basis of medical evaluation and radiographic examination


Exclusion criteria

  • Having arthritic disorders other than OA
  • Having other known organic disease that would significantly affect function (e.g. chronic obstructive pulmonary disease)
  • Patients receiving disability support payments for OA


Baseline characteristics

Baseline characteristics were similar among all treatment groups

Intervention group: pain coping skills (N = 32 randomly assigned, 31 analysed)

Location of OA: 100% knee

Mean percentage above ideal weight (mean (SD)): 21.0 (17.6)

PROGRESS-Plus

Place of residence: US

Sex: 22% male, 78% female

Age (mean (SD)), years: 62.4 (12.0)

Control group: arthritis education (N = 36 randomly assigned, 35 analysed)

Location of OA: 100% knee

Mean percentage above ideal weight (mean (SD)): 27.2 (29.0)

PROGRESS-Plus

Place of residence: US

Sex: 31% male, 69% female

Age (mean (SD)), years: 66.0 (9.5)

Control group: standard care control (N = 31 randomly assigned, 28 analysed)

Location of OA: 100% knee

Mean percentage above ideal weight (mean (SD)): 24.1 (22.2)

PROGRESS-Plus

Place of residence: US

Sex: 32% male, 68% female

Age (mean (SD)), years: 63.0 (13.0)


InterventionsIntervention: pain coping skills

Description: Participants in this condition received a cognitive-behavioural intervention. To help participants understand the rationale for training in pain coping skills, a simplified version of Melzack and Wall’s gate control model was used to show that pain is a complex experience affected by thoughts, feelings and behaviours. Relaxation, imagery and distraction techniques were introduced as methods for controlling pain through attention diversion. Cognitive restructuring was utilised to help participants recognise and modify irrational cognitions related to pain

During the six months of follow-up, participants were called three times by their group therapists (one month, two months and four months after completion of treatment). The calls focused on a review of progress since completion of the group and a discussion of participants' use of pain coping skills

Intended audience: people with knee OA

Mode: group sessions(six to nine participants)

Personnel: psychologist and nurse

Delivery method: face-to-face

Language: English

Format: standard format

Location: -

Duration: 10 weekly sessions, each of 90-minute duration

Additional treatment during trial: unclear

Were the following heiQ components addressed?

Health-directed behaviour: no

Positive and active engagement in life: no

Emotional well-being: no

Self-monitoring and insight: yes

Constructive attitudes and approaches: yes

Skill and technique acquisition: yes

Social integration and support: no

Health service navigation: no

Comparator: arthritis education

Type: attention control

Description: The arthritis education intervention used a lecture-discussion format developed by Lorig. Four topics were discussed: the nature of OA, treatment methods, exercise and mobility and function

During the six months of follow-up, participants were called three times by their group therapists (one month, two months and four months after completion of treatment). The calls focused on a review of progress since completion of the group and answering any questions that participants had about their arthritis and its treatment

Frequency: 10 weekly sessions of 90-minute duration

Additional treatment during trial: unclear

Comparator: standard care control

Type: usual care

Description: Participants in this condition continued with their routine care for OA

Additional treatment during trial: Participants did not attend any group sessions for arthritis education or pain coping skills training


OutcomesOutcome assessed at: baseline, post-treatment and six months of follow-up

Primary outcomes of study

  • Constructive attitudes and approaches (Coping Strategies Questionnaire (CSQ) subscales)
  • Pain (AIMS subscale pain, 0 to 10, lower score is better)
  • Function self-reported (AIMS subscale physical disability, 0 to 10, lower score is better)
  • Emotional distress (AIMS subscale psychological disability, 0 to 10, lower score is better)
  • Medication use
  • Direct observation of motor pain behaviour (videotapes)


NotesWe extracted the following outcomes at post-treatment (short term) and at six months (intermediate term) for the analyses in this review: pain (AIMS subscale pain), function self-reported (AIMS subscale physical disability), emotional distress (AIMS subscale psychological disability) and dropouts (proportion of missing participants)

The trial was supported by NIAMS Grant No. RO1 AM NS 35270


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low riskQuote: “(…) and were randomly assigned (using a random number table) (…)”

Comment: Method of randomisation with low risk of bias is used

Allocation concealment (selection bias)Unclear riskComment: No information provided on allocation concealment

Blinding of participants and personnel (performance bias)
All outcomes
High riskComment: These types of interventions and controls make it impossible for authors to blind participants and personnel, which introduces a risk of performance bias

Blinding of outcome assessment (detection bias)
All outcomes
High riskComment: Outcomes were assessed through self-report and might therefore have been biased, as participants were not blinded

Incomplete outcome data (attrition bias)
All outcomes
Low riskComment: The authors provide no information on the manner in which the analyses were performed and how missing data were handled. However, the rate of dropout was low and equal among treatment groups

Selective reporting (reporting bias)Unclear riskComment: Constructive attitudes and approaches and medication use not reported

Other biasLow riskComment: No other potential sources of bias were identified

Keefe 1996

MethodsStudy design:RCT, three arms, non-blinded

Country in which trial was carried out:US

Method of recruitment of participants:unclear

Setting:unclear

Was the sample size justified with a priori calculation of effect size/power?no justification for sample size provided

Length of follow-up:12 months

Dropouts:Nine (30%) dropped out from SA-CST, five (17%) dropped out from CST, four (14%) dropped out from AE-SS. No reasons provided


ParticipantsInclusion criteria

  • Diagnosed as having OA of the knee
  • Being married
  • No arthritic disorder other than OA
  • No other known organic disease that would significantly affect function (e.g. chronic obstructive pulmonary disease)
  • Were not receiving disability support payments for OA


Exclusion criteria:none

Baseline characteristics

Baseline characteristics were similar among all treatment groups

Intervention group: coping skills training (CST) (N = 29 randomly assigned, 26 analysed post-treatment, 24 analysed after six and 12 months of FU)

Location of OA: 100% knee

PROGRESS-Plus

Place of residence: US

Sex: 48% males, 52% females

Age (mean), years: 61.4

Intervention group: spouse-assisted CST (N = 30 randomly assigned, 27 analysed post-treatment, 25 analysed after six months, 21 analysed after 12 months of FU)

Location of OA: 100% knee

PROGRESS-Plus

Place of residence: US

Sex: 40% males, 60% females

Age (mean), years: 63.5

Control group: arthritis educationspousal support (N = 29 randomly assigned, 28 analysed post-treatment, 26 analysed after six months, 25 analysed after 12 months of FU)

Location of OA : 100% knee

PROGRESS-Plus

Place of residence: US

Sex: 32% male, 68% female

Age (mean), years: 62.8


InterventionsIntervention: coping skills training (CST)

Description: “The rationale for the CST intervention was based on Melzack and Wall’s gate control model of pain, which views pain as a complex experience, affected by thoughts, feelings, and behaviours. Patients were told that a major focus of the group was developing a menu of pain-coping skills. Three sets of coping skills were included: attention diversion skills, activity-based skills, and cognitive coping strategies”

Intended audience: married participants with knee OA

Mode: group sessions (four to six participants, without their spouses)

Personnel: nurse and a psychologist, both trained in the CST programme

Delivery method: ace-to-face

Language: English

Format: standard format

Location: -

Duration: 10 weekly sessions of two hours

Additional treatment during trial: unclear

Were the following heiQ components addressed?

Health-directed behaviour: no

Positive and active engagement in life: no

Emotional well-being: no

Self-monitoring and insight: yes

Constructive attitudes and approaches: yes

Skill and technique acquisition: yes

Social integration and support: no

Health service navigation: no

Intervention: spouse assistedCST

Description: Each session included training in pain coping skills and couples skills. Participants and their spouses were provided with a rationale that emphasised (1) OA is a couples issue that can affect each partner and his or her relationships, (2) developing ways to cope with OA is an important task for each couple and (3) the reactions of the spouse can influence the participant's ability to cope with pain. Pain coping strategies were the same as in the CST group. Couples skills were designed to supplement and reinforce the participant's pain coping skills; they included communication skills, behavioural rehearsal, mutual goal setting, joint home and in vivo practice and maintenance training

Intended audience: married participants with knee OA

Mode: group sessions (four to six participants and their spouses)

Personnel: nurse and psychologist, both trained in the CST programme

Delivery method: face-to-face

Language: English

Format: standard format

Location: -

Duration: ten weekly sessions of two hours

Additional treatment during trial: unclear

Were the following heiQ components addressed?

Health-directed behaviour: no

Positive and active engagement in life: no

Emotional well-being: no

Self-monitoring and insight: yes

Constructive attitudes and approaches: yes

Skill and technique acquisition: yes

Social integration and support: yes

Health service navigation: no

Comparator: arthritis educationspousal support

Type: alternate intervention

Description: The group was provided a detailed overview of the nature of OA, methods of diagnosis, medical and surgical treatments, home remedies and methods for maintaining mobility and flexibility. Spouses attended each session along with the participant. Spouses were encouraged to participate fully in all discussions of educational material

Frequency: 10 weekly group sessions (four to six participants and their spouses) of two hours

Additional treatment during trial: unclear


OutcomesOutcome assessed at: baseline, post-treatment and six and 12 months of follow-up

Primary outcomes of study

  • Marital adjustment (Dyadic Adjustment Scale (DAS))
  • Self-management (Arthritis Self-Efficacy Scale (ASES) subscales pain (10 to 100), function (10 to 100) and other symptoms (10 to 100), higher scores are better)
  • Pain (AIMS subscale pain, 0 to 10, lower score is better)
  • Function self-reported (AIMS subscale physical disability, 0 to 10, lower score is better)
  • Emotional distress (AIMS subscale psychological disability, 0 to 10, lower score is better)
  • Constructive attitudes and approaches (Coping Strategies Questionnaire (CSQ) subscales coping attempts, pain control and rational thinking)
  • Pain behaviour (videotape)


NotesWe extracted the following outcomes at post-treatment (short term) and at 12 months (intermediate term) for the analyses in this review:

self-management (ASES mean), pain (AIMS subscale pain), function self-reported (AIMS subscale physical disability), emotional distress (AIMS subscale psychological disability), constructive attitudes and approaches (CSQ subscale coping attempts) and dropouts (proportion of missing participants)

The trial was supported by NIAMS Grant AR-35270

Data analysis: Intervention groups were judged to be similar enough to combine by pooling data


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Unclear riskQuote: “All patients were randomly assigned to 1 of 3 treatment conditions (…)”

Comment: no information provided on randomisation method

Allocation concealment (selection bias)Unclear riskComment: no information provided on allocation concealment

Blinding of participants and personnel (performance bias)
All outcomes
High riskThese types of controls and interventions make blinding of participants and personnel not possible, which may have introduced bias

Blinding of outcome assessment (detection bias)
All outcomes
High riskComment: no information provided on blinding of outcome assessors; however, most outcomes are subjective, and participants are not blinded to group allocation, introducing a risk of detection bias

Incomplete outcome data (attrition bias)
All outcomes
High riskComment: Dropout rate immediately post-treatment showed no significant differences. After 12 months, dropout seems unequal and is moderate to high (30%, 17% and 14%). Reasons are not provided. Furthermore, only completers are included in the analyses

Selective reporting (reporting bias)Low riskComment: All outcome measures listed in the Methods section are reported in Results

Other biasLow riskComment: No other potential sources of bias were identified

Keefe 2004

MethodsStudy design:RCT, multi-centre, four arms, non-blinded

Country in which trial was carried out: US

Method of recruitment of participants:Participants were recruited from rheumatology clinics and advertisements placed in newspapers

Setting:outpatients and general population

Was the sample size justified with a priori calculation of effect size/power? no, sample size not justified

Length of follow-up: 12 weeks (as long as intervention takes)

Dropouts:Two (11%) dropped out from SA-CST, one (5%) dropped out from SA-CST + Ex, two (11%) dropped out from usual care, no one dropped out from the exercise group. No reasons provided


ParticipantsInclusion criteria

  • Diagnosed with knee OA by a board-certified rheumatologist
  • Being married


Exclusion criteria

  • Comorbid medical conditions that could affect their health status over the course of the trial (e.g. a recent myocardial infarction)
  • An abnormal cardiac response to exercise (e.g. exercise-induced ventricular tachycardia, abnormal blood pressure response)
  • Other known organic disease that would contraindicate safe participation in the study (e.g. chronic obstructive pulmonary disease, congestive heart failure, cancer)


Baseline characteristics

Participants in the spouse-assisted CST alone and spouse-assisted CST plus exercise conditions had higher pretreatment levels of pain than those in the two other conditions

Intervention group: spouse-assisted CST (N = 18 randomly assigned, 18 analysed)

Location of OA: 100% knee

PROGRESS-Plus

Place of residence: US

Sex: 50% male, 50% female

Age (mean (SD)), years: 60.0 (12.2)

Intervention group: spouse-assisted CST + exercise (N = 20 randomly assigned, 20 analysed)

Location of OA: 100% knee

PROGRESS-Plus

Place of residence: US

Sex: 35% male, 65% female

Age (mean (SD)), years: 60.2 (9.1)

Control group: exercise (N = 16 randomly assigned, 16 analysed)

Location of OA: 100% knee

PROGRESS-Plus

Place of residence: US

Sex: 62.5% male, 37.5% female

Age (mean (SD)), years: 60.3 (8.7)

Control group: standard care (N = 18 randomly assigned, 18 analysed)

Location of OA: 100% knee

PROGRESS-Plus

Place of residence: US

Sex: 38.9% male, 61.1% female

Age (mean (SD)), years: 57.6 (14.3)


InterventionsIntervention: SA-CST

Description: Couples were provided with a rationale that emphasised that (1) pain is a complex experience, which, as the Gate Control Theory suggests, can be influenced by thoughts, feelings and behaviours; (2) participants and their spouses can acquire and maintain skills for managing pain through frequent practice; and (3) because OA is a couples issue that affects each partner and his or her relationships, involving the spouse in training can be quite helpful. Training sessions emphasised active learning. Group leaders provided feedback and suggestions to enhance the efficacy of skills practice. Each session involved two major components: (1) training in pain coping skills, and (2) training in couples skills designed to supplement and reinforce the participant's pain coping skills

Intended audience: participants with knee OA and their spouses

Mode: group sessions (three to five couples)

Personnel: PhD-level psychologists who had been trained in SA-CST

Delivery method: face-to-face

Language: English

Format: standard format

Location: -

Duration: 12 once-weekly, two-hour sessions (12 weeks) (total: 24 hours)

Additional treatment during trial: unclear

Were the following heiQ components addressed?

Health-directed behaviour: no

Positive and active engagement in life: no

Emotional well-being: no

Self-monitoring and insight: yes

Constructive attitudes and approaches: yes

Skill and technique acquisition: yes

Social integration and support: yes

Health service navigation: no

Comparator: exercise

Type: alternate intervention

Description: The exercise programme included (1) cardiopulmonary endurance training, (2) strength training and (3) flexibility/range of motion training

Frequency: 60-minute sessions, three times a week, for 12 weeks (total: 26 hours)

Additional treatment during trial: unclear

Comparator: standard care

Type: usual care

Description: Participants continued to receive their routine care; neither they nor their spouses attended coping skills training sessions or exercise sessions

Additional treatment during trial: unclear

Note:The SA-CST + exercise programme is a combination of two programmes described above, with a total duration of 50 hours


OutcomesOutcome assessed at: baseline and post-treatment

Primary outcomes of study

  • Functional performance (aerobic fitness (maximal effort bicycle ergometry exercise test))
  • Functional performance (muscle strength (maximal effort leg extensions, leg flexions and bicep curls)
  • Constructive attitudes and approaches (Coping Strategies Questionnaire (CSQ) subscales)
  • Self-management (Arthritis Self-Efficacy Scale (ASES), 30 to 300, higher score is better)
  • Marital adjustment (Dyadic Adjustment Scale (DAS))
  • Pain (AIMS subscale pain, 0 to 10, lower score is better)
  • Emotional distress (AIMS subscale psychological disability, 0 to 10, lower score is better)


NotesWe extracted the following outcomes at post-treatment (short term) for the analyses in this review: self-management (ASES), pain (AIMS subscale pain), emotional distress (AIMS subscale psychological disability), constructive attitudes and approaches (CSQ subscale coping attempts) and dropouts (proportion of missing participants)

This research was supported by National Institute of Arthritis and Musculoskeletal Diseases Grant No. AR-35270

Data analysis: Intervention groups were judged to be similar enough to combine by pooling data


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Unclear riskQuote: “Patients completed a baseline evaluation and were then randomly assigned to one of four conditions”

Comment: no information on method of randomisation provided

Allocation concealment (selection bias)Unclear riskComment: no information provided on allocation concealment

Blinding of participants and personnel (performance bias)
All outcomes
High riskComment: This type of control and intervention makes blinding of participants impossible, and this may have influenced the results

Comment: This type of control and intervention makes blinding of personnel impossible; usual care probably is not influenced by performance bias, but not blinding the personnel in the other intervention groups may have introduced bias

Blinding of outcome assessment (detection bias)
All outcomes
High riskComment: No information was provided on blinding of outcome assessors; however, most outcomes are subjective, and participants were not blinded to group allocation, introducing a risk of detection bias

Incomplete outcome data (attrition bias)
All outcomes
Low riskQuote: “Over the course of the study, five subjects dropped out of the study, two from the spouse-assisted coping skills training condition, one from the spouse-assisted coping skills training plus exercise training condition, and two from the standard care control condition”

Comment: Low dropout rate, equally distributed among treatment groups. No information was provided on the analysis techniques used

Selective reporting (reporting bias)Low riskComment: All outcomes listed in the Methods section are reported in the Results

Other biasHigh riskQuote: “No group differences were found on any of the pre-treatment measures except for the AIMS Pain measure. (…) Patients in the spouse-assisted CST alone and spouse-assisted CST plus exercise conditions had higher pre-treatment levels of pain than those in the two other conditions”

Comment: A higher baseline pain level in both intervention treatment arms could possibly enlarge the observed treatment effect (type I error)

Lorig 2008

MethodsStudy design:RCT, single-centre, two arms, non-blinded

Country in which trial was carried out: US

Method of recruitment of participants:Participants were recruited by links to the study website placed on established websites, online newsletters and discussion groups. Calendar announcements and articles in newspapers also directed subjects to the study website. Potential participants could leave their e-mail address

Setting:general population

Was the sample size justified with a priori calculation of effect size/power?No justification of the sample size was provided

Length of follow-up:one year

Dropouts:126 (29%) dropped out of the intervention group; 78 (18%) dropped out of the control group. No reasons provided


ParticipantsInclusion criteria

  • A self-reported diagnosis of OA, RA or fibromyalgia (confirmation requested from participants' physicians, of whom 68% replied and confirmed diagnosis in all but six cases)
  • Aged 18 years or older
  • Access to a computer with Internet and e-mail capabilities
  • Agreed to one to two hours per week of log-on time spread over at least three sessions per week for six weeks
  • Able to complete the online questionnaire


Exclusion criteria

  • Being in active treatment for cancer for one year or less
  • Participating in the small-group ASMP or the Chronic Disease Self-Management Program


Baseline characteristics (including RA and FM participants)

Usual care control participants had an average of ˜ one chiropractic visit more in the past six months

Intervention group: Internet-based ASMP (OA subgroup: N = 275 randomly assigned, 134 analysed)

PROGRESS-Plus

Place of residence: US

Race, ethnicity and culture: non-Hispanic white 90.9%

Sex: 10.2% male, 89.8% female

Education (mean (SD)): 15.6 (3.09) years of education

Social capital: 65.5% married

Age (mean (SD)), years: 52.2 (10.9)

Control group: usual care (OA subgroup: N = 276 randomly assigned, 158 analysed)

PROGRESS-Plus

Place of residence: US

Race, ethnicity and culture: non-Hispanic white 93.7%

Sex: 9.5% male, 90.5% female

Education (mean (SD)): 15.7 (3.11) years of education

Social capital: 71.1% married

Age (mean (SD)), years: 52.5 (12.2)

Note:The six participants for whom diagnosis was not confirmed by their physician were dropped from analysis.


InterventionsIntervention: Internet-based ASMP

Description: The Internet ASMP consists of password-protected, interactive, Web-based instruction; Web-based bulletin board discussion; tools that the participants can use individually (e.g. exercise logs, medication diaries, tailored exercise programmes); and the Arthritis Helpbook. The programme focused on reduction of pain and improvement of function. Topics covered were self-management principles, goal setting/action plans, pain management, relaxation/cognitive pain management, problem-solving steps, fitness/exercise, feedback/problem solving, difficult emotions, healthy eating, osteoporosis, fatigue and energy conservation, medication, depression, working with your healthcare professional, evaluating treatment plans and sleep

Intended audience: people with OA, RA and/or fibromyalgia

Mode: individual

Personnel: a pair of peer moderators

Delivery method: Internet

Language: English

Format: tailored to individual needs and standard format

Location: anywhere with access to computer with Internet

Duration: six weeks, at least three times per week for one to two hours

Additional treatment during trial: Participants were allowed to continue usual care

Were the following heiQ components addressed?

Health-directed behaviour: yes

Positive and active engagement in life: yes

Emotional well-being: yes

Self-monitoring and insight: yes

Constructive attitudes and approaches: yes

Skill and technique acquisition: yes

Social integration and support: no

Health service navigation: yes

Comparator: usual care

Type: usual care

Description: not specified

Additional treatment during trial: unclear


OutcomesOutcome assessed at: baseline and six and 12 months of follow-up

Primary outcomes of study

  • Pain (10-cm VAS, 0 to 10, lower score is better)
  • Function self-reported (fatigue on 10-cm VAS, 0 to 10, lower score is better)
  • Function self-reported (Activities Limitation Scale, 0 to 4, lower score is better)
  • Emotional distress (Health Distress Scale, 0 to 5, lower score is better)
  • Function self-reported (Health Assessment Questionnaire (HAQ), 0 to 3, lower score is better)
  • Global OA scores (Self-Rated Global Health Scale, 0 to 5, lower score is better)
  • Health-directed activity (aerobic exercise (minutes/wk), higher is better)
  • Health-directed activity (stretching and strengthening exercise (minutes/wk), higher is better)
  • Health-directed activity (practice of stress management (times/wk), higher is better)
  • Health service navigation (communication with physician, 0 to 5, higher score is better)
  • Health service utilisation (visits to physician, emergency room, chiropractor, physical therapist and nights in hospital)
  • Self-management (Arthritis Self-Efficacy Scale (ASES), 1 to 10, higher score is better)


NotesWe extracted the following outcomes at 12 months (intermediate term) for the analyses in this review: self-management (ASES), pain (VAS), global OA scores (Self-Rated Global Health Scale), function self-reported (HAQ), emotional distress (health distress scale), health-directed activity (practice of stress management), health service navigation (communication with physician) and dropouts (proportion of missing participants)

This study was supported by NIH Center for Disease Control’s Arthritis Grant (AR-43538)

Control participants were sent a $10 Amazon.com certificate for each questionnaire completed

Data analysis: Change scores were combined with end point scores using generic inverse variance


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Unclear riskQuote: “Following completion of the online questionnaire, participants were randomized to either the intervention group or to a control group”

Comment: no information on the method of randomisation provided

Allocation concealment (selection bias)Unclear riskComment: no information provided on the allocation concealment

Blinding of participants and personnel (performance bias)
All outcomes
High riskComment: Participants were not blinded during the trial, which introduces a risk of performance bias. Although personnel also were not blinded, it is unlikely that this has biased the results because the physicians providing usual care were not involved in the trial

Blinding of outcome assessment (detection bias)
All outcomes
High riskComment: Participants self-reported the outcomes and were not blinded from their allocated treatment

Incomplete outcome data (attrition bias)
All outcomes
High riskQuote: “All analyses were also run substituting last-reported value for any missing cases (intent-to-treat [ITT] analyses)”

Comment: A large number of participants dropped out with no reasons provided. Numbers of dropouts differed between groups (29% in the intervention group vs 18% in the control group). Also, non-completers at six and 12 months differed significantly from completers (non-completers e.g. were younger and had significantly higher levels of health distress and activity limitation at baseline)

Selective reporting (reporting bias)Low riskComment: All outcomes listed in the Methods section were reported in the Results

Other biasLow riskComment: No other potential sources of bias were identified

Maisiak 1996

MethodsStudy design: RCT, single-centre, three arms, outcome assessment blinded

Country in which trial was carried out: US

Method of recruitment of participants: Recruitment over a two-year period from lists of previous callers to the Arthritis Information Service (AIS) of Alabama and subscribers to an arthritis newsletter, and from newspaper advertisements

Setting: primary care and outpatients

Sample size justified with a priori calculation of effect size/power? yes

Length of follow-up: nine months

Dropouts: After nine months of follow-up, eight (6%) dropped out from usual care, 11 (8%) dropped out from symptom monitoring and seven (7%) dropped out from treatment counselling (11 of 26 were in the OA subgroup)


ParticipantsInclusion criteria

  • A diagnosis of primary OA of the hip or knee or a diagnosis of primary RA made by the patient’s physician (letter documenting diagnosis and physician’s confidence in that diagnosis (no ACR criteria had to be provided)
  • Some current pain or some current disability due to arthritis
  • ≥ 21 years of age
  • Able to communicate by telephone over a nine-month period
  • Residing in Alabama


Exclusion criteria

  • Serious (non-rheumatologic) comorbidities that might affect study participation


Baseline characteristics (include RA patients ˜ 55%)

Baseline characteristics were similar in all treatment groups

Intervention group: treatment counselling (N = 135 randomly assigned, 128 included in analysis, including RA patients; 62 in OA subgroup)

Mean disease duration: 16.5 years (including RA patients)

PROGRESS-Plus

Place of residence: Alabama

Race, ethnicity and culture: 87% white

Sex: 4% male, 96% female

Education: 12.1 mean years of schooling (including RA patients)

Age (mean), years: 60.1 (SD not specified)

Disability: 43% see a specialist (including RA patients)

Control group: symptom monitoring (N = 135 randomly assigned, 124 included in analysis, including RA patients; 54 in OA subgroup)

Mean disease duration: 15.9 years (including RA patients)

PROGRESS-Plus

Place of residence: Alabama

Race, ethnicity and culture: 85% white

Sex: 6% male, 94% female

Education: 12.0 mean years of schooling (including RA patients)

Age (mean), years: 60.7

Disability: 51% see specialist (including RA patients)

Control group: usual care (N = 135 randomly assigned, 127 included in analysis, including RA patients; 54 in OA subgroup)

Mean disease duration: 15.7 years (including RA patients)

PROGRESS-Plus

Place of residence: Alabama

Race, ethnicity and culture: 82% white

Sex: 13% male, 87% female

Education: 12.2 mean years of schooling (including RA patients)

Age (mean), years: 60.5

Disability: 50% see specialist (including RA patients)


InterventionsIntervention group: treatment counselling

Description: Six categories of patient behaviour were targeted for potential change, including patient-physician communication, medication compliance, removing barriers to medical care, symptom reviews, self-care activities and stress control using a 13-page written, structured protocol especially designed for counselling patients

Intended audience: people with OA or RA

Mode: individual

Personnel: trained counsellors

Delivery method: telephone and mail

Language: English

Format: structured protocol with options for tailoring to individual needs

Location: home

Duration: 20 minutes per session, five times at two-week intervals during first three months, and six more times at four-week intervals during the last six months

Additional treatment during trial: unclear

Were the following heiQ components addressed?

Health-directed behaviour: yes

Positive and active engagement in life: no

Emotional well-being: no

Self-monitoring and insight: yes

Constructive attitudes and approaches: no

Skill and technique acquisition: yes

Social integration and support: no

Health service navigation: yes

Control group: symptom monitoring

Type: attention control

Description: The purpose was to provide a detailed review of the participant's symptoms and to provide attention to the participant in an amount equal to that provided to participants in the treatment counselling group. During each session, the specialist would administer questions regarding symptom assessment from the AIMS2. The symptom monitoring specialist was not allowed to ask any other questions concerning arthritis and did not provide any advice to the participant

Frequency: 20 minutes per session, five times at two-week intervals during first three months, and six more times at four-week intervals during the last six months

Additional treatment during trial: unclear

Other relevant information: The symptom monitoring specialists were temporary, part-time staff who tended to be college students with little or no training in arthritis education or counselling. Each received two hours of training on administering the AIMS2 questions by phone

Control group: usual care

Type: usual care

Description: Participants in the usual care control group were not contacted by the study staff outside of their three assessments (baseline, six and nine months)

Additional treatment during trial: no restriction in using any other outside sources of assistance, including the Arthritis Information Service in Alabama


OutcomesOutcomes assessed at: baseline and three, six and nine months of follow-up

Primary outcome

  • Global OA scores (AIMS-2 total health status, 0 to 10, lower score is better)


Secondary outcome

  • Number of visits to physician


NotesWe extracted the following outcome at nine months (intermediate term) for the analyses in this review: global OA scores (AIMS-2 total health status) and dropouts (proportion of missing participants)

Supported by a Multipurpose Arthritis and Musculoskeletal Diseases Center grant from the National Institute of Arthritis and Musculoskeletal and Skin Diseases

Author (R Maisiak) sent additional information about the trial on request


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low riskQuote: “The random sequence was generated by a standard random number software program on a computer” [additional information provided by author]

Comment: This method of randomisation has a low risk of introducing bias

Allocation concealment (selection bias)Unclear riskQuote: “The sequence of assignments was printed on paper and stored in a locked cabinet” [additional information provided by author]

Comment: Allocation seems appropriately concealed; however, it is not clear whether allocation was concealed for the telephone interviewers who assigned participants to treatment

Blinding of participants and personnel (performance bias)
All outcomes
High riskComment: Participants were not blinded to treatment allocation, which might have introduced bias

Comment: Personnel were not blinded to treatment allocation, which might have introduced bias

Blinding of outcome assessment (detection bias)
All outcomes
High riskQuote: “The assessment interviewers were mostly college students who were not told of the overall purpose of the study, and were blinded to the group assignment of the patients they were interviewing”

Comment: Outcomes were assessed by blinded personnel; however, most outcomes are subjective, and participants are not blinded to group allocation

Incomplete outcome data (attrition bias)
All outcomes
Low riskComment: Although no intent-to-treat analysis seems to be performed, the quantity of missing data is very low (6%) for the OA group, and reasons for dropout are unlikely to have biased the results significantly. It is unclear how missing data were divided between treatment groups

Selective reporting (reporting bias)Unclear riskQuote: “Other measures were also included in the assessment and were unrelated to this report”

Comment: Several other measures were assessed but not reported. It is unclear what these measures were

Other biasLow riskComment: No other potential sources of bias were identified

Martire 2007

MethodsStudy design:RCT, single-centre, three arms, non-blinded

Country in which trial was carried out:US

Method of recruitment of participants:not specified

Setting:outpatients

Was the sample size justified with a priori calculation of effect size/power?no justification of sample size provided

Length of follow-up:six months

Dropouts:21 (39%) dropped out of usual care, 12 (13%) dropped out of PES and 17 (17%) dropped out of CES. Reasons for dropout in the usual care group were dissatisfaction with assignment to this group (six), health problems in partner or family members (four) and unknown (11).


ParticipantsInclusion criteria

  • Diagnosed with hip or knee OA
  • 50 years of age or older
  • Married
  • Pain of at least moderate intensity on most days over the past month
  • Difficulty with at least one instrumental activity of daily living (e.g. household tasks, driving)
  • Received assistance from the spouse with at least one instrumental activity of daily living


Exclusion criteria

  • Attended the Arthritis Self-Management Program in the past five years
  • Had a comorbid diagnosis of fibromyalgia or rheumatoid arthritis


Baseline characteristics

Participants in the CES group had more depressive symptoms than participants in the PES group

Intervention group: PES (N = 89 randomly assigned, 89 analysed)

Location of OA: hip and knee

Duration of OA (mean (SD)): 15.3 (11.8) years

PROGRESS-Plus

Place of residence: US

Sex: 28% male, 72% female

Education (mean (SD)): 14.6 (1.7) years of education

Age (mean (SD)), years: 68.0 (8.0)

Intervention group: CES (N = 99 randomly assigned, 99 analysed)

Location of OA: hip and knee

Duration of OA (mean (SD)): 14.3 (9.4) years

PROGRESS-Plus

Place of residence: US

Sex: 26% male, 74% female

Age (mean (SD)), years: 69.2 (7.2)

Control group (N = 54 randomly assigned, 54 analysed)

Location of OA: hip and knee

Duration of OA (mean (SD)): 16.1 (12.0) years

PROGRESS-Plus

Place of residence: US

Sex: 28% male, 72% female

Age (mean (SD)), years: 68.4 (7.5)


InterventionsIntervention: PES

Description: The protocol was based on the Arthritis Self-Management Program (Lorig). No spouses, family members, or friends participated in PES sessions. Major components were information about arthritis, self-management strategies, benefits of exercise, communication skills and effectively coping with negative emotions. At the end of the session, each participant set a health-related goal; the next session, each participant reported on his or her success in meeting this goal and received feedback from the group. Up to five monthly booster sessions were conducted via telephone in the six months of follow-up after treatment

Intended audience: people with hip or knee OA

Mode: group sessions (four to six participants)

Personnel: trained facilitator

Delivery method: face-to-face

Language: English

Format: tailored to individual needs and standard format

Location: -

Duration: weekly sessions of two hours for six weeks

Additional treatment during trial: Participants also received usual care

Were the following heiQ components addressed?

Health-directed behaviour: yes

Positive and active engagement in life: yes

Emotional well-being: yes

Self-monitoring and insight: yes

Constructive attitudes and approaches: yes

Skill and technique acquisition: yes

Social integration and support: no

Health service navigation: no

Intervention: CES

Description: same as PES; topics were framed as couple’s issues whenever possible. Up to five monthly booster sessions were conducted via telephone in the six months of follow-up after treatment

Intended audience: people with hip and knee OA

Mode: group sessions (four to six participants and their spouses)

Personnel: trained facilitator

Delivery method: face-to-face

Language: English

Format: tailored to individual needs and standard format

Location: -

Duration: once a week, session of two hours, six weeks

Additional treatment during trial: Participants also received usual care

Were the following heiQ components addressed?

Health-directed behaviour: yes

Positive and active engagement in life: yes

Emotional well-being: yes

Self-monitoring and insight: yes

Constructive attitudes and approaches: yes

Skill and technique acquisition: yes

Social integration and support: yes

Health service navigation: no

Comparator

Type: usual care

Description: Individuals in this group maintained their regular medical regimens and saw their rheumatologist as needed, and no attempt was made to alter the usual care that these individuals received from their physicians

Additional treatment during trial: Couples in this group did not participate in any arthritis self-management programme or receive any surgical interventions during their study participation


OutcomesOutcome assessed at: baseline, post-treatment and six months of follow-up

Primary outcomes of study

  • Global OA scores (WOMAC total score 0 to 96, lower score is better)
  • Pain (WOMAC subscale pain, 0 to 20, lower score is better)
  • Function self-reported (WOMAC subscale physical function, 0 to 68, lower score is better)
  • Emotional distress (CES-D, 0 to 60, lower score is better)
  • Self-management (Arthritis Self-Efficacy Scale (ASES), 30 to 300, higher score is better)
  • Marital satisfaction (marital adjustment test)


Secondary outcomes of study

  • Measures for spouses: perceived stress, depressive symptoms, caregiver mastery, critical attitudes and marital satisfaction


NotesWe extracted the following outcomes at post-treatment (short term) and at six months (intermediate term) for the analyses in this review: self-management (ASES), pain (WOMAC subscale pain), global OA scores (WOMAC), function self-reported (WOMAC subscale function) and dropouts (proportion of missing participants)

The trial was supported by several grants (Grant P50 HL65111-65112; (Pittsburgh Mind-Body Center) from the National Heart, Lung, and Blood institute; Grants K01 MH065547 and P30 MH52247 from the National institute of Mental Health; Grant K07 AG000923 from the National Institute on Aging; Grant R01 NR008272 from the National Institute of Nursing Research; Grant P01 AR50245 from the National Institute of Arthritis and Musculoskeletal Diseases; and Grant to F.J. Keefe from the Arthritis Foundation)

The author (L Martire) provided additional information about the trial on request

Data analysis: Intervention groups were judged to be similar enough to combine by pooling data


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)High riskQuote: “The first group of accumulated couples was randomly assigned to conditions 1, 2, or 3 using a lottery method. The next group of couples at that site was randomly assigned to the two remaining two study conditions, and so on” [information provided by the author]

Comment: Method of random sequence generation has high risk of bias

Allocation concealment (selection bias)Low riskQuote: “This information was kept in a locked drawer of my office” [information provided by the author]

Comment: Allocation was concealed

Blinding of participants and personnel (performance bias)
All outcomes
High riskComment: This type of intervention makes blinding of participants and personnel to treatment allocation not possible, introducing a risk of bias

Blinding of outcome assessment (detection bias)
All outcomes
High riskQuote: “All assessments were carried out by interviewers who were not involved in intervention implementation”

Comment: Blinding of outcome assessors remains unclear; however, most outcomes are subjective, and participants are not blinded to group allocation, introducing a risk of detection bias

Incomplete outcome data (attrition bias)
All outcomes
High riskQuote: “There was a different rate of attrition across the three groups (…). The reasons for attrition from the usual care group were dissatisfaction with assignment to this group (n=6) (…)."

Comment: A significantly different rate of attrition among treatment groups is noted. Dropout rate is acceptable in both intervention groups (13% in PES, 17% in CES) but very high in the usual care group (39%). An intention-to-treat approach is used to analyse data

Selective reporting (reporting bias)Unclear riskComment: Emotional distress and marital satisfaction were not reported

Other biasHigh riskQuote: “Post hoc analyses (…) indicated that individuals with OA in the CES group had more pre-intervention depressive symptoms than those in the PES group. Thus, the effects of intervention were not examined for this outcome (…)”

Comment: Difference in baseline depressive symptoms may have introduced bias

Comment: Compliance was equal in intervention groups but not very high (24% in PES and 28% in CES did not receive intervention; 33% in both PES and CES attended all sessions)

Maurer 1999

MethodsStudy design: RCT (stratified by disease severity), multi-centre, two arms, outcome assessment blinded

Country in which trial was carried out:US

Method of recruitment of participants:Several methods were used to recruit participants; most patients were involved from direct physician referrals from with study-affiliated clinics

Setting:outpatients

Was the sample size justified with a priori calculation of effect size/power? yes

Length of follow-up:four weeks

Dropouts: Seven (13%) dropped out from the education group; eight (14%) dropped out from the exercise group. No reasons were provided


ParticipantsInclusion criteria

  • Knee OA according to the current ACR criteria (1999)
  • A score of 1, 2 or 3 on the Kellgren radiographic scale
  • At least one osteophyte other than at the tibial spine (with severity below Kellgren grade 4)
  • Between 50 and 80 years of age
  • No drugs for their arthritis other than stable doses of analgesics or NSAIDs
  • Mild to moderate knee pain for at least the previous three months


Exclusion criteria

  • Concurrently receiving physical therapy
  • Actively involved in any other pharmaceutical or exercise study
  • Had undergone isokinetic strength training within the previous three years
  • Had significant cardiovascular disease
  • More than mild knee swelling
  • Large popliteal cysts
  • Knee instability
  • Major knee or hip surgery on the side to be treated
  • Systemic disease other than OA that might affect muscle function
  • Severe osteopenia
  • A history of fracture in the area of the joint to be treated
  • Paresis of the lower extremity


Baseline characteristics

Baseline characteristics were similar among all treatment groups

Intervention group: education group (N = 56 randomly assigned, 49 analysed)

Location of OA: 100% knee

Body weight (mean (SD)): 190.4 (35.0) lb

Duration of OA (mean (SD)): 13.1 (11.7) years

PROGRESS-Plus

Place of residence: US

Sex: 64% male, 36% female

Age (mean (SD)), years: 64.5 (8.4)

Control group: exercise group (N = 57 randomly assigned, 49 analysed)

Location of OA: 100% knee

Body weight (mean (SD)):183.8 (32.8) lb

Duration of OA (mean (SD)): 9.7 (9.0)

PROGRESS-Plus

Place of residence: US

Sex: 53% male, 47% female

Age (mean (SD)), years: 66.3 (8.8)


InterventionsIntervention: education group

Description: All participants were provided with educational information about OA. Several Arthritis Foundation pamphlets were distributed, and four lectures and discussions were conducted. The programme consisted of (1) a lecture by a rheumatologist on the disease process of OA and its clinical characteristics, (2) a video discussing joint protection and OA self-management techniques, (3) a session on nutrition guidelines and a guide to relevant community services (provided by a dietician and a social worker, respectively) and (4) a discussion on coping with pain and disability led by a psychologist

Intended audience: people with knee OA

Mode: group sessions

Personnel: healthcare professionals

Delivery method: face-to-face

Language: English

Format: standard format

Location: -

Duration: four sessions in eight weeks

Additional treatment during trial: unclear

Were the following heiQ components addressed?

Health-directed behaviour: yes

Positive and active engagement in life: no

Emotional well-being: no

Self-monitoring and insight: yes

Constructive attitudes and approaches: no

Skill and technique acquisition: yes

Social integration and support: no

Health service navigation: no

Comparator: exercise group

Type: alternate intervention

Description: Participants underwent strength training of the knee extensor muscle groups unilaterally with the dynamometer three times a week for eight weeks. During the exercise sessions, a total of 27 repetitions were performed as three sets of three extensions at each of the following angular velocities: 90 gr/s, 120 gr/s and 150 gr/s. Between each set of three repetitions, the velocity was not adjusted while the participant rested for one minute

Frequency: three times a week for eight weeks

Additional treatment during trial: unclear


OutcomesOutcome assessed at: baseline, post-treatment and four weeks of follow-up

Primary outcomes of study

  • AIMS-2 subscales (including mobility, walk and bend, extremity function, household and self-care, bodily pain, social and psychological factors)
  • SF-36 MOS total score and subscales (including pain, physical functioning, role-physical, role-emotional, general health, vitality, social functioning, mental health)
  • Global OA score (WOMAC VAS, 0 to 240, lower score is better)
  • Pain (WOMAC VAS, 0 to 50, lower score is better)
  • Function self-reported (WOMAC VAS, 0 to 170, lower score is better)
  • Pain during 50 feet walking with moderate pace (10-point scale, 0 to 10, lower score is better)
  • Pain during 50 feet walking with maximum pace (10-point scale, 0 to 10, lower score is better)
  • Pain during ascending and descending half a flight of stairs (10-point scale, 0 to 10, lower score is better)
  • How much does your knee limit your ability to perform your daily activities (open-ended question)
  • Treatments before and at time of investigation
  • Clinical findings: disease duration, pattern of involved joints, duration of morning stiffness (as assessed by physician)
  • Knee range of motion (goniometer)
  • Heel-to-buttocks distance at maximal knee flexion
  • Joint effusion (palpation, yes or no)
  • Isokinetic peak torque at 90 degrees/s and 120 degrees/s (dynamometer)
  • Isometric strength (dynamometer)
  • Participant evaluation based on the extent to which the participant's pain limited his or her activity level
  • Physician evaluation based on the extent to which the participant's pain limited his or her activity level (according to the physician)


NotesWe extracted the following outcomes at post-treatment (short term) and at six months (intermediate term) for the analyses in this review: global OA scores (WOMAC), pain (WOMAC subscale pain), function self-reported (WOMAC subscale function) and dropouts (proportion of missing participants)

This trial was supported by a Health Services Research and Development grant from the Department of Veterans Affairs

The author (R Schumacher) provided additional information about the trial on request

Data analysis: As this study provided no information on variance, we used the SD from WOMAC total score, WOMAC subscale pain and WOMAC subscale function from an observational trial that was judged to have a population similar enough to the population of this study (Wolfe 1999)


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low riskQuote: “Patients were assigned by a random number generator to either the exercise group (…) or to the education group (…)”

Comment: This method of randomisation has a low risk of selection bias

Allocation concealment (selection bias)Unclear riskComment: No information provided on allocation concealment

Blinding of participants and personnel (performance bias)
All outcomes
High riskComment: In this type of intervention and control, participants and personnel cannot be blinded from treatment allocation, possibly introducing a risk of bias

Blinding of outcome assessment (detection bias)
All outcomes
High riskComment: Self-reported outcomes were assessed with the use of validated questionnaires. Other outcomes were assessed by blinded personnel; however, most outcomes are subjective, and participants are not blinded to group allocation

Incomplete outcome data (attrition bias)
All outcomes
Low riskQuote: “Of the 113 who enrolled, 15 either dropped out or were excluded (…) In none of these cases termination believed to be caused by a study-related event; (…)”

Comment: The numbers of dropouts are equal in the two groups, and the rate of dropout is not too high (13% and 14%). Reasons for dropout remain unclear. No information is provided on the type of analysis performed

Selective reporting (reporting bias)Low riskAll important outcomes for this review were reported

Other biasLow riskComment: No other potential sources of bias were identified

Mazzuca 1997

MethodsStudy design:RCT, single-centre, two arms, non-blinded

Country in which trial was carried out:US

Method of recruitment of participants:Participants were identified by a computerised medical record system containing all outpatient information for the host facility

Setting:primary care

Was the sample size justified with a priori calculation of effect size/power?no justification provided

Length of follow-up: 12 months

Dropouts:19 (18%) dropped out of the intervention group, and 21 (20%) dropped out of the control group. Reasons for dropout were moving without forwarding address (18), inconvenience (15), unrelated illness (five) and death (two)


ParticipantsInclusion criteria

  • Radiographically confirmed OA of the knee (knee radiograph reported as showing “OA” or “osteophyte”)
  • With a clinic visit in the past year
  • A pharmacy record indicating a current prescription of a salicylate or other NSAID or a pure analgesic (acetaminophen or opioid analgesic)
  • Participants needed to be accessible by telephone
  • Participants needed to score seven (of a possible 10) on the Mini-Mental State Examination


Exclusion criteria

  • Concurrent systemic inflammatory rheumatic disease
  • Medical comorbidity that would render the participant unable to participate fully in study procedures (e.g. terminal conditions such as chronic obstructive pulmonary disease, end-stage renal disease, heart failure, or malignancy with anticipated life expectancy less than one year)
  • Alcohol abuse or a psychiatric disorder
  • Previous or planned knee arthroplasty
  • Treatment by a rheumatologist in the past year


Baseline characteristics

Baseline characteristics were similar among all treatment groups

Intervention group: self-care education (N = 105, 82 analysed)

Location of OA: 100% knee

Duration of OA (mean (SD)): 14.6 (17.9) years

PROGRESS-Plus

Place of residence: inner city, US

Race, ethnicity and culture: 71% African American

Occupation: 50% unemployed, 19% employed, 30% retired

Sex: 15% male, 85% female

Education (mean (SD)): 9.7 (2.6) years of education

Socioeconomic status: 97% annual income ≤ $20,000

Social capital: 75% living alone

Age (mean (SD)), years: 62.8 (12.2)

Disability (mean (SD)): 1.5 (1.0) comorbid conditions

Control group: attention control (N = 106 randomly assigned, 83 analysed)

Location of OA: 100% knee

Duration of OA (mean (SD)): 13.5 (13.7) years

PROGRESS-Plus

Place of residence: inner city, US

Race, ethnicity and culture: 67% African American

Occupation: 56% unemployed, 11% employed, 33% retired

Sex: 15% male, 85% female

Education (mean (SD)): 9.7 (3.5) years

Socioeconomic status: 96% annual income ≤ $20.000

Social capital: 72% living alone

Age (mean (SD)), years: 62.0 (11.0)

Disability (mean (SD)): 1.7 (1.1) comorbid conditions


InterventionsIntervention: self-care education

Description: Participants received a course of individualised arthritis self-care instructions based on needs demonstrated in the diagnostic assessment and through preliminary communication with the primary care physician. Core content areas included quadriceps strengthening exercises, control of joint pain with thermal modalities and joint protection. Participants for whom NSAIDs or salicylates were prescribed were instructed on the importance of taking their medication with food or milk. Toward the end of the teaching session, all participants were encouraged to identify the vocational and avocational roles that were most threatened by their knee OA. A brief problem-solving exercise then occurred. At the end of the session, each participant was given a copy of the Arthritis Foundation Information series pamphlet, “Osteoarthritis”, and a printed set of instructions for isometric quadriceps exercises. Participants were contacted by phone at one week and one month after initial instruction. The phone contacts were unscripted but were structured to ensure that (1) compliance with self-care recommendations was assessed and reinforced, as appropriate, (2) misconceptions were clarified and (3) continued participation in the study was encouraged

Intended audience: people with knee OA

Mode: individual

Personnel: experienced arthritis nurse educator under the supervision of a rheumatologist

Delivery method: face-to-face and by telephone

Language: English

Format: tailored to individual needs

Location: general practitioner clinic

Duration: one session of 30 to 60 minutes and thereafter a telephone call (five to 10 minutes) at one week and at one month after initial instruction

Additional treatment during trial: unclear

Were the following heiQ components addressed?

Health-directed behaviour: yes

Positive and active engagement in life: no

Emotional well-being: no

Self-monitoring and insight: no

Constructive attitudes and approaches: no

Skill and technique acquisition: yes

Social integration and support: no

Health service navigation: no

Comparator: attention control

Type: attention control

Description: All participants viewed the synchronised slide-tape presentation “Arthritis, it’s not just growing old”. This 20-minute audiovisual presentation was designed for the general public, with the primary purposes of defining common types of arthritis in adults and encouraging those who think they may have arthritis to seek medical care. Participants also were given a current issue of the "IUMAMDC Newsletter". A brief follow-up telephone contact was made one week and one month after attention-placebo treatment but only for the purpose of reinforcing continued participation in the study

Frequency: once at 20 minutes, followed by two phone calls, each of five to 10 minutes

Additional treatment during trial: unclear


OutcomesOutcome assessed at: baseline and four, eight and 12 months of follow-up

Primary outcomes of study

  • Function self-reported (HAQ disability, 0 to 3, lower score is better)
  • Pain (Health Assessment Questionnaire (HAQ) subscale pain on VAS, 0 to 10, lower score is better)
  • Pain walking and at rest (VAS, 0 to 10, lower score is better)
  • Quality of life (Quality of Well-Being Scale (QWB), 0 to 1, lower score is better)


Secondary outcomes of study

  • Healthcare utilisation and costs


NotesWe extracted the following outcomes at 12 months (intermediate term) for the analyses in this review: pain (HAQ subscale pain), function self-reported (HAQ subscale disability), quality of life (QWB) and dropouts (proportion of missing participants)

The trial was supported in part by a grant from NIAMS (AR-20582)


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)High riskQuote: “If contact occurred during a time when the nurse educator was available, the subject was enrolled in the E-group and received self-care education. Subjects recruited at other times were assigned to the AC condition”

Comment: This method of random sequence generation has a high risk of bias

Allocation concealment (selection bias)Unclear riskComment: no information provided on allocation concealment

Blinding of participants and personnel (performance bias)
All outcomes
High riskComment: Participants were not blinded, which may have introduced bias. Although personnel also were not blinded, risk of bias remains low because participants in the control group seem to have no contact with the study personnel

Blinding of outcome assessment (detection bias)
All outcomes
High riskComment: Outcomes were assessed through self-report; therefore outcome assessment was not blinded

Incomplete outcome data (attrition bias)
All outcomes
High riskQuote: “(…) omission from the analyses of any subjects who did not complete all assessments (i.e., those who were lost to followup or who missed interim assessments).”

Quote: “Specifically, the 23 subjects in group AC with incomplete followup data tended to have higher baseline HAQ disability scores (i.e., they had poorer function) than the 23 dropouts from group E”

Comment: No intention-to-treat analysis was performed. Although numbers of dropouts were equal in the two groups (18% vs 20%), differences in characteristics of dropouts were evident between these groups, which may have biased results

Selective reporting (reporting bias)Low riskComment: Outcomes listed in the Methods were reported in the Results

Other biasLow riskComment: No other potential sources of bias were identified

Mazzuca 2004

MethodsStudy design:cluster-RCT, multi-centre, two arms, outcome assessment blinded

Country in which trial was carried out: US

Method of recruitment of participants:screening of the computer records of a large HMO

Setting:primary care

Was the sample size justified with a priori calculation of effect size/power?no justification provided

Length of follow-up:12 months

Dropouts:31 (28%) dropped out from the intervention group, and 21 (28%) dropped out from the control group. No reasons provided


ParticipantsInclusion criteria

  • A clinical diagnosis of knee OA in the participant's medical record that satisfied the ACR clinical criteria for knee OA
  • Under treatment for chronic knee pain by primary care physician


Exclusion criteria:none

Baseline characteristics

Women constituted a marginally larger proportion of participants in group E than in group C (78% vs 65%)

Intervention group: nurse-directed intervention (N = 111 randomly assigned, 80 analysed)

Location of OA: 100% knee

PROGRESS-Plus

Place of residence: US

Race, ethnicity and culture: 72% white

Sex: 22% male (n = 24), 78% female (n = 87)

Education: 86% 12 or more years

Social capital: 64% married

Age (mean (SD)), years: 61.8 (12.5)

Control group: waiting list (N = 75 randomly assigned, 54 analysed)

Location of OA: 100% knee

PROGRESS-Plus

Place of residence: US

Race, ethnicity and culture: 69% white

Sex: 35% male (n = 26), 65% female (n = 49)

Education: 89% 12 or more years

Social capital: 60% married

Age (mean (SD)), years: 61.8 (11.9)


InterventionsIntervention: nurse-directed intervention

Description: Care for knee OA in group E was guided by an algorithm designed to achieve the following two goals: (1) first-line management with non-pharmacological treatment modalities and acetaminophen; and (2) pharmacological management of knee pain according to a stepped protocol designed to minimise the risk of NSAID therapy. The algorithm prescribed a set of non-pharmacological measures, including isometric quadriceps exercises, application of heat or cold, instruction in behavioural principles of joint protection (including weight loss), use of shoes with well-cushioned soles and a cane or other assistive device for ambulation (if necessary). Participants were advised to supplement these measures with acetaminophen (1 g up to three to four times daily) for relief of knee pain

After 30 to 60 minutes of self-care instructions, participants were given ≥ two weeks to adopt these measures and evaluate their effects, at which point they were contacted by telephone by the nurse to assess their progress. Participants who reported a satisfactory response had their self-care efforts reinforced; continued progress was followed biweekly for the remainder of the study period

The algorithm for participants with newly diagnosed knee OA (18 weeks) differed from the algorithm for participants who already took an NSAID (10 weeks)

Intended audience: people with OA of the knee

Mode: individual

Personnel: arthritis nurse educator, with concurrence of the primary care physician

Delivery method: face-to-face and telephone

Language: English

Format: standard format

Location: GP clinic and at home

Duration: one session of 30 to 60 minutes; after two weeks biweekly telephone calls (five to 10 minutes) for 10 or 18 weeks

Additional treatment during trial: unclear

Were the following heiQ components addressed?

Health-directed behaviour: yes

Positive and active engagement in life: no

Emotional well-being: no

Self-monitoring and insight: no

Constructive attitudes and approaches: no

Skill and technique acquisition: yes

Social integration and support: no

Health service navigation: no

Comparator

Type: waiting list (usual care)

Description: Participants in the control group (delayed intervention) received routine OA care. In addition, their primary care physicians received the care algorithm as part of the HMO’s in-service education programme. No resource or consultation routinely available to participants with OA within the HMO and deemed appropriate by the primary care physician was denied to participants in the control group

Additional treatment during trial: unclear


OutcomesOutcome assessed at: baseline and three, six and 12 months of follow-up

Primary outcomes of study

  • Pain (WOMAC subscale pain, 5 to 25, lower score is better)
  • Function self-reported (WOMAC subscale function, 17 to 85, lower score is better)


NotesWe extracted the following outcomes at 12 months (intermediate term) for the analyses in this review: pain (WOMAC subscale pain), function self-reported (WOMAC subscale function) and dropouts (proportion of missing participants)

This study was supported in part by a grant from the National Institutes of Health (P60 AR20582)

Data analysis: For all outcomes, standard errors were converted into SD


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Unclear riskQuote: “One hundred eleven patients were seen at HMO sites randomized to group E; 75 received care at sites randomize to group C”

Comment: no information provided on the method of random sequence generation

Allocation concealment (selection bias)Unclear riskComment: no information provided on the concealment of allocation

Blinding of participants and personnel (performance bias)
All outcomes
High riskComment: Participants could not be blinded, with the possible risk of performance bias. The primary physicians also were not blinded to treatment allocation, thereby introducing a risk of performance bias

Blinding of outcome assessment (detection bias)
All outcomes
High riskQuote: “Study personnel who were blind to group assignment conducted the interviews”

Comment: Outcome assessors were blinded to group assignment; however, most outcomes are subjective, and participants are not blinded to group allocation

Incomplete outcome data (attrition bias)
All outcomes
High riskQuote: “Subjects who discontinued the interview schedule prematurely did not differ from those completing the study with respect to sex, age, ethnicity, education level, marital status or WOMAC scores”

Comment: Although dropouts did not differ from completers, the number of participants dropping out is high (28% in both groups), and it is unclear whether an intention-to-treat analysis is performed

Selective reporting (reporting bias)Low riskComment: Outcomes listed in the Methods section were reported in the Results

Other biasHigh riskQuote: “In addition, their PCP received the care algorithm as part of the HMO’s in-service education program”

Comment: The primary care physicians who provided the usual care to the control group participants also received the intervention algorithm, possibly introducing contamination

Quote: "Because the HMO site was the unit of randomization and the 2 nurses worked at different care sites, patient and site were included in the models as random effects"

Comment: The cluster design was accounted for in the statistical analysis

McKnight 2010

MethodsStudy design:RCT (stratified by sex), single-centre, three arms, non-blinded.

Country in which trial was carried out: US

Method of recruitment of participants: recruitment from the local community by mass mailings, television/newspaper advertisements and flyers

Setting: general population

Was the sample size justified with a priori calculation of effect size/power? yes

Length of follow-up: 24 months

Dropouts:27 (30%) dropped out from the strength training group (21 discontinued because of no interest, personal reasons, knee replacement, time commitment or other reasons; six were lost-to-follow-up), 20 (23%) dropped out from the self-management group (10 discontinued because of no compliance, time commitment, inflammatory arthritis; 10 were lost-to-follow-up), 25 (26%) dropped out from the combination group (20 discontinued because of health problems, no interest, time commitment, no compliance and other reasons; five were lost-to-follow-up)


ParticipantsInclusion criteria

  • Kellgren/Lawrence classification grade 2 radiographic evidence of knee OA in one or both knees
  • Age 35 to 64 years
  • History of pain on most days (i.e. four or more days in a week) in one or both knees for at least four months during the year before study entry
  • Duration of symptoms (defined as pain on most days for at least four months in one year) of less than five years
  • Some level of disability due to knee pain for at least three of the following items: descending or ascending stairs, walking, kneeling or performing daily activities


Exclusion criteria

  • Any uncontrolled medical condition that could prevent safe participation in the study (e.g. uncontrolled heart disease, blood pressure or respiratory condition)
  • Any neurological condition that could affect coordination
  • Inflammatory arthritis (e.g. rheumatoid or psoriatic arthritis)
  • Participates in vigorous (e.g. exercise, walking, household chores) physical activity for longer than 120 minutes per week
  • Participates in any form of resistance training
  • History of knee surgery
  • Kellgren and Lawrence grade 3 or 4 radiographic evidence of OA in one or both knees
  • Body mass index greater than 37.5 kg/m2
  • History of a knee corticosteroid injection in the three months before study entry
  • Plans to move from the local area
  • Plans to become pregnant during the study


Baseline characteristics

Baseline characteristics were similar in all treatment groups

Intervention group: self-management group (N = 87 randomly assigned, 87 analysed)

Location of OA: 100% knee

BMI (mean (SD)): 27.9 (4.1) kg/m2

PROGRESS-Plus

Place of residence: US

Race, ethnicity and culture: 96.3% white

Sex: 26.3% male, 74.7% female

Education: 55.9% college educated

Age (mean (SD)), years: 52.6 (6.5)

Intervention group: combination group (N = 95 randomly assigned, 95 analysed)

Location of OA: 100% knee

BMI (mean (SD)): 27.4 (4.1) kg/m2

PROGRESS-Plus

Place of residence: US

Race, ethnicity and culture: 86.3% white

Sex: 24.0% male, 76.0% female

Education: 59.1% college educated

Age (mean (SD)), years: 51.9 (7.7)

Control group: strength training (N = 91 randomly assigned, 91 analysed)

Location of OA: 100% knee

BMI (mean (SD)): 27.9 (4.5) kg/m2

PROGRESS-Plus

Place of residence: US

Race, ethnicity and culture: 92.6% white

Sex: 19.8% male, 80.2% female

Education: 74.1% college educated

Age (mean (SD)), years: 53.3 (7.2)


InterventionsIntervention: self-management

Description: The two-phase self-management intervention targeted coping and self-efficacy skills. Phase 1 (nine months) consisted of classroom sessions. These were followed by telephone calls designed to boost knowledge and behaviours from classroom sessions, as well as to provide practical, one-on-one problem-solving discussions to tailor the treatment to each participant’s needs. The telephone sessions continued into phase 2, when they were staggered. Coping skills focused on promoting increased adaptive strategies and reducing avoidant or passive strategies. Self-efficacy skills focused on increasing perceptions of control for physical functioning, pain management and other ancillary arthritis symptoms. Staff taught self-management skills using educational and behavioural methods, including homework assignments and active involvement/practice during treatment sessions.

Intended audience: People with knee OA

Mode: both group sessions and individual phone calls

Personnel: programme manager and local health professionals

Delivery method: face-to-face and telephone

Language: English

Format: tailored to individual needs

Location: ‘classroom’

Duration: Phase 1 consisted of nine months of 12 weekly sessions, each of 90 minutes (60% didactic, 40% interactive) and (once-)weekly telephone calls; phase 2 consisted of telephone calls staggering to biweekly, monthly and then bimonthly

Additional treatment during trial: unclear

Were the following heiQ components addressed?

Health-directed behaviour: yes

Positive and active engagement in life: no

Emotional well-being: no

Self-monitoring and insight: yes

Constructive attitudes and approaches: yes

Skill and technique acquisition: yes

Social integration and support: no

Health service navigation: no

Intervention: combination

Description: The combined group concurrently participated in both strength training and self-management courses, with slight alterations to ensure equivalence of contact time across the treatment groups. Specifically, the staff contacted participants in the combined treatment group less often than participants in the strength training and self-management programmes during phase 2. Otherwise, the combined group participated in the full, independent treatment protocols for both strength training and self-management programmes

Intended audience: people with knee OA

Mode: both group sessions and individual phone calls

Personnel: programme manager, local health professionals and physiotherapists

Delivery method: face-to-face and telephone

Language: English

Format: tailored to individual needs and standard format

Location: ‘classroom’

Duration: Phase 1 consisted of nine months of 12 weekly sessions, each of 90 minutes (60% didactic, 40% interactive) and (once-)weekly telephone calls for the SMP-programme and three sessions per week, each of 60 minutes, for strength training; phase 2 consisted of telephone calls staggering to biweekly, monthly and then bimonthly for the SMP-programme and strength training sessions every two weeks in the first six weeks and then monthly for a total of 15 months

Additional treatment during trial: unclear

Were the following heiQ components addressed?

Health-directed behaviour: yes

Positive and active engagement in life: no

Emotional well-being: no

Self-monitoring and insight: yes

Constructive attitudes and approaches: yes

Skill and technique acquisition: yes

Social integration and support: no

Health service navigation: no

Comparator: strength training

Type: alternate intervention

Description: Strength training consisted of two phases. Phase 1 (nine months) consisted of sessions supervised by expert physical trainers, targeted to improve stretching and balance, range of motion and flexibility, as well as isotonic muscle strengthening. Phase 2 (15 months) focused on developing self-directed long-term exercising habits

Frequency: three sessions per week for nine months (phase 1), then every two weeks in first six weeks and then monthly for a total of 15 months (phase 2). Each session took 60 minutes

Additional treatment during trial: unclear


OutcomesOutcome assessed at: baseline and three, nine, 18 and 24 months of follow-up

Primary outcomes of study

  • Functional performance (maximum voluntary isometric lower body strength, range of motion (FOCUS), work capacity (ERGOS work simulator), timed get-up-and-go test, stair climbing)
  • Pain (10-cm VAS, 0 to 100, lower score is better)
  • Pain (SF-36 subscale pain and WOMAC subscale pain)
  • Function self-reported (WOMAC subscales stiffness and disability)
  • Function self-reported (SF-36 subscale function)
  • Function self-reported (VAS for disability, 0 to 100, lower score is better)


Secondary outcomes of study

  • Coping efficacy, self-management and health-related quality of life (Client Satisfaction Questionnaire (CSQ))
  • Self-management (Arthritis Self-Efficacy Scale (ASES))
  • Emotional distress (Positive and Negative Affect Schedule (PANAS))
  • Quality of life (EuroQol)
  • Social integration and support (Medical Outcomes Social Support Survey)


NotesOnly outcomes on dropouts could be extracted because presentation of data was insufficient

Funding provided by the National Institute of Arthritis and Musculoskeletal and Skin Diseases (Grant R01-AR-047595)


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low riskQuote: “Two-hundred seventy-three participants were stratified by sex and randomly assigned by the project coordinator via a random number table to 1 of the 3 treatment groups”

Comment: This method of randomisation has low risk of introducing bias

Allocation concealment (selection bias)Unclear riskComment: no information provided on allocation concealment

Blinding of participants and personnel (performance bias)
All outcomes
High riskComment: Participants and personnel were not blinded to treatment allocation, which may have introduced bias

Blinding of outcome assessment (detection bias)
All outcomes
High riskQuote: “The Knee Study was a 24-month unblinded RCT”

Comment: The authors specifically report that this was an unblinded study, which means that outcome assessors were not blinded, and risk of detection bias is present

Incomplete outcome data (attrition bias)
All outcomes
Low riskQuote: “Missing data were handled with a multiple-imputation procedure, imputing 5 complete data sets to provide complete data for our intent-to-treat analysis“

Comment: An ITT analysis was performed with multiple imputation of missing data. Loss to follow-up was similar among treatment groups, with reasons provided and similar, although the dropout rate was fairly high (30%, 23% and 26%)

Selective reporting (reporting bias)High riskComment: Only outcomes on dropouts could be extracted because data presentation was insufficient

Other biasHigh riskQuote: “Overall compliance was higher during phase 1 (67.5%) compared with phase 2 (50.3%), with negligible differences between the groups”

Comment: Overall and group-specific adherence to the allocated treatment was low

Murphy 2008

MethodsStudy design: RCT, multi-centre, two arms, outcome assessment blinded

Country in which trial was carried out: US

Method of recruitment of participants: Participants were recruited through fliers and on-site presentations

Setting: senior housing facilities/senior centres

Was the sample size justified with a priori calculation of effect size/power?no justification for sample size provided

Length of follow-up:six months

Dropouts:Four (14%) dropped out of intervention group (one because of medical problems, three before intervention with no baseline data), and one (4%) dropped out of control group (medical problems)


ParticipantsInclusion criteria

  • Symptomatic hip or knee OA as determined by ACR clinical criteria through an on-site examination by one of the study rheumatologists
  • Reporting of OA symptoms (i.e. pain, stiffness, fatigue) that caused difficulty or the need for personal assistance in at least one of four activities of daily living (i.e. bathing, transferring, toileting, walking)
  • Age 62 or older
  • Ambulating with or without walking aid
  • English-speaking
  • No significant cognitive impairment (score of 5 or greater on the six-item screener)


Exclusion criteria

  • Hip or knee surgery within the previous nine months
  • A condition in which exercise would be contraindicated (e.g. uncontrolled hypertension, recent surgery, severe pain during exercise)
  • Current participation in physical/occupational therapy
  • Dementia
  • Unable to give consent


Baseline characteristics

Baseline characteristics were similar among all treatment groups

Intervention group: exercise + activity strategy training (N = 28 randomly assigned, 25 analysed)

Location of OA: 67% knee, 22% hip and knee, 11% hip [total study population]

BMI (mean (SD)): 30.1 (6.5) kg/m2

PROGRESS-Plus

Place of residence: US

Race, ethnicity and culture: 93% white

Sex: 7% male, 93% female

Education: 75% had some college to advanced degree

Social capital: 25% married

Age (mean (SD)), years: 75.8 (7.1)

Disability (mean (SD)): 1.5 (1.4) total chronic conditions, 4.4 (2.1) painful or stiff joints

Control group: exercise + education (N = 26 randomly assigned, 26 analysed)

Location of OA: 67% knee, 22% hip and knee, 11% hip [total study population]

BMI (mean (SD)): 30.0 (4.8) kg/m2

PROGRESS-Plus

Place of residence: US

Race, ethnicity and culture: 89% white

Sex: 15% male, 85% female

Education: 58% had some college to advanced degree

Social capital: 19% married

Age (mean (SD)), years: 74.8 (7.3)

Disability (mean (SD)): 1.0 (1.2) total chronic conditions, 4.6 (2.1) painful or stiff joints


InterventionsIntervention: exercise + activity strategy training

Description: The AST sessions involved education, group discussion and demonstration and practice of techniques to facilitate activity performance. Participants practiced strategies for symptom management. Physical activity enhancement was encouraged by addressing individual barriers and group problem solving to build in additional physical and other valued activities into daily routines. Once an occupational therapist went to the resident of the participant to guide individualised instruction on in-home strategies

Exercise was equal in the two groups (control and intervention) and consisted of progressive resistance exercises using ankle cuff weights by which extra weight could be added. The programme was tailored to individual participants as needed. The programme took 45 minutes to perform, including warm-up and cool-down

During the next six months, participants met for two additional group sessions (spaced two months apart) to review the exercise programme and highlight main points from the AST components

Intended audience: people with knee and hip OA

Mode: group sessions (and one individual session)

Personnel: occupational therapists

Delivery method: face-to-face

Language: English

Format: tailored to individual needs

Location: at the housing site of the participants (senior homes)

Duration: two sessions per week, 1.5-hour sessions, for four weeks

Additional treatment during trial: unclear

Were the following heiQ components addressed?

Health-directed behaviour: yes

Positive and active engagement in life: no

Emotional well-being: no

Self-monitoring and insight: no

Constructive attitudes and approaches: yes

Skill and technique acquisition: yes

Social integration and support: no

Health service navigation: no

Comparator

Type: alternate intervention

Description: Exercise plus education group; exercise was the same as in the intervention group, the health education programme was based on educational materials from the Arthritis Foundation on topics such as managing pain, importance of exercise, diet, arthritis and medication options

During the next six months, participants met for two additional group sessions (spaced two months apart) to review the exercise programme and highlight main points from the education components

Frequency: two sessions of 1.5 hours per week for four weeks

Additional treatment during trial: unclear


OutcomesOutcome assessed at: baseline, post-treatment and after six months of follow-up

Primary outcomes of study

  • Pain (WOMAC subscale pain, 0 to 20, lower score is better)
  • Functional performance (physical activity (CHAMPS, Actiwatch-S))


Secondary outcomes of study

  • Self-management (Arthritis Self-Efficacy Scale (ASES) subscales pain and other symptoms)
  • Functional performance (six-minute walk test, timed up-and-go test)


NotesWe extracted the following outcomes at post-treatment (short term) for the analyses in this review: self-management (ASES subscales pain and other symptoms), pain (WOMAC subscale pain), functional performance (six-minute walking distance) and dropouts (proportion of missing participants)

The trial was supported by the National Centre for Medical Rehabilitation Research (Grant K01-HD-045293) and the Office of the Vice President for Research at the University of Michigan

The author (S Murphy) provided additional information on the trial on request


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low riskQuote: “Participants were randomly assigned after baseline assessment at each site into 1 of the 2 interventions in blocks of 2 using a random digit table”

Comment: This method of randomisation has a low risk of introducing bias

Allocation concealment (selection bias)Unclear riskComment: no information given on allocation concealment

Blinding of participants and personnel (performance bias)
All outcomes
High riskComment: Blinding of participants and personnel to treatment allocation is not possible in this type of control and intervention, thereby introducing a risk of performance bias

Blinding of outcome assessment (detection bias)
All outcomes
High riskQuote: “All testing was done at each site by trained assessors blinded to the group assignment of participants”

Comment: Outcome assessors were blinded; however, most outcomes are subjective, and participants are not blinded to group allocation

Incomplete outcome data (attrition bias)
All outcomes
Low riskQuote: “An intent-to-treat analysis was performed using the last observation carried forward (in this case, baseline values when post-test values were missing)”

Comment: Intention-to-treat approach was used in analyses. Dropout rate was low both groups (4/28 in intervention group, 1/26 in control group) for similar reasons

Selective reporting (reporting bias)Unclear riskQuote: "Participants then returned for follow up testing 6 months after baseline. Only baseline to posttest outcomes are presented here"

Comment: Six-month outcomes were not reported

Other biasLow riskComment: No other potential sources of bias were identified

Nunez 2006

MethodsStudy design:RCT, single-centre, two arms, outcome assessment blinded

Country in which trial was carried out:Spain

Method of recruitment of participants:Participants were referred by the orthopaedic surgery department (100 consecutive patients meeting the inclusion criteria were enrolled)

Setting: outpatients

Was the sample size justified with a priori calculation of effect size/power? yes

Length of follow-up:six months

Dropouts: Eight (16%) dropped out of the intervention group (two died, two had severe pathology, one lost contact, three dropped out), 12 (24%) dropped out of the control group (two died, one had severe pathology, two lost contact, two transferred to other communities, five dropped out)


ParticipantsInclusion criteria

  • Knee OA according to the Kellgren and Lawrence criteria
  • On waiting list for total knee replacement less than six months


Exclusion criteria

  • Functional illiteracy
  • Inflammatory musculoskeletal disease
  • Metabolic or neoplastic disease
  • Severe psychopathology or comorbidity, defined as a diagnosis in the medical record severe enough that the participant could not complete the TEFR programme


Baseline characteristics

Baseline characteristics were similar among all treatment groups

Intervention group: TEFR (N = 51 randomly assigned, 43 analysed)

Location of OA: 100% knee

Duration of OA (mean (SD)): 12.06 (9.65) months

PROGRESS-Plus

Place of residence: Spain, urban

Occupation: 88% retired or housewives, 10% permanently disabled, 2% active (sick leave)

Sex: 24% male, 76% female

Social capital: 63% had a family, 35% was alone, 2% lived with a carer

Age (mean (SD)), years: 72.59 (6.20)

Disability: 86% had comorbidities, 37% had prior prostheses

Control group (N = 49 randomly assigned, 37 analysed)

Location of OA: 100% knee

Duration of OA (mean (SD)): 11.61 (11.45) months

PROGRESS-Plus

Place of residence: Spain, urban

Occupation: 80% retired or housewives, 16% permanently disabled, 4% active (sick leave)

Sex: 25% male, 65% female

Social capital: 72% had a family, 24% were alone, 4% lived with a carer

Age (mean (SD)), years: 69.45 (6.79)

Disability: 82% had a comorbidity, 29% had a prior prosthesis


InterventionsIntervention

Description: The TEFR programme was based on theories of social learning and self-management and was carried out using active learning strategies. The programme consisted of two individual visits at first week and at three months, and group group sessions in weeks three and four. In group sessions, when possible, a relative or a significant other accompanied the participants. All participants were provided with written information on the contents of the sessions. The contents were centred on consequences of the disease in daily life and included principles of economy/energy conservation and joint protection; evaluation and control of pain; treatment recommended for the management of knee OA; demonstration and use of assistive devices and tables of physical exercises with no burden on the lower limbs; and general exercises to mobilise the joints and strengthen the musculature of the rest of the body. Participants were instructed to practise the exercises at home during the week before the second group session

Intended audience: people with knee OA

Mode: group sessions (10 to 12 participants) and individual sessions (both two sessions)

Personnel: trained health educator

Delivery method: face-to-face

Language: Spanish

Format: both tailored to individual needs and standard format

Location: -

Duration: Total duration of the programme was three months; at week one and month three, individual visits lasted 30 minutes, at weeks three and four, group visits lasted 90 minutes

Additional treatment during trial: Both groups of participants received 3 to 4 g/d of paracetamol alone or ≤ 2 g/d of paracetamol combined with 2,400 mg/d of ibuprofen or NSAIDs. The dosage of NSAIDs used was varied according to individual participant needs

Were the following heiQ components addressed?

Health-directed behaviour: yes

Positive and active engagement in life: no

Emotional well-being: no

Self-monitoring and insight: no

Constructive attitudes and approaches: no

Skill and technique acquisition: yes

Social integration and support: no

Health service navigation: no

Comparator

Type: usual care

Description: Participants received conventional (pharmacological) treatment only. Participants received 3 to 4 g/d of paracetamol alone or ≤ 2 g/d of paracetamol combined with 2,400 mg/d of ibuprofen or NSAIDs. The dosage of NSAIDs used was varied according to individual participant needs

Additional treatment during trial: none


OutcomesOutcome assessed at: baseline and after six months of follow-up

Primary outcomes of study

  • Pain (WOMAC subscale pain, 0 to 20, lower score is better)
  • Function self-reported (WOMAC subscale function, 0 to 68, lower score is better)
  • Stiffness (WOMAC subscale stiffness, 0 to 8, lower score is better)


Secondary outcomes of study

  • Quality of life (SF-36 subscales, 0 to 100, higher score is better)
  • Medical treatment (dose of analgesics and NSAIDs per week, number of visits to general physicians, cost of visits to general physicians)


NotesWe extracted the following outcomes at six months (intermediate term) for the analyses in this review: positive and active engagement in life (SF-36 subscale role emotional), pain (WOMAC subscale pain), global OA scores (WOMAC), function self-reported (WOMAC subscale function), quality of life (SF-36 subscale general health perception), emotional distress (SF-36 subscale mental health), social integration and support (SF-36 subscale social function) and dropouts (proportion of missing participants)

Data analysis: For the outcome global OA scores, we added the subscales of WOMAC (pain, stiffness and physical function) provided by the author to get the WOMAC total score; we estimated the SD with the formula provided in Table 7.7a in the Cochrane Handbook for Systematic Reviews of Interventions and the SDs from the subscales (this method was chosen in close consultation with a biostatistician)


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low riskQuote: “A randomization table generated by an ad hoc program based on the pseudorandomized routine of the STATA 5.0 statistical package was used”

Comment: appropriate method of randomisation, with low risk of selection bias

Allocation concealment (selection bias)Unclear riskComment: no information provided on allocation concealment

Blinding of participants and personnel (performance bias)
All outcomes
High riskComment: The types of interventions and controls make it impossible for authors to blind participants to their allocated treatment, which might have introduced performance bias Personnel also were not blinded during the trial, but it is unlikely that physicians providing usual care would have been influenced by the fact that participants were randomly assigned to a control group

Blinding of outcome assessment (detection bias)
All outcomes
High riskQuote: “All assessments were performed by an independent, blinded investigator”

Comment: Outcome assessors were blinded; however, most outcomes are subjective, and participants are not blinded to group allocation

Incomplete outcome data (attrition bias)
All outcomes
High riskQuote: “The crude logistics regression analysis showed that there was no evidence that the loss of patients was related to any of the baseline characteristics or to either of the groups (…)”

Comment: The rate of dropout is moderately high but unequal between groups (16% in the intervention group, 24% in the control group). However, analysis showed that dropout rates were not dependent on baseline characteristics or group allocation. No intention-to-treat analysis was performed

Selective reporting (reporting bias)Low riskComment: All outcomes listed in the Methods were reported in the Results

Other biasLow riskComment: No other potential sources of bias were identified

Victor 2005

Methods


ParticipantsInclusion criteria

  • A diagnosis of OA made clinically and confirmed with X-ray
  • Aged 45 years or older
  • Knee pain due to OA


Exclusion criteria

  • No informed consent given
  • Unable to speak or understand English
  • Psychiatric illness
  • Other diseases (e.g. cardiac)


Baseline characteristics

In the control group, more participants lived alone, more participants came from a non-white race and the levels of disability in this group were significantly lower than in the intervention group

Intervention group: self-management group (N = 120, 87 analysed after one month of FU, 72 analysed after 12 months of FU)

Location of OA: 100% knee

Duration of OA: 55% had OA for longer than three years

PROGRESS-Plus

Place of residence: UK

Race, ethnicity and culture: 48% non-white ethnic group

Occupation: 32% employed, 34% professional or managerial job

Sex: 25% male, 75% female

Education: 40% higher education

Socioeconomic status: 25% living alone, 70% home owners

Social capital: 57% married

Age (mean (SD)), years: 62 (11)

Disability: 57% OA in both knees, 61% OA in other joints, 24% limiting long-term illness

Control group: waiting list (N = 73 randomly assigned, 56 analysed after one month of FU, 53 analysed after 12 months of FU)

Location of OA: 100% knee

Duration of OA: 55% OA for longer than three years

PROGRESS-Plus

Place of residence: UK

Race, ethnicity and culture: 80% non-white ethnic group

Occupation: 28% employed, 33% professional or managerial job

Sex: 31% male, 69% female

Education: 34% higher education

Socioeconomic status: 46% living alone, 62% home owners

Social capital: 45% married

Age (mean (SD)), years: 65 (11)

Disability: 63% OA in both knees, 67% OA in other joints, 25% limiting long-term illness


InterventionsIntervention: self-management group

Description: Goals were to inform participants about OA, its causes and effects; to increase self-efficacy by developing strategies and skills in coping with pain, joint protection and exercise; and to improve self-esteem and quality of life by sharing experiences and group support. The structured programme covered clinical information, participatory activities to promote increased function and skills development in coping

Intended audience: people with OA of the knee

Mode: group sessions (six to eight participants)

Personnel: research nurses

Delivery method: face-to-face and written (booklet)

Language: English

Format: structured format

Location: space made available by the GP (e.g. waiting room, treatment or consulting room)

Duration: four sessions, each of one hour

Additional treatment during trial: usual care

Were the following heiQ components addressed?

Health-directed behaviour: yes

Positive and active engagement in life: no

Emotional well-being: no

Self-monitoring and insight: no

Constructive attitudes and approaches: yes

Skill and technique acquisition: yes

Social integration and support: yes

Health service navigation: no

Comparator: waiting list

Type: waiting list

Description: Participants from the control practices received only the booklet and were put on a waiting list for the PEP. After completion of the trial, they were offered the full intervention

Additional treatment during trial: usual care


Outcomes


Notes


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)High riskQuote: “The unit of minimisation was the practice rather than individual patients”

Comment: Trial used cluster-randomisation (per practice) instead of individual participant randomisation, which introduces a risk of bias

Allocation concealment (selection bias)Low riskQuote: “Minimisation status as either waiting list or intervention was concealed from practices until they had achieved their patient recruitment quota”

Comment: Allocation was concealed during participant recruitment

Blinding of participants and personnel (performance bias)
All outcomes
High riskComment: Participants and personnel were not blinded from treatment allocation during the intervention period, introducing a risk of bias

Blinding of outcome assessment (detection bias)
All outcomes
High riskQuote: “At baseline and 12 months direct interviews were undertaken by interviewers ‘blind’ to the status of participants”

Comment: Outcome assessors were blinded to treatment allocation; however, most outcomes were subjective, and participants were not blinded to group allocation

Incomplete outcome data (attrition bias)
All outcomes
High riskQuote: “Loss to follow-up was significantly greater amongst those with lower ‘coping’ scores as measured by the AHI, lower socio-domain SF-36 scores and lower physical health status”

Comment: Loss to follow-up is unequally distributed (48% in intervention group vs 27% in control group), large in both groups and significantly greater among persons with specific characteristics. No intention-to-treat analysis was performed but a per-protocol analysis was performed (only completers)

Selective reporting (reporting bias)Unclear riskComment: outcomes at three and six months not reported

Other biasLow riskQuote: "Thus, we analysed the data at the patient level, but using regression with adjustment for clustering. The Stata statistical software package (Stata Corp) includes a facility for such cluster-adjusted regression, in which the standard error estimates are based on robust estimates of variance"

Comment: The cluster design was accounted for in the statistical analysis

Wetzels 2005

MethodsStudy design: RCT (block-randomisation), multi-centre, two arms, outcome assessment blinded

Country in which trial was carried out: The Netherlands

Method of recruitment of participants: Participants were recruited through their GP from their practice medical records

Setting: primary care

Was the sample size justified with a priori calculation of effect size/power? yes

Length of follow-up:six months

Dropouts: Six dropped out from the intervention (12%), and one dropped out from the control group (2%). Reasons were motivation problems, moved elsewhere, hip/knee surgery, too severe problems of comorbidity and treatment by a geriatric specialist


ParticipantsInclusion criteria

  • A clinical diagnosis of hip or knee OA, registered in the participant's practice medical history record as free text or as ICPC-code L89 (OA of the knee) or L90 (OA of the hip)
  • Aged 65 years or older


Exclusion criteria

  • Had undergone hip or knee replacement operation
  • Had been referred for a replacement operation
  • GP thought patient was not suitable for participating (because of severe psychosocial circumstances or a terminal disease)


Baseline characteristics

Baseline characteristics were similar in all treatment groups

Intervention group: self-management programme (N = 51 randomly assigned, 40 analysed)

Location of OA: 52.9% knee, 17.6% hip, 29.4% both hip and knee

PROGRESS-Plus

Place of residence: urban, The Netherlands

Sex: 23.5% male, 76.5% female

Education: 54% primary or lower secondary, 46% upper secondary or further

Age (mean (SD)), years: 75.63 (6.68)

Control group (N = 53 randomly assigned, 48 analysed)

Location of OA: 54.7% knee, 22.6% hip, 22.6% both hip and knee

PROGRESS-Plus

Place of residence: urban, The Netherlands

Sex: 24.5% male, 75.5% female

Education: 50% primary or lower secondary, 50% upper secondary or further

Age (mean (SD)), years: 73.47 (6.01)


InterventionsIntervention: self-management programme

Description: The intervention included education and self-management of OA symptoms and consisted of three parts: (1) prepreparation for home visit of the nurse using a booklet and health status charts; (2) 30-minute home visit by a nurse to discuss prepared ‘homework’ and goal setting (changing behaviour); and (3) follow-up phone call after three months for evaluation and adaptation of goals

Intended audience: people with OA

Mode: individual

Personnel: family practice nurse (certified educated in rheumatology)

Delivery method: face-to-face, follow-up by telephone

Language: Dutch

Format: tailored to individual needs

Location: home

Duration: one 30-minute nurse home visit and one follow-up phone call after three months

Additional treatment during trial: unclear

Were the following heiQ components addressed?

Health-directed behaviour: yes

Positive and active engagement in life: yes

Emotional well-being: no

Self-monitoring and insight: yes

Constructive attitudes and approaches: yes

Skill and technique acquisition: no

Social integration and support: no

Health service navigation: no

Comparator

Type: information only

Description: Participants in the control group received only the educational leaflet about OA

Additional treatment during trial: unclear


OutcomesOutcome assessed at: baseline and six months of follow-up

Primary outcomes of study

  • Function self-reported (Dutch AIMS-SF subscale physical, 7 to 35, lower score is better)
  • Global OA scores (Dutch AIMS-SF subscale symptoms, 3 to 15, lower score is better)
  • Social integration and support (Dutch AIMS-SF subscale social, 4 to 20, lower score is better)
  • Emotional distress (Dutch AIMS-SF subscale affect, 5 to 25, lower score is better)
  • Functional performance (timed up-and-go test (TUG) below 12 seconds, percentage of persons (0 to 100), higher score is better)


Secondary outcomes of study

  • Participant-reported number of contacts with GP and physiotherapist
  • Use of pain medication


NotesWe extracted the following outcomes at six months (intermediate term) for the analyses in this review: function self-reported (Dutch AIMS-SF subscale function), emotional distress (Dutch AIMS-SF subscale affect), social integration and support (Dutch AIMS-SF subscale social) and dropouts (proportion of missing participants)

Funding was provided by The Netherlands Organisation for Health Research and Development (ZonMW, number 920-03-252)

Data analysis: Change scores were combined with end point scores using generic inverse variance. For the outcome function of performance, we converted dichotomous outcomes to continuous outcomes, so we could combine these data with other available continuous data for this outcome (Cochrane Handbook for Systematic Reviews of Interventions, Chapter 9.4.6).


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low riskQuote: “An independent statistician made randomization lists in advance for each practice. To ensure similar number of patient from different practices in each group, block-randomisation (blocks of two) was used“

Comment: Adequate randomisation technique was used

Allocation concealment (selection bias)Low riskQuote: “This procedure was performed by a research assistant who was blinded for patients’ characteristics“

Comment: Participant characteristics were concealed during treatment allocation, reducing the risk of selection bias

Blinding of participants and personnel (performance bias)
All outcomes
High riskComment: Participants were not blinded from their allocated treatment during trial, introducing a risk of performance bias

Comment: Personnel administering the intervention had contact only with the intervention group and not with the control group. Therefore, it is unlikely that this provided risk of bias, even though personnel were not blinded to treatment allocation

Blinding of outcome assessment (detection bias)
All outcomes
High riskQuote: “A research assistant measured in all patients the post-intervention TUG, at this stage he was blinded for intervention-control condition“

Comment: The outcome assessor was blinded; however, most outcomes are subjective, and participants are not blinded to group allocation

Incomplete outcome data (attrition bias)
All outcomes
Low riskQuote: “Due to several reasons seven patients withdrew (…) and nine patients did not respond to the final patient self-assessment questionnaire (Figure 1)”

Quote: “Data from dropouts and lost to follow-up cases was not available, therefore only cases with data from baseline and after 6 months were included“

Comment: Although a strict intention-to-treat analysis was not performed, it is unlikely that this has affected the results significantly, as not many participants dropped out or were lost to follow-up at six months. Although dropout rates were not equally distributed between the two groups (12% in intervention vs 2% in control group), reasons were provided, and so dropout does not seem to introduce a high risk of bias

Selective reporting (reporting bias)Low riskComment: All outcomes specified in the Methods section are reported

Other biasHigh riskQuote: “The validity and reliability of the TUG might be compromised by the fact that the test was performed at home, on different chairs, and by different observers. At baseline the assessors of TUG times were not blinded for the assignment of subjects to treatment group“

Comment: The risk that results from the timed up-and-go test are not valid or reliable is high

Yip 2007

MethodsStudy design: RCT, multi-centre, two arms, non-blinded

Country in which trial was carried out:China

Method of recruitment of participants: not specified

Setting:outpatients

Was the sample size justified with a priori calculation of effect size/power?yes

Length of follow-up: 12 months

Dropouts

  • After four months, 21 (24%) dropped out from the intervention group (10 being busy, three no interest, three walking problems, four cannot contact), 41 (44%) dropped out from the control group (two excluded, 19 being busy, eight no interest, six walking problems, five cannot contact, one passed away)
  • After one year, 16 (36%) dropped out from the intervention group, and 24 (52%) dropped out from the control group


Note: In the one-year follow-up paper, only participants who completed at least two of three follow-up assessments were included (N = 45 in intervention group; N = 50 in control group)


ParticipantsInclusion criteria

  • Diagnosis of OA of the knee confirmed by medical history and a physical examination (by a registered nurse or a physiotherapist) based on the clinical criteria of the ACR
  • Capable of completing the questionnaire verbally


Exclusion criteria

  • Bed bound, wheelchair bound or experienced loss of balance while standing
  • Knee replacement
  • Currently having active physiotherapy such as hydrotherapy or strengthening exercises
  • Currently receiving acupuncture treatments


Baseline characteristics

Baseline characteristics were similar among all treatment groups

Intervention group: ASMP + exercise (N = 88 randomly assigned, 88 analysed after four months, 40 analysed after one year of FU)

Location of OA: 100% knee

Duration of OA (mean (SD)): 8.31 (7.3) years

PROGRESS-Plus

Place of residence: Hong Kong, China

Race, ethnicity and culture: Asian

Occupation: 26.1% housewife, 9.1% professional and administration, 64.8% service provider and non-professional workers

Sex: 15.9% male, 84.1% female

Education: 87.5% Form 3 level or below, 12.5% above Form 3

Social capital: 69.3% married and living together, 30.7% single

Age (mean (SD)), years: 65.6 (9.7)

Disability: number of joints affected: 6.8% one joint, 46.6% two joints, 29.5% three joints, 17.1% four or more joints

Control group (N = 94 randomly assigned, 94 analysed after four months, 37 analysed after one year of FU)

Location of OA: 100% knee

Duration of OA (mean (SD)): 7.85 (6.3) years

PROGRESS-Plus

Place of residence: Hong Kong, China

Race, ethnicity and culture: Asian

Occupation: 26.6% housewife, 8.5% professional and administration, 64.9% service provider and non-professional worker

Sex: 16.0% male, 84.1% female

Education: 87.2% Form 3 level or below, 12.8% above Form 3

Social capital: 70.2% married and living together, 29.8% single

Age (mean (SD)), years: 64.02 (10.3)

Disability: number of joints affected: 14.9% one joint, 51.1% two joints, 18.1% three joints, 15.9% four or more joints


InterventionsIntervention: ASMP + exercise

Description: The programme was based on the ASMP by Lorig. The programme focused on teaching participants how to cope with and manage common knee OA consequences, such as arthritis pain, fatigue, daily activity limitations and stress. It was designed to give participants skills they could use to optimise their ability to manage their condition. An action plan using three types of exercise was promoted and reinforced weekly during the programme. These included stretching, walking and Tai Chi. A pedometer was given to act as a positive reinforcement in walking (not used as an outcome measure). Participants set their goal on exercise practice and received positive feedback from a nurse every week

Intended audience: people with knee OA

Mode: group sessions (10 to 15 participants)

Personnel: registered nurses, lay person tutor who suffered from knee OA

Delivery method: face-to-face

Language: Chinese

Format: standard format

Location: -

Duration: once-weekly sessions of two hours' length, six weeks' total duration

Additional treatment during trial: usual care (conventional orthopaedic treatment)

Were the following heiQ components addressed?

Health-directed behaviour: yes

Positive and active engagement in life: no

Emotional well-being: yes

Self-monitoring and insight: yes

Constructive attitudes and approaches: yes

Skill and technique acquisition: yes

Social integration and support: no

Health service navigation: yes

Comparator

Type: usual care

Description: routine orthopaedic treatment (treatment prescribed by orthopaedic doctors or outpatient clinic) with no other treatment

Additional treatment during trial: none


OutcomesOutcome assessed at: baseline, one week, four months and 12 months of follow-up

Primary outcomes of study

  • Pain (Visual Analogue Scale (VAS), 0 to 100, lower score is better)
  • Function self-reported (fatigue intensity on VAS, 0 to 100, lower score is better)
  • Health-directed activity (frequency, duration of light exercise, hours per week, higher score is better)
  • Function self-reported (modified Health Assessment Questionnaire (mHAQ), 0 to 100, lower score is better)
  • Functional performance (range of motion both knees (goniometer))
  • Functional performance (muscle strength of hamstrings/quadriceps)
  • Unplanned arthritis-related medical consultations


Secondary outcomes of study

  • Self-management (Arthritis Self-Efficacy Scale (ASES) subscales pain (5 to 50), other symptoms (6 to 60), higher score is better)
  • Use of self-management techniques (use of cold/hot compresses, share the load among various joints, use large joint to carry heavy load, avoid maintaining weight on the same joint for prolonged periods of time)
  • Pain at night (VAS, 0 to 100, lower score is better)
  • Pain during walking (VAS, 0 to 100, lower score is better)
  • Pain sitting to standing position (VAS, 0 to 100, lower score is better)
  • Global OA scores (self-rated health, 1 to 5, higher score is better)


NotesWe extracted the following outcomes at one week (short term) and 12 months (intermediate term) for the analyses in this review: self-management (ASES subscale pain), pain (VAS), global OA scores (self-rated health), function self-reported (mHAQ), health-directed activity (hours of light exercise/week) and dropouts (proportion of missing participants)

The trial was funded by partial support of the SN Departmental Research Committee and the Hong Kong Polytechnic University, School of Nursing

The author (J Sit) provided additional information about the trial on request

Data analysis: For the outcome self-management in OA, we could choose between ASES subscale pain and ASES subscale other and chose for ASES subscale pain, as we judged that pain was a more measurable aspect of self-management. The direction of benefit for global OA scores was reversed ( i.e. multiplied by -1) in the analysis


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low riskQuote: “(…) before being assigned to an intervention or control group by reference to a random number table”

Comment: Appropriate method of random sequence generation was used

Allocation concealment (selection bias)Unclear riskComment: no information given on allocation concealment

Blinding of participants and personnel (performance bias)
All outcomes
High riskComment: Blinding of participants is not possible, and therefore risk of bias is high. Blinding of personnel is not possible either. However, it is unlikely that the treating physician in the control group was influenced by the randomisation of participants

Blinding of outcome assessment (detection bias)
All outcomes
High riskQuote: “Both of them were trained (…) and were not involved in delivering the intervention”

Comment: Blinding of outcome assessors remains unclear; however, most outcomes are subjective, and participants are not blinded to group allocation, introducing a risk of detection bias

Incomplete outcome data (attrition bias)
All outcomes
High riskQuote: “The drop-out group suffered from higher arthritic pain than the participant group. During the previous 16 weeks the drop-out group had visited the doctor for unplanned arthritis-related problems more frequently”

Quote: “Analyses of intervention effects with and without an intent-to-treat basis were done and compared. (…) An intent-to-treat basis is presented”

Comment: Although the authors state that an ITT analysis is presented, one-year follow-up results are presented only for participants who completed two or more assessments. A high dropout rate was noted in both groups after four months (44% in the control group, 24% in the intervention group). The dropout rates differ between groups, although provided reasons are similar, and differences were found between completers and non-completers. In the one-year analysis, dropout is high as well and differs between groups (52% in the control group, 36% in the intervention group)

Selective reporting (reporting bias)Low riskComment: all outcomes reported

Other biasHigh riskComment: The cultural difference between the Western population and the Asian population has to be taken into account. This study was performed in China, and assessments were done using face-to-face interviews. It is possible that the outcome improvement was not related to the content of the programme but was simply a result of the participants' behaving according to their expectations of what the researchers were looking for (Hawthorne effect). Response bias might have occurred as a result of face-to-face interviewing and cultural traditions

 
Characteristics of excluded studies [ordered by study ID]

StudyReason for exclusion

Barlow 2000Mixed arthritis population without subgroup data for OA (we attempted to contact the authors but did not receive a response)

Bezalel 2010Intervention did not fulfil our criteria for a self-management programme

Coleman 2010Study compared two different self-management programmes so did not fulfil our inclusion criteria

Ehrlich-Jones 2001Mixed arthritis population without subgroup data for OA (we attempted to contact the authors but did not receive a response)

Ettinger 1997Intervention did not fulfil our criteria for a self-management programme

Fernandes 2009Intervention did not fulfil our criteria for a self-management programme

Fernandes 2010Intervention did not fulfil our criteria for a self-management programme

Focht 2005Intervention did not fulfil our criteria for a self-management programme

Goeppinger 1989Mixed arthritis population without subgroup data for OA

Hoogeboom 2010Study compared two different self-management programmes so did not fulfil our inclusion criteria

Laforest 2008Mixed arthritis population without subgroup data for OA (we attempted to contact the authors but did not receive a response)

Laforest 2008aMixed arthritis population without subgroup data for OA (we attempted to contact the authors but did not receive a response)

Lindroth 1989Mixed arthritis population without subgroup data for OA (we attempted to contact the authors but did not receive a response)

Lorig 1985Mixed arthritis population without subgroup data for OA

Lorig 1998Study compared two different self-management programmes so did not fulfil our inclusion criteria

Lorig 1999aMixed arthritis population without subgroup data for OA (we attempted to contact the authors but did not receive a response)

Lorig 1999bMixed arthritis population without subgroup data for OA

Lorig 2005Mixed arthritis population without subgroup data for OA

Martire 2003aStudy compared two different self-management programmes so did not fulfil our inclusion criteria

Martire 2008Study compared two different self-management programmes so did not fulfil our inclusion criteria

Murphy 2010Study compared two different self-management programmes so did not fulfil our inclusion criteria

Nour 2006Mixed arthritis population without subgroup data for OA (we attempted to contact the authors but did not receive a response)

Solomon 2002Mixed arthritis population without subgroup data for OA (we attempted to contact the authors but did not receive a response)

 
Characteristics of studies awaiting assessment [ordered by study ID]
Allen 2011

Methods

Participants

Interventions

Outcomes

NotesAwaiting assessment

Coleman 2012

Methods

Participants

Interventions

Outcomes

NotesAwaiting assessment

Hurley 2012

Methods

Participants

Interventions

Outcomes

NotesAwaiting assessment

Schlenk 2011

Methods

Participants

Interventions

Outcomes

NotesAwaiting assessment

Somers 2012

Methods

Participants

Interventions

Outcomes

NotesAwaiting assessment

Stukstette 2011

Methods

Participants

Interventions

Outcomes

NotesAwaiting assessment

Von Korff 2012

Methods

Participants

Interventions

Outcomes

NotesAwaiting assessment

Wu 2011

Methods

Participants

Interventions

Outcomes

NotesAwaiting assessment

 
Characteristics of ongoing studies [ordered by study ID]
Allen 2012

Trial name or title

Methods

Participants

Interventions

Outcomes

Starting date

Contact information

NotesAwaiting assessment

Bennell 2012

Trial name or title

Methods

Participants

Interventions

Outcomes

Starting date

Contact information

NotesAwaiting assessment

 
Comparison 1. SMP versus attention control

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Self-management of OA1Mean Difference (IV, Random, 95% CI)Totals not selected

 2 Pain3Std. Mean Difference (Random, 95% CI)Subtotals only

    2.1 Short term
1Std. Mean Difference (Random, 95% CI)-0.62 [-1.11, -0.13]

    2.2 Intermediate term
3Std. Mean Difference (Random, 95% CI)-0.26 [-0.44, -0.09]

 3 Global OA scores1Mean Difference (IV, Random, 95% CI)Totals not selected

 4 Function—self-reported3Std. Mean Difference (Random, 95% CI)Subtotals only

    4.1 Short term
1Std. Mean Difference (Random, 95% CI)-0.13 [-0.49, 0.23]

    4.2 Intermediate term
3Std. Mean Difference (Random, 95% CI)-0.19 [-0.50, 0.11]

 5 Quality of life1Mean Difference (IV, Random, 95% CI)Totals not selected

 6 Withdrawals5937Risk Ratio (M-H, Random, 95% CI)1.11 [0.78, 1.57]

 7 Emotional distress2Std. Mean Difference (IV, Random, 95% CI)Subtotals only

    7.1 Short term
168Std. Mean Difference (IV, Random, 95% CI)-0.37 [-0.85, 0.11]

    7.2 Intermediate term
2409Std. Mean Difference (IV, Random, 95% CI)0.02 [-0.18, 0.21]

 
Comparison 2. SMP versus usual care/no treatment/wait list

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Self-management of OA11Std. Mean Difference (Random, 95% CI)Subtotals only

    1.1 Short term
5721Std. Mean Difference (Random, 95% CI)0.22 [-0.00, 0.45]

    1.2 Intermediate term
101647Std. Mean Difference (Random, 95% CI)0.14 [0.00, 0.27]

    1.3 Long term
1195Std. Mean Difference (Random, 95% CI)0.23 [-0.05, 0.51]

 2 Engagement in life3Std. Mean Difference (IV, Random, 95% CI)Subtotals only

    2.1 Short term
1143Std. Mean Difference (IV, Random, 95% CI)-0.23 [-0.57, 0.11]

    2.2 Intermediate term
3357Std. Mean Difference (IV, Random, 95% CI)0.01 [-0.20, 0.21]

 3 Pain14Std. Mean Difference (Random, 95% CI)Subtotals only

    3.1 Short term
6Std. Mean Difference (Random, 95% CI)-0.26 [-0.41, -0.10]

    3.2 Intermediate term
13Std. Mean Difference (Random, 95% CI)-0.17 [-0.26, -0.08]

    3.3 Long term
1Std. Mean Difference (Random, 95% CI)-0.18 [-0.45, 0.09]

 4 Global OA scores7Std. Mean Difference (Random, 95% CI)Subtotals only

    4.1 Short term
2Std. Mean Difference (Random, 95% CI)-0.34 [-0.59, -0.09]

    4.2 Intermediate term
7Std. Mean Difference (Random, 95% CI)-0.28 [-0.39, -0.17]

    4.3 Long term
1Std. Mean Difference (Random, 95% CI)-0.29 [-0.56, -0.02]

 5 Function—self-reported13Std. Mean Difference (Random, 95% CI)Subtotals only

    5.1 Short term
5Std. Mean Difference (Random, 95% CI)-0.01 [-0.19, 0.18]

    5.2 Intermediate term
13Std. Mean Difference (Random, 95% CI)-0.16 [-0.25, -0.08]

    5.3 Long term
1Std. Mean Difference (Random, 95% CI)-0.27 [-0.55, 0.01]

 6 Function—performance2Std. Mean Difference (Random, 95% CI)Subtotals only

    6.1 Short term
1Std. Mean Difference (Random, 95% CI)0.33 [-0.07, 0.73]

    6.2 Intermediate term
2Std. Mean Difference (Random, 95% CI)0.06 [-0.24, 0.36]

 7 Quality of life8Std. Mean Difference (Random, 95% CI)Subtotals only

    7.1 Short term
2Std. Mean Difference (Random, 95% CI)0.14 [-0.47, 0.75]

    7.2 Intermediate term
8Std. Mean Difference (Random, 95% CI)0.03 [-0.08, 0.14]

    7.3 Long term
2Std. Mean Difference (Random, 95% CI)0.10 [-0.10, 0.31]

 8 Withdrawals163738Risk Ratio (M-H, Random, 95% CI)0.99 [0.74, 1.33]

 9 Emotional distress9Std. Mean Difference (Random, 95% CI)Subtotals only

    9.1 Short term
3Std. Mean Difference (Random, 95% CI)0.01 [-0.44, 0.45]

    9.2 Intermediate term
8Std. Mean Difference (Random, 95% CI)0.11 [-0.06, 0.28]

 10 Health-directed activity3Std. Mean Difference (Random, 95% CI)Subtotals only

    10.1 Short term
1Std. Mean Difference (Random, 95% CI)0.67 [0.37, 0.97]

    10.2 Intermediate term
3Std. Mean Difference (Random, 95% CI)0.25 [0.05, 0.46]

 11 Skill and technique acquisition1Mean Difference (IV, Random, 95% CI)Totals not selected

 12 Constructive attitudes and approaches2Std. Mean Difference (IV, Random, 95% CI)Totals not selected

    12.1 Short term
1Std. Mean Difference (IV, Random, 95% CI)0.0 [0.0, 0.0]

    12.2 Intermediate term
1Std. Mean Difference (IV, Random, 95% CI)0.0 [0.0, 0.0]

 13 Social integration and support3Std. Mean Difference (IV, Random, 95% CI)Subtotals only

    13.1 Short term
1143Std. Mean Difference (IV, Random, 95% CI)-0.19 [-0.52, 0.15]

    13.2 Intermediate term
3357Std. Mean Difference (IV, Random, 95% CI)-0.08 [-0.30, 0.14]

 14 Health service navigation2Std. Mean Difference (Random, 95% CI)0.15 [-0.03, 0.34]

 
Comparison 3. SMP versus information only

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Self-management of OA3Std. Mean Difference (IV, Random, 95% CI)Subtotals only

    1.1 Short term
190Std. Mean Difference (IV, Random, 95% CI)0.06 [-0.36, 0.47]

    1.2 Intermediate term
3760Std. Mean Difference (IV, Random, 95% CI)0.20 [-0.04, 0.44]

 2 Engagement in life1Mean Difference (IV, Random, 95% CI)Totals not selected

    2.1 Short term
1Mean Difference (IV, Random, 95% CI)0.0 [0.0, 0.0]

    2.2 Intermediate term
1Mean Difference (IV, Random, 95% CI)0.0 [0.0, 0.0]

 3 Pain3751Std. Mean Difference (IV, Random, 95% CI)-0.07 [-0.21, 0.08]

 4 Global OA scores3Std. Mean Difference (IV, Random, 95% CI)Subtotals only

    4.1 Short term
189Std. Mean Difference (IV, Random, 95% CI)0.09 [-0.33, 0.50]

    4.2 Intermediate term
3751Std. Mean Difference (IV, Random, 95% CI)-0.06 [-0.28, 0.16]

 5 Function—self-reported4854Std. Mean Difference (IV, Random, 95% CI)-0.09 [-0.22, 0.05]

 6 Function—performance1Mean Difference (IV, Random, 95% CI)Totals not selected

 7 Quality of life2648Std. Mean Difference (IV, Random, 95% CI)0.05 [-0.10, 0.21]

 8 Withdrawals41251Risk Ratio (M-H, Random, 95% CI)1.60 [0.75, 3.40]

 9 Emotional distress3775Std. Mean Difference (IV, Random, 95% CI)-0.00 [-0.30, 0.30]

 10 Health-directed activity1Mean Difference (IV, Random, 95% CI)Totals not selected

    10.1 Short term
1Mean Difference (IV, Random, 95% CI)0.0 [0.0, 0.0]

    10.2 Intermediate term
1Mean Difference (IV, Random, 95% CI)0.0 [0.0, 0.0]

 11 Social integration and support2Std. Mean Difference (IV, Random, 95% CI)Subtotals only

    11.1 Short term
190Std. Mean Difference (IV, Random, 95% CI)-0.02 [-0.44, 0.40]

    11.2 Intermediate term
2181Std. Mean Difference (IV, Random, 95% CI)-0.02 [-0.39, 0.35]

 12 Health service navigation1Mean Difference (IV, Random, 95% CI)Totals not selected

    12.1 Short term
1Mean Difference (IV, Random, 95% CI)0.0 [0.0, 0.0]

    12.2 Intermediate term
1Mean Difference (IV, Random, 95% CI)0.0 [0.0, 0.0]

 13 Skill and technique acquisition1Mean Difference (IV, Random, 95% CI)Totals not selected

    13.1 Short term
1Mean Difference (IV, Random, 95% CI)0.0 [0.0, 0.0]

    13.2 Intermediate term
1Mean Difference (IV, Random, 95% CI)0.0 [0.0, 0.0]

 14 Constructive attitudes and approaches1Mean Difference (IV, Random, 95% CI)Totals not selected

    14.1 Short term
1Mean Difference (IV, Random, 95% CI)0.0 [0.0, 0.0]

    14.2 Intermediate term
1Mean Difference (IV, Random, 95% CI)0.0 [0.0, 0.0]

 
Comparison 4. SMP versus non-SMP intervention

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Self-management of OA3Std. Mean Difference (IV, Random, 95% CI)Subtotals only

    1.1 Short term
3186Std. Mean Difference (IV, Random, 95% CI)0.42 [-0.05, 0.89]

    1.2 Intermediate term
170Std. Mean Difference (IV, Random, 95% CI)0.54 [0.05, 1.04]

 2 Pain5Std. Mean Difference (Random, 95% CI)Subtotals only

    2.1 Short term
5Std. Mean Difference (Random, 95% CI)0.03 [-0.29, 0.36]

    2.2 Intermediate term
2Std. Mean Difference (Random, 95% CI)-0.18 [-0.56, 0.19]

 3 Global OA scores1Std. Mean Difference (Random, 95% CI)Totals not selected

    3.1 Short term
1Std. Mean Difference (Random, 95% CI)0.0 [0.0, 0.0]

 4 Function—self-reported3Std. Mean Difference (Random, 95% CI)Subtotals only

    4.1 Short term
3Std. Mean Difference (Random, 95% CI)0.23 [-0.03, 0.48]

    4.2 Intermediate term
2Std. Mean Difference (Random, 95% CI)-0.17 [-0.54, 0.20]

 5 Function—performance1Std. Mean Difference (Random, 95% CI)Totals not selected

 6 Quality of life2Std. Mean Difference (IV, Random, 95% CI)Subtotals only

    6.1 Short term
157Std. Mean Difference (IV, Random, 95% CI)0.24 [-0.28, 0.76]

    6.2 Intermediate term
2243Std. Mean Difference (IV, Random, 95% CI)-0.01 [-0.28, 0.26]

    6.3 Long term
1178Std. Mean Difference (IV, Random, 95% CI)-0.23 [-0.55, 0.10]

 7 Withdrawals7919Risk Ratio (M-H, Random, 95% CI)0.86 [0.69, 1.09]

 8 Emotional distress3Std. Mean Difference (IV, Random, 95% CI)Subtotals only

    8.1 Short term
3192Std. Mean Difference (IV, Random, 95% CI)0.13 [-0.28, 0.55]

    8.2 Intermediate term
2118Std. Mean Difference (IV, Random, 95% CI)0.18 [-0.19, 0.55]

 9 Constructive attitudes and approaches2Std. Mean Difference (IV, Random, 95% CI)Subtotals only

    9.1 Short term
2135Std. Mean Difference (IV, Random, 95% CI)0.92 [0.49, 1.34]

    9.2 Intermediate term
170Std. Mean Difference (IV, Random, 95% CI)0.62 [0.12, 1.12]

 
Comparison 5. SMP versus acupuncture

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Global OA scores1Std. Mean Difference (Random, 95% CI)Totals not selected

    1.1 Short term
1Std. Mean Difference (Random, 95% CI)0.0 [0.0, 0.0]

    1.2 Intermediate term
1Std. Mean Difference (Random, 95% CI)0.0 [0.0, 0.0]

 2 Pain1Std. Mean Difference (Random, 95% CI)Totals not selected

    2.1 Short term
1Std. Mean Difference (Random, 95% CI)0.0 [0.0, 0.0]

    2.2 Intermediate term
1Std. Mean Difference (Random, 95% CI)0.0 [0.0, 0.0]

 3 Function self-reported1Std. Mean Difference (Random, 95% CI)Totals not selected

    3.1 Short term
1Std. Mean Difference (Random, 95% CI)0.0 [0.0, 0.0]

    3.2 Intermediate term
1Std. Mean Difference (Random, 95% CI)0.0 [0.0, 0.0]

 4 Function performance1Std. Mean Difference (Random, 95% CI)Totals not selected

 5 Withdrawals1Risk Ratio (M-H, Random, 95% CI)Totals not selected

 
Comparison 6. Subgroup analysis

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Self-management in OA7Std. Mean Difference (Random, 95% CI)0.14 [-0.08, 0.36]

    1.1 White, educated, female
3Std. Mean Difference (Random, 95% CI)0.29 [0.07, 0.50]

    1.2 NOT white, educated, female
4Std. Mean Difference (Random, 95% CI)0.03 [-0.29, 0.36]

 2 Function self-reported8Std. Mean Difference (Random, 95% CI)-0.12 [-0.23, -0.01]

    2.1 White, educated, female
3Std. Mean Difference (Random, 95% CI)-0.20 [-0.37, -0.02]

    2.2 NOT white, educated, female
5Std. Mean Difference (Random, 95% CI)-0.06 [-0.21, 0.08]

 3 Pain9Std. Mean Difference (Random, 95% CI)-0.17 [-0.28, -0.06]

    3.1 White, educated, female
4Std. Mean Difference (Random, 95% CI)-0.11 [-0.30, 0.07]

    3.2 NOT white, educated, female
5Std. Mean Difference (Random, 95% CI)-0.20 [-0.35, -0.05]

 4 Withdrawals123095Risk Ratio (M-H, Random, 95% CI)1.04 [0.81, 1.34]

    4.1 White, educated, female
72129Risk Ratio (M-H, Random, 95% CI)1.05 [0.76, 1.45]

    4.2 NOT white, educated, female
5966Risk Ratio (M-H, Random, 95% CI)1.04 [0.63, 1.69]

 
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

OutcomesIllustrative comparative risks* (95% CI)Relative effect
(95% CI)
No of participants
(studies)
Quality of the evidence
(GRADE)
Comments

Assumed riskCorresponding risk

Attention controlSMP

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 measuredSee commentSee commentNot estimable-See commentNo 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 1,000130 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).
 
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

OutcomesIllustrative comparative risks* (95% CI)Relative effect
(95% CI)
No of participants
(studies)
Quality of the evidence
(GRADE)
Comments

Assumed riskCorresponding risk

Usual care/No treatment/Wait listSMP

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

FunctionSelf-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 1,000171 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.
 
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

OutcomesIllustrative comparative risks* (95% CI)Relative effect
(95% CI)
No of participants
(studies)
Quality of the evidence
(GRADE)
Comments

Assumed riskCorresponding risk

Information onlySMP

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 1,000389 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.
 
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

OutcomesIllustrative comparative risks* (95% CI)Relative effect
(95% CI)
No of participants
(studies)
Quality of the evidence
(GRADE)
Comments

Assumed riskCorresponding risk

Non-SMP interventionSMP

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 commentSee commentNot estimable0
(0)
See commentNo 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

FunctionSelf-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 1,000209 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.
 
Table 1. PROGRESS-Plus

StudyPROGrESSADHL

Allen 2010US54% white,

46% non-white,

2% Hispanic
38.3% employed92.7% male,

7.3% female
33.3% high school education or less25.7% inadequate income69.3% married60.1 (10.4)32.3% fair or poor health-

Berman 2004US69.1% 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 2004US72% 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/y56.5% married, 25% divorced,

13% single,

25% widowed
70.8 (6.1)16% joint replacement surgeryDomain social support for health

Buszewicz 2006UK> 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 accommodation68.6 (8.4)--

Calfas 1992US97.5% white,

2.5% non-white
82.5% retired, 5% full-time employed, 12.5% part-time employed27.5% male,

72.5% female
--70% married, 20% divorced, 10% widowed,

5% other
67.0 (7.3)--

Cronan 1997US92.3% white,

2.5% African American,

2.2% Hispanic,

1.7% Asian,

1.4% other
-36.3% male, 64.0% female31.2% high school, 37.7% some college, 28.7% college degree27.2% < $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 2009Australia-11.6% employed, 5.4% home duties, 82.7% retired39.5% 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 alone67.5 (10.8)32.9% on waiting list for hip replacementDomains social support for health, navigating health system, actively managing my health

Hansson 2010Sweden------62.5 (9.4)--

Heuts 2005The Netherlands-40.5% paying job, 35.5% no paying job40.5% male,

59.5% female
25.5% low,

30.5% middle, 20% high
--51.6 (5.1)--

Hopman-Rock 2000The Netherlands--17% male,

83% female
21.5% primary, 49.5% secondary, 23.5% college/university-67.5% living together/married, 27% living alone65.3 (5.5)2.5 (1.6) other chronic conditions-

Hughes 2004US72.2% white,

22.1% African American, 2.6% Hispanic,

2.1% Asian-Pacific Islander, 1.1% other
-16.8% male, 83.3% female10.5% less than high school, 20.0% high school, 69.6% more than high school33.1% income less than $20,000-73.422.4% ARA class I, 64.4% ARA class II, 13.3% ARA class III-

Hurley 2007UK--29.7% male, 70.3% female---67 (range 50 to 91)--

Jessep 2009UK--30.5% male, 69.5% female---66.5 (range 51 to 81)--

Keefe 1990US--28.3% male, 71.7% female---63.9 (11.5)--

Keefe 1996US--40% male,

60% female
---62.6--

Keefe 2004US--46.9% male, 53.4% female---59.5 (11.1)--

Lorig 2008US92.3% non-Hispanic white-9.9% male, 90.2% female15.7 (3.1) years of education-68.3% married52.4 (11.6)-Domain ability to actively engage with healthcare

providers

Maisiak 1996US85% white-7.7% male, 92.3% female12.1 years of schooling--60.448% see a specialist-

Martire 2007US--27.3% male, 72.7% female---68.6 (7.6)--

Maurer 1999US--58.5% male, 41.5% female---65.4 (8.6)--

Mazzuca 1997US69% African American53% unemployed, 15% employed, 31.5% retired15% male, 85% female9.7 (3.1) years of education96.5% annual income ≤ $20,00073.5% living alone62.4 (11.6)1.6 (1.1) comorbid conditions-

Mazzuca 2004US70.5% white-28.5% male, 71.5% female87.5% 12 or more years of education-62% married61.8 (12.2)--

McKnight 2010US91.7% white-23.4% male, 77.0% female63.0% college educated--52.6 (7.2)--

Murphy 2008US91% white-11% male, 89% female66.5% some college to advanced degree-22% married75.3 (7.1)4.5 (2.1) painful or stiff joints-

Nunez 2006Spain-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 carer71.1 (6.7)33% prior prostheses-

Ackerman 2012Australia68.5% Australian-born26% paid employment, 62% retired, 8.5% not working because of medical condition, 3.5% unemployed40% male, 60% female12% 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 partner65.1 (10.9)-Domains social support for health, navigating

health system, actively managing

my health

Victor 2005UK64% non-white ethnic group30% employed, 33.5% professional or managerial job28% male, 72% female37% higher education66% home owner35.5% living alone, 51% married63.1 (11.1)60% OA in both knees, 64% OA in other joints-

Wetzels 2005The Netherlands--24% male, 76% female52% primary or lower secondary, 48% upper secondary or further--74.5 (6.4)--

Yip 2007Hong Kong100% Asian26.4% housewife, 8.8% professional and administration, 64.9% service provider and non-professional16.0% male, 84.1% female87.4% Form 3 level or below, 12.7% above Form 3-69.8% married and living together, 30.3% single64.8 (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.
 
Table 2. Additional characteristics of included studies

StudyNoLocation of OABMIDuration of OA

Allen 201052380% knee, 15% hip, 5% knee and hip31.8 (6.7)16.1 (12.2) years

Berman 2004570100% knee-50.4% < 5 years,

20.7% 6 to 10 years,

28.7% > 10 years

Blixen 200432--8.3 (IQR 2 to 24) years

Buszewicz 2006812---

Calfas 199240--10% < 1 year,

5% 1 to 5 years,

92.5% > 5 years

Cronan 1997363--7.0 (5.5) years since diagnosis

Crotty 200915232.9% hipHeight 167.2 (10.1) cm

Weight 85.8 (21.0) kg
-

Hansson 20101144.5% hip, 34% knee, 32% hand, 29% more locations35% BMI 20 to 25,

38.5% BMI 25 to 30,

27% BMI > 30
-

Heuts 2005297-28.1 (5.0)-

Hopman-Rock 2000120-27.6 (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 2004215---

Hurley 2007431100% knee30.2 (range 18 to 51)6 (IQR 3 to 13) years

Jessep 200964100% knee29.5 (range 1 to 47)12.5 (range 0.5 to 55) years

Keefe 199099100% knee24.2 (23.6)% above ideal weight-

Keefe 199688100% knee--

Keefe 200472100% knee--

Lorig 2008551---

Maisiak 1996405--16.0 years

Martire 2007242Hip and knee-15.1 (10.9) years

Maurer 1999113100% kneeWeight 187.1 (33.9) lb11.4 (10.5) years

Mazzuca 1997211100% knee-14.0 (15.9) years

Mazzuca 2004186100% knee--

McKnight 2010273100% knee27.7 (4.2)-

Murphy 20085467% knee, 11% hip, 22% hip and knee30.1 (5.7)-

Nunez 2006100100% knee-11.9 (10.6) months

Ackerman 201212031% hip, 62.5% knee, 6.5% hip and knee24.5 (IQR 25 to 35)-

Victor 2005193100% knee-55% > 3 years

Wetzels 200510453.8% knee, 20.1% hip, 26% hip and knee--

Yip 2007182100% 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.
 
Table 3. Summary of comparisons

Study

Country
SettingIntervention (N)

Mode/Personnel/Delivery method/Duration
Attention control (N)

Mode/Personnel/Delivery method/Duration

Allen 2010

US
Primary careSelf-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
OutpatientsCognitive-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
OutpatientsPain 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 outpatientsTreatment 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 careEducation (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)

StudySettingIntervention (N)Information only (N)

Buszewicz 2006

UK
Primary careSelf-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 populationFit & 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
OutpatientsArthritis 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 careSelf-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: -

StudySettingIntervention (N)Usual care/Waiting list/No treatment (N)

Allen 2010

US
Primary careSelf-management intervention (174)

M: individual

P: health professionals

De: telephone

Du: 12 calls in 12 months
Usual care (174)

 

Blixen 2004

US
OutpatientsTelephone 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 populationEducation 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 populationCombination 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
OutpatientsPatient 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 carePEPOA (61)

M: group

P: health professionals

De: face-to-face

Du: five sessions in five weeks
Usual care (53)

Heuts 2005

The Netherlands
Primary careSelf-management programme (149)

M: group

P: trained physiotherapists

De: face-to-face

Du: six sessions
Usual care (148)

Hopman-Rock 2000

The Netherlands
General populationHealth 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 careIndividual rehabilitation (146)

M: individual

P: physiotherapist

De: face-to-face

Du: 12 sessions in six weeks
Usual care (140)

Hurley 2007

UK
Primary careGroup rehabilitation (132)

M: group

P: physiotherapist

De: face-to-face

Du: 12 sessions in six weeks
Usual care (140)

Keefe 1990

US
OutpatientsPain 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 populationSpouse-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 populationSpouse-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 populationInternet-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 outpatientsTreatment counselling (135)

M: individual

P: trained counsellors

De: telephone

Du: 11 sessions in nine months
Usual care (135)

Martire 2007

US
OutpatientsPatient-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
OutpatientsCouple-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 careNurse-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
OutpatientsSelf-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 careSelf-management (120)

M: group

P: research nurse

De: face-to-face

Du: four sessions
Waiting list (73)

Yip 2007

China
OutpatientsArthritis self-management programme (88)

M: group

P: nurses

De: face-to-face

Du: six sessions in six weeks
Usual care (94)

StudySettingIntervention (N)Alternate intervention (N)

Berman 2004

US
General populationEducation (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 populationEducation 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 populationCombination 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
OutpatientsESCAPE-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 specifiedCoping 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 specifiedSpouse-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 populationSpouse-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 populationSpouse-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
OutpatientsEducation (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 populationSelf-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 populationCombination 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 facilitiesExercise + 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.
 
Table 4. heiQ-components addressed in interventions

InterventionHealth-directed activityPositive and active engagement in lifeEmotional distressSelf-monitoring and insightConstructive attitudes and approachesSkill and technique acquisitionSocial integration and supportHealth 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