Methods

Study Design: Cluster‐RCT

Participants

Setting: Primary care

Country: Germany.

Participants: All 883 family physicians in 2 German regions were invited to participate

118 practices (126 GPs) agreed to participate and were randomised at practice level. 1 practice withdrew and 1 was excluded because it did not recruit any participants. Total participants recruited: 1378

Condition: Low back pain    

Inclusion criteria for patients were LBP as presenting symptom on the day of recruitment, written consent to participate in the study, and age above 19 years. Exclusion
criteria were insufficient German language skills, pregnancy, and isolated thoracic pain.

Interventions

Practices were randomised into 2 intervention and 1 "control" group.

1. Intervention: Distribution of guidelines on low back pain, 3 interactive seminars, 2 individual academic‐detailing sessions, patient leaflets (educational material + outreach visits +educational meetings)

2. Intervention: Distribution of guidelines on low back pain, 3 interactive seminars, 2 individual academic‐detailing sessions, patient leaflets. Also, motivational counselling session for GPs and 20‐hour training for 2 nurses per practice. Patients recruited received 3 counselling sessions by the nurses (patient‐directed component).

3."Control": Distribution of guidelines on low back pain (educational material)                          

Outcomes

GP outcomes: None

Patient outcomes: Functional capacity (measured by Hannover Functional Ability Questionnaire), days in pain, days of sick leave physical activity, quality of life (measured with EuroQol), and Fear Avoidance Beliefs questionnaire

Notes

We were unable to confirm the results and calculate the standardised mean differences (SMD) due to non‐reported standard deviations

Sources of funding: The study was funded by the German Ministry for Education and Research (BMBF, FKZ 01 EM 0113).

Conflicts of interest as declared by the authors: Federal funds were received in support of this work. No benefits in any form have been or will be received from a commercial party related
directly or indirectly to the subject of this manuscript.

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

“Practices were assigned to the 3 study arms by central permuted block randomisation with allocation concealment.”

Allocation concealment (selection bias)

Low risk

“Practices were assigned to the 3 study arms by central permuted block randomisation with allocation  concealment.”

Protection against contamination

Unclear risk

Allocation was by practice but it is unclear if communication between intervention and control practices could have occurred

Baseline outcomes similar

Low risk

No important differences present across study groups.

Baseline characteristics similar

Low risk

Baseline characteristics were reported and were similar between the 2 groups

Blinding of outcome assessment (detection bias)
All outcomes

Low risk

"Outcome measure was objective and recorded by interviewers and trained nurses."

Incomplete outcome data (attrition bias)
All outcomes

Low risk

All outcomes were reported

Selective reporting (reporting bias)

Unclear risk

No study protocol was published in order to be able to verify this.

Other bias

Unclear risk

Possibility of bias during participant recruitment but GPs were asked to recruit consecutive patients

Methods

Study design: RCT

Participants

Country: Canada (Quebec)

Condition: osteoporosis

Participants: 1314 women without osteoporosis treatment were randomised.

Inclusion Criteria:

Women, aged 50 years and over

Not residing in a long‐term care hospital before the fracture

Able to understand the programme information and consent form

Must voluntarily accept to participate in this programme and sign the consent form

Participants must have a fragility or traumatic fracture of one of the following sites: wrist, forearm, humerus, scapula, clavicle, sternum, thoracic or lumbar vertebrae, pelvis, sacrum, hip, femur, proximal and distal tibia, fibula (including ankle), and foot

Participants must be able to answer the questionnaires via phone interviews

Exclusion Criteria:

Unable to understand the purpose of the programme

Participants with a traumatic fracture of one of the following sites: cervical, skull and face, hand and finger, toe, metatarsus, and patella

Pathological fracture

Women currently participating in a clinical trial requiring them to take a medication for osteoporosis

Interventions

Experimental group 1: Written educational material on osteoporosis for the physician (distribution of educational material) plus education of patients with advice to see their GP and give them written material (patient‐directed component)

Experimental group 2: 15‐minute educational video on osteoporosis as well as written documentation on osteoporosis for the physician (distribution of educational material) plus education of participants with written material and video and advice to see their GP and give them written material (patient‐directed component)

Control group: No intervention. However the control group completed a questionnaire on osteoporosis which may have increased their awareness

Outcomes

Treatment for osteoporosis (using bisphonates, raloxifene, nasal calcitonin or teriparatide)

Notes

The analysis of the delivery of the reading material to physicians was completed as post hoc observation

Conflicts of interest and funding sources as declared by the authors: Conflicts of interest Dr. Bessette has received research grants from
Abbott, Amgen, Bristol‐Myers‐Squibb, Eli Lilly, Merck, Pfizer, and Roche, has received consulting fees or other remuneration from Abbott,
Amgen, Merck, Novartis, Pfizer, and Roche and has participated on the speakers bureau for Amgen, Novartis, Merck, Pfizer, Roche, andWarner
Chilcott. Dr. Brown has received research grants from Abbott, Amgen, Bristol‐Myers‐Squibb, Eli Lilly, Pfizer, and Roche, has received consulting fees
or other remuneration from Abbott, Amgen, Eli Lilly, Novartis, Merck, and Warner Chilcott and has participated on the speakers bureau for Eli
Lilly, Amgen, Novartis, Merck, and Warner Chilcott. Dr. Davison has received consulting fees or other remuneration
from Amgen and Servier and has participated on the Speakers’ Bureau for Amgen, Merck Frosst Warner Chilcott and Servier.
Dr. Ste‐Marie has received research grants from the Alliance for Better Bone Health and Novartis, has received consulting fees or other
remuneration fromtheAlliance for Better Bone Health,Amgen,Novartis, Eli Lilly, and Servier and has participated on the Speakers’ Bureau for the
Alliance for Better Bone Health, Amgen, Novartis, Eli Lilly, Servier, and Merck. No other authors have a conflict or interest to disclose
The ROCQ program was funded by Merck Frosst Canada, Inc., Warner Chilcott, Sanofi‐Aventis group, Amgen Canada Inc., Eli Lilly
Canada, Inc., and Novartis Pharmaceuticals Canada, Inc. None of the funding sources had a role in the collection, analysis, or interpretation
of the data or in the decision to publish this article.

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Unclear risk

The manner of randomisation has not been reported

Allocation concealment (selection bias)

Unclear risk

Not reported

Protection against contamination

Low risk

There were 4452 physicians available to treat 1174 included patients and we therefore felt the risk of contamination to be small

Baseline outcomes similar

Low risk

There were no statistically significant baseline differences among the groups for any investigated variable

Baseline characteristics similar

Low risk

The distribution of baseline characteristics was similar among the groups

Blinding of outcome assessment (detection bias)
All outcomes

Unclear risk

There was no reported blinding of outcome assessment

Incomplete outcome data (attrition bias)
All outcomes

High risk

The main outcomes (treatment rates) were reported as percentages. The analysis of the delivery of the reading material to physicians was completed as post hoc observation

Selective reporting (reporting bias)

Low risk

The main outcomes were treatment rates. The protocol of the study was published.

Other bias

Unclear risk

The analysis of the delivery of the reading material to physicians was completed as post hoc observation

Methods

Study design: RCT

Participants

Setting: Primary care

Country: Canada

462 providers, 428 patients

Condition: Acute low back pain.

Inclusion criteria: The patients included in this study were all residents of British Columbia, Canada, aged between 19 and 65 years.
They had as their chief complaint, acute low back pain and an accepted claim with the Workers’ Compensation
Board of British Columbia relating to an injury that was thought to be causative. All patients included in the study
satisfied the Quebec Task Force Classification of Spinal Disorders criteria for categories 1 or 2 and had symptoms
for more than 2 weeks and less than 4 weeks.

Interventions

1.Distribution of educational materials to GP only + 3 reminders at 0 ‐ 4 weeks (via letters), 5 ‐ 12 weeks and after 12 weeks

2.Distribution of educational materials to participant and GP + 3 reminders (both to participant and GP) at 0 ‐ 4 weeks, 5 ‐ 12 weeks and after 12 weeks (distribution of educational material, reminders and patient‐directed component)

3. Control: no educational material, usual care 

Outcomes

GP outcomes: Concordance with specific clinical guideline‐derived history‐taking items, physical examination procedures and treatment recommendations

Patient outcome: None

Conflicts of interest and sources of funding as declared by the authors: FDA device/drug status: not applicable.
This research was supported by the Workers’ Compensation Board of British Columbia, Canada. No funds were received from a commercial
entity related to this manuscript.

Notes

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

“Random number generator” used                                

Allocation concealment (selection bias)

High risk

 In group 2 GPs received “ a letter from a study physician regarding a specific named patient.”                                                                                                          

Protection against contamination

High risk

Randomisation happened at participant level and it is not clear if the same physician was part of both the intervention and the control group

Also, there is a risk of contamination if communication occurred between physicians allocated in different groups who worked in the same practice

Baseline outcomes similar

Unclear risk

Not specified

Baseline characteristics similar

Unclear risk

Not specified

Blinding of outcome assessment (detection bias)
All outcomes

Unclear risk

Not reported

Incomplete outcome data (attrition bias)
All outcomes

Low risk

No missing outcome data

Selective reporting (reporting bias)

Unclear risk

No study protocol published and therefore this could not be verified

Other bias

Unclear risk

It is unclear if either participants or GPs were blinded (possibility of performance bias)

Methods

Study design: Randomised trial, no control

Participants

Setting: Primary care

Country: USA
149 GPs were sent letters, 258 patients were recruited, 200 patients were contacted by GPs
Condition: osteoporosis detected by heel ultrasound

Fifty‐nine men (mean age 62.8) and 199 women (mean age 58.7) were involved in the survey; thirty‐seven men and 163 women were able to be questioned.

Of Caucasian patients 169 of 223 were reached and of African American patients 30 of 35 were reached.

Interventions

1. Patient‐mediated: extended letter to physician about patient’s risk of osteoporosis after USS screening result including advice on management (educational material)
2. Patient‐mediated: short letter about patient’s risk of osteoporosis

Outcomes

Number of participants contacted by GPs following distribution of reminders

Ordering BMD scan within 6 months

Prescription of osteoporosis medication

Notes

There was no control group

Conflicts of interest and sources of funding as declared by the authors: Merck, Procter and Gamble, and Aventis pharmaceuticals

for unrestricted grants for purchase of supplies and to support student activities related to the health fairs.

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Unclear risk

“Randomly assigned” is the only information provided

Allocation concealment (selection bias)

Unclear risk

Not reported

Protection against contamination

Unclear risk

Randomisation happened at physician level but it is not clear if physicians within the same practice were allocated in different groups

Baseline outcomes similar

Unclear risk

Not specified

Baseline characteristics similar

Unclear risk

Not specified

Blinding of outcome assessment (detection bias)
All outcomes

Unclear risk

Not reported

Incomplete outcome data (attrition bias)
All outcomes

High risk

Outcomes for only 63% of participants followed up

Selective reporting (reporting bias)

Unclear risk

No published protocol and therefore unable to verify this

Other bias

Unclear risk

Potential bias when recruiting participants: 149 GPs were sent letters, 258 patients were recruited, 200 patients were contacted by GPs

Also, not clear if participants were blinded or if they were aware that they were taking part in the study (potential performance bias)

Methods

Study design: CBA

2 intervention and 2 control sites (2 divisions of general practice in Adelaide)

Participants

Setting: Primary care

Country: Australia

87 GPs were recruited in the intervention group. 90 in the control group. GPs were eligible to participate if they were members in one of these
Divisions and were working ≥ 0.5 full time equivalent (FTE). All GPs in the two Divisions who met the selection criteria were invited to participate. 

Condition: Shoulder pain

Interventions

1. Two sessions of academic detailing (outreach session) on shoulder assessment + educational material (DVD) + guideline + follow‐up session 3 months afterwards (distribution of educational material)

2. Control group: 90 randomly‐selected GPs who received no extra training

Outcomes

GP outcomes: Knowledge score before, immediately after and 3 months after academic detailing + requests for ultrasound and Xray imaging

Notes

Conflicts of interest and funding resources as declared by the authors: Funded by the Diagnostic Imaging Reform Implementation Package, Diagnostic
Imaging Section, the Australian Department of Health and Ageing.

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

High risk

Not a randomised study

Allocation concealment (selection bias)

High risk

Not done

Protection against contamination

Unclear risk

There is a possibility of contamination if communication occurred between physicians allocated in different groups but working in the same practice

Baseline outcomes similar

Unclear risk

Not specified

Baseline characteristics similar

Low risk

Baseline characteristics of the intervention and control providers are reported as similar

Blinding of outcome assessment (detection bias)
All outcomes

Unclear risk

Not reported

Incomplete outcome data (attrition bias)
All outcomes

Low risk

All outcomes reported

Selective reporting (reporting bias)

Unclear risk

No study protocol published and therefore unable to verify this

Other bias

High risk

Possible recruitment (self selection) bias of participants                                                               

Methods

Study design: RCT

Participants

Setting: Primary care

Country: France

180 GPs, randomisation at GP level.

842 patient participants were recruited by the GPs. Patients over 49 years of age could enter the study if they had radiographic confirmation of OA

of the knee or hip for at least 6 months, had pain intensity on motion > or equal to 40 mm on a 100 mm visual analogue scale (VAS) the day before inclusion

; and were suitable for treatment with acetaminophen.

Condition: Osteoarthritis pain management

Interventions

1. Course on chronic pain management (3 x 4‐hourly group sessions), 8 postal letters emphasising recommendations, patient leaflet with 5 statements about pain relief (educational meeting/workshop plus distribution of educational material)

2. Control, unrelated presentation received

Outcomes

Outcomes: Change in the intensity of pain on motion as measure on a 100 mm VAS + Lequesne index score + Womac scores + Global perception of change + Acetaminophen use

Notes

Conflicts of interest ad sources of funding as declared by the authors: Supported and sponsored by Sanofi‐Aventis OTC, Direction Medicale, Gentilly, France.

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

“Randomization stratified according to practice location and date of qualification."

Allocation concealment (selection bias)

High risk

Not done

Protection against contamination

Unclear risk

There is a possibility of contamination if communication occurred between GPs allocated to different groups but working in the same practice

Baseline outcomes similar

Low risk

No important differences present across study groups

Baseline characteristics similar

Low risk

Baseline characteristics are reported as similar

Blinding of outcome assessment (detection bias)
All outcomes

Unclear risk

Not reported

Incomplete outcome data (attrition bias)
All outcomes

Low risk

Missing data balanced in numbers across groups. Similar reasons for missing data across groups

Selective reporting (reporting bias)

Unclear risk

No evidence of published protocol

Other bias

Unclear risk

Possibility of bias during recruitment of patients by GPs. Unclear if participants were blinded

Methods

Study design: RCT

Participants

Country: Canada, Ontario

Participants: patients over the age of 55, able to give consent and identified to be at risk of future fracture

Patients were eligible for inclusion in the study if they were community‐dwelling, aged 55 years or older, able
to give informed consent, and were identified to be at risk for future fracture according to one of the following
criteria:
1. attended the hospital Fracture Clinic for a non‐pathological fracture of the vertebrae, hip or wrist or
had a BMD in the past year with a T‐score of ≤‐2.0

2. attended the hospital Emergency Department with a fall and found to be at high risk for falls as defined
by a Timed Up and Go [25] result of greater than 14 seconds; or,
3. were self‐referred or referred by a health care provider because of perceived high risk of fracture and
identified as a high risk for falls defined by a Timed Up and Go result of greater than 14 seconds.
Patients already receiving appropriate pharmacological therapy for osteoporosis as outlined in the Osteoporosis
Canada guidelines were excluded from the study.

Interventions

Intervention group: The results of the patient's recent BMD test (patient‐mediated) and patient‐specific advice on prescribing according to the Osteoporosis Canada guidelines (reminders and educational material) were sent to the participant's physician. The participant received personalised counselling on osteoporosis from a research nurse, a written summary of the proposed management plan and educational material (patient‐directed component)

Control group: usual care

Outcomes

The main outcome was prescribing of osteoporosis medication (alendronate, risedronate, raloxifene) 6 months after the intervention

Notes

The control participants received the intervention 6 months after randomisation (delayed protocol group)

Conflicts of interest and sources of funding as declared by the authors: Financial support for completion of the study was given by Merck Frosst
Canada Ltd., Sanofi‐Aventis Pharma Inc., Proctor & Gamble Pharmaceuticals Canada Inc., Eli Lilly Canada Inc., and the Greenshield Foundation.
Equipment (e.g. office space, computers, telephones) was contributed in kind by the Group Health Centre, Algoma Public Health, Sault Area Hospital,
Algoma Community Care Access Centre, and the Slips, Trips and Falls Committee of Sault Ste. Marie Safe Communities Partnership, all located in
Sault Ste. Marie, Ontario. The Ontario Ministry of Health and Long‐term Care provided additional support.
PMC is supported by the Algoma District Medical Group in Sault Ste. Marie, Ontario. SES was supported by a salary award from the Alberta Heritage
Foundation for Medical Research (Health Scholar) when this study was completed and holds a Canada Research Chair in Knowledge Translation
and Quality of Care. LRD is supported by a Canadian Institutes of Health Research Rx&D Health Research Foundation Research Career Award. SRM is
supported by salary awards from the Alberta Heritage Foundation for Medical Research (Health Scholar) and the Canadian Institutes for Health
Research (New Investigator).

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

A computer‐generated randomisation scheme was used

Allocation concealment (selection bias)

High risk

Participants and treating physicians were not blinded

Protection against contamination

Unclear risk

Not reported

Baseline outcomes similar

Low risk

No statistically significant baseline differences were detected among the groups

Baseline characteristics similar

Low risk

Baseline characteristics were similar among the groups

Blinding of outcome assessment (detection bias)
All outcomes

High risk

Outcome assessors could not be blinded; however the primary source of data was obtained from the Group Health Centre Electronic Medical record

Incomplete outcome data (attrition bias)
All outcomes

Low risk

Selective reporting (reporting bias)

Low risk

All outcomes were reported. The protocol of the study was published

Other bias

Low risk

No other bias identified

Methods

Study design: Cluster‐RCT

56 cluster practices were in the intervention group and 63 in the control group

Participants

Setting: Primary care

Country: Canada 

119 GP practices (these were randomised), 270 patients (the unit of analysis).

Condition: Osteoporosis

Participants patients inclusion criteria: Participants included postmenopausal women who had sustained a wrist fracture (confirmed by
x‐ray). Women currently taking osteoporosis therapies (e.g., risedronate, raloxifene, alendronate, teriparatide) were
excluded, but we did not exclude women on hormone therapy (HT), since they may have been taking HT for
menopausal symptoms. Women who had a traumatic wrist fracture (based on description of fracture), or were unable
to communicate in English or give consent were also excluded. Women who had a previous BMD test were not
excluded, since a previous test could be a predictor of receiving osteoporosis therapy.

Interventions

1. Personalised letter to GP from research co‐ordinator 2 weeks and 2 months post‐fracture (patient‐mediated and reminders) incorporating advice on management, recommended therapies and a treatment algorithm (distribution of educational material). Participants also received a letter 2 weeks and 2 months post‐fracture with advice to see their GP and an educational booklet (patient‐directed component)

2. Control: no information, usual care

Outcomes

Outcomes: Proportion of women who stated that their primary care physician had: Discussed osteoporosis with them + started them on osteoporosis therapy within 6 months of fracture + BMD testing within 6 months + changes in the participant’s knowledge of osteoporosis using the Osteoporosis Knowledge Questionnaire (OPQ). Outcomes were assessed by telephone interviews

Notes

Conflicts of interest and sources of funding as declared by the authors: This trial was funded by a peer‐reviewed grant from the
Canadian Institutes of Health Research (KTS 62358).

The study did not provide sufficient information to allow the re‐calculation of adjusted for clustering effect sizes.

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

“Computer generated list of random numbers in a blinded fashion”

Allocation concealment (selection bias)

Low risk

As above

Protection against contamination

Unclear risk

Although the practices were randomised using a cluster design, it is still unclear if communication between physicians or patients of different groups was possible

Baseline outcomes similar

Low risk

No important differences present between the groups

Baseline characteristics similar

Low risk

Baseline characteristics were reported as similar

Blinding of outcome assessment (detection bias)
All outcomes

Unclear risk

Not clear if telephone interviews were conducted blindly

Incomplete outcome data (attrition bias)
All outcomes

Low risk

Similar reasons and rates of dropouts between both groups

Selective reporting (reporting bias)

Unclear risk

No evidence of published protocol

Other bias

Low risk

No other bias identified

Methods

Study design: cluster‐RCT

Participants

Setting: primary care

Country: UK
24 practices (practice was the unit of randomisation, stratified by primary care group; 3 primary care groups), 2187 patients
Condition: acute LBP

Patients were eligible for this study if they were aged between 18 and 64 years, registered with a GP in
Birkenhead, Wallasey or West Wirral Primary Care Groups (PCGs), and had consulted their GP about an episode of
acute low back pain for which they had not already sought advice during the preceding 6 months.

Interventions

1. Educational outreach visit + guidelines (educational material)+ poster of guidelines + referral forms with guidelines + access to fast‐track physiotherapy and a back clinic                      

2. Control: standard practice

Outcomes

Rate of referral for lumbar spine x‐ray within 3 months

Number of sickness certificates issued

Number of prescribed opioids or muscle relaxants

Number referred to secondary care

Number referred to physio or educational programme

Notes

Conflicts of interest and sources of funding as declared by the authors: None reported

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

Random allocation, each centre was given a unique identifier, stratification by primary care group

Allocation concealment (selection bias)

Low risk

Central allocation

Protection against contamination

Unclear risk

Although the practice was the unit of randomisation, it is not clear if communication between practices could affect the results

Baseline outcomes similar

Unclear risk

Not specified

Baseline characteristics similar

Low risk

Baseline characteristics are reported as similar

Blinding of outcome assessment (detection bias)
All outcomes

High risk

“Only one research assistant was employed and blind outcome assessment was not possible.”

Incomplete outcome data (attrition bias)
All outcomes

Low risk

No missing outcome data

Selective reporting (reporting bias)

Unclear risk

Insufficient information as no published protocol

Other bias

Low risk

No other bias identified

Methods

Study design: cluster‐RCT

Participants

Setting: Primary care

Country: UK
247 practices randomised; practice was unit of randomisation and analysis‐stratified randomisation by radiology department (3 radiology departments) and practice size.

The audit and
feedback intervention was delivered to all eligible GPs
according to study design.
Condition: acute LBP or knee pain

Interventions

2 x 2 factorial design

1. Distribution of educational materials + audit and feedback (number of practice referrals compared with peers)
 2. Distribution of educational materials + reminders (messages on x‐ray results)
 3. Distribution of educational materials + audit and feedback + reminders
 4. Distribution of educational materials (guideline) (control group)

Outcomes

Number of lumbar or knee radiographs requested per 1000 patients for 2 years

Notes

Intervention fidelity: Measured attachment rate of educational reminder messages to x‐ray reports

Conflicts of interest and sources of funding as reported by the authors: The study was funded by the UK National Health Service Research and
Development Primary Secondary Interface Programme. The Health Service Research Unit, University of Aberdeen, is funded by the Chief
Scientist Office of the Scottish Office Department of Health. The Centre for Health Services Research, University of Newcastle upon Tyne, and the
Health Services Research Unit, University of Aberdeen are members of the Medical Research Council Health Services Research Collaboration.

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

Computer‐generated random number tables. 

Allocation concealment (selection bias)

Low risk

 Performed centrally by statistician

Protection against contamination

Low risk

The intervention was individualised messages and feedback to practices and therefore the risk of it being disseminated to other practices is low

Baseline outcomes similar

High risk

There was baseline imbalance between the groups

Baseline characteristics similar

Unclear risk

Not specified

Blinding of outcome assessment (detection bias)
All outcomes

Low risk

Outcomes were measured objectively by radiology departments

Incomplete outcome data (attrition bias)
All outcomes

Low risk

No missing outcome data

Selective reporting (reporting bias)

Unclear risk

No published protocol

Other bias

Unclear risk

The intervention of attaching messages to radiology reports was not consistently applied across sites. The site where the messages were attached by an operator pressing a key had an attachment rate of 40% while the other 2 sites had a rate of close to 100%                                                          

Methods

Study design: Cluster‐RCT

Participants

Setting: Primary care

Country: Netherlands

67 GPs eligible to participate, 41 of these completed outcome reports, 531 participants

Condition: Low back pain

The participating GPs were asked to recruit consecutive patients with a new episode of low back pain as the main reason
for consultation. Low back pain was defined as pain, discomfort, stiffness, or fatigue between the lower edge of the shoulder
blades and the gluteal fold either with or without radiation to the legs. Patients who were pregnant, younger than 16 years of
age, or not familiar with the Dutch language were excluded. Only patients who were diagnosed with “nonspecific low back
pain” and no “red flags” present were included in the analyses.

Interventions

1. Two‐hour workshop (negotiation skills) , guideline on low back pain and guidance on low back pain for occupational physicians, 2 scientific articles, a patient education tool and a management decision tool (distribution of educational materials)

2. Control: no intervention, usual care

Outcomes

Number of referrals to a therapist (physical, exercise, or manual therapist)

Prescription of pain medication on a time‐contingent basis

Prescription of paracetamol versus NSAIDs

Adequacy of patient education.

Notes

Conflicts of interest and sources of funding as declared by the authors: The manuscript submitted does not contain information about medical
device(s)/drug(s).
No funds were received in support of this work. No benefits in any form have been or will be received from a commercial party related
directly or indirectly to the subject of this manuscript.

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

"Computer‐generated random‐ list of numbers” used

Allocation concealment (selection bias)

High risk

“Research team knew which practices received which intervention."

Protection against contamination

Unclear risk

The unit of allocation was the GP so there is a risk of communication between GPs allocated to different groups but working in the same practice

Baseline outcomes similar

Unclear risk

Not specified

Baseline characteristics similar

Unclear risk

Baseline characteristics were reported as similar

Blinding of outcome assessment (detection bias)
All outcomes

High risk

“Research team knew which practices received which intervention."

Incomplete outcome data (attrition bias)
All outcomes

Low risk

Intention‐to‐treat analysis performed                                                                                                                                 

Selective reporting (reporting bias)

Unclear risk

No protocol was published (although the protocol is mentioned in the study)

Other bias

High risk

Possible bias as GPs recruited participants. Possible recollection or report bias due to self‐reported outcomes                                                                                                                               

Methods

Study design: RCT            

Participants

Setting: Primary care

Country: USA 
15 practices, 159 providers, 327 participants (randomisation at patient level)
Condition: women aged 50 to 89 who had suffered a fracture (any type) and therefore high likelihood of osteoporosis

The goal in participant selection was to identify older patients who had fractures indicating an increased
risk of osteoporosis, had not received a BMD measurement or a medication for osteoporosis, and did not have
medical conditions or other factors that would contraindicate the interventions.

Interventions

1. Reminders: electronic medical record message about participant’s risk of osteoporosis + distribution of education materials (with guidelines)
2. Reminders + distribution of education materials + patient‐directed component (educational material and reminder to see the GP)
3. Control: standard practice

Outcomes

GP outcomes: proportion of study population who received medication for osteoporosis or a BMD test within 6 months after the intervention

Participant outcomes: regular physical activity, total caloric expenditure, total calcium intake and patient satisfaction

Notes

Conflicts of interest and sources of funding as declared by the authors: This study was supported by a research
contract through Merck & Co. Inc. The funding organization was not involved in the design or conduct of the study;
the collection, management, analysis, or interpretation of the data; or in the preparation or approval of this manuscript.

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

Computerised random number generator used                                                                                                                            

Allocation concealment (selection bias)

Low risk

 "The study statistician randomised and assigned participants to the study groups"

Protection against contamination

Unclear risk

Possible contamination as the participant was the unit of randomisation

Baseline outcomes similar

Unclear risk

Not specified

Baseline characteristics similar

Low risk

Baseline characteristics were reported as similar

Blinding of outcome assessment (detection bias)
All outcomes

Low risk

Study analyst “was blinded to the treatment groups.”

Incomplete outcome data (attrition bias)
All outcomes

Low risk

No missing outcome data

Selective reporting (reporting bias)

Unclear risk

Protocol mentioned in the study but not published

Other bias

Unclear risk

Unclear if participants were blinded                                                                             

Methods

Study design: Cluster‐RCT

Participants

Setting: Primary care

Country: Australia
78 practices, 92 GPs were randomised and participated in the study (randomisation at practice level)
Condition: Low back pain

Patient participant inclusion criteria were people presenting with acute (less than three months duration) non‐specific LBP and
aged 18 years or older.

Interventions

Intervention group: 2 facilitated, interactive, educational workshops aiming to facilitate behaviour change plus distribution of educational DVDs to all physicians

Control group: usual care

Outcomes

Primary outcomes were patient outcomes but due to low numbers of patients recruited, these were not measured.

Secondary outcomes included self‐reported behavioural change and number of x‐ray and CT requests

Notes

Not all physicians participated in the full intervention. Only 36 (61%) attended the workshops and an additional 6 watched the DVDs. However the analysis included all physicians

Conflicts of interest and sources of funding as declared by the authors: The authors have declared that no competing interests exist. The IMPLEMENT trial was funded by the Australian National Health and Medical Research Council (NHMRC) by way of a Primary Health Care Project
Grant (334060). SDF and DAO are supported by NHMRC Early Career Fellowships. RB is supported in part by a NHMRC Practitioner Fellowship. The funders had no
role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

An independent and blinded statistician implemented the randomisation (computer‐generated random numbers) after stratifying practices based on the number of GPs and whether the practice was rural or not

Allocation concealment (selection bias)

Low risk

Allocation was concealed from the investigators until baseline data had been collected from GPs

Protection against contamination

Low risk

The whole practice was randomised to reduce risk of contamination

Baseline outcomes similar

Unclear risk

There is no information on baseline outcomes (on x‐ray and CT numbers)

Baseline characteristics similar

High risk

There was some baseline imbalance, with control GPs more likely to identify themselves as having an interest in low back pain (24% versus 9%) and more GPs in the intervention group undertaking low back pain continuing education in the past year (16% versus 5%)

Blinding of outcome assessment (detection bias)
All outcomes

Low risk

Investigators not involved in the intervention who entered the data and the statistician were blinded to group allocation until the statistical analysis was completed

Incomplete outcome data (attrition bias)
All outcomes

High risk

The primary outcomes were not measured due to low numbers of participants recruited

Selective reporting (reporting bias)

High risk

As above. Also, subgroup analysis to investigate the effect on GPs who attended the workshop (as per protocol) was not done

Other bias

High risk

Not all physicians participated in the full intervention. Only 36 (61%) attended the workshops and an additional 6 watched the DVDs. However the analysis included all physicians

Participants could not be blinded. Primary outcomes not measured. Reported outcomes were self reported

Methods

Study design: RCT

Participants

Setting: Primary care

Country: Northern Ireland

40 GP principals randomised

Condition: Shoulder pain

Interventions

1. Educational meeting/workshop on shoulder management and injection technique training on mannequins

2. As above plus injection training on real patients

Outcomes

Number of shoulder injections

Referrals to physiotherapy and injection clinics over last 6 months

Level of confidence (10 cm VAS)

Notes

Conflicts of interest and sources of funding as declared by the authors: None reported.

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Unclear risk

Not reported in the study

Allocation concealment (selection bias)

Unclear risk

Not reported in the study

Protection against contamination

Low risk

The randomisation was at physician level. The intervention required the physician to be present to practise their skills and therefore contamination is unlikely

Baseline outcomes similar

Low risk

No important differences between the groups

Baseline characteristics similar

Low risk

No important differences between the groups

Blinding of outcome assessment (detection bias)
All outcomes

High risk

Not done, this was self reported

Incomplete outcome data (attrition bias)
All outcomes

Unclear risk

"One GP's assessment return after training was incomplete and another failed to make a return. Both of these were in the "mannequin only" training group". It is unclear what the exact proportion of missing data was and if this could bias the results

Selective reporting (reporting bias)

Unclear risk

No published protocol of the study

Other bias

High risk

Results were based on self reporting by GPs. GPs were not blinded

Methods

Study design: RCT

Participants

Setting: Primary care

Country: USA

59 GPs

59 patients: workers 18‐60 years old with VDPQ scores suggesting a high risk of prolonged work disability (i.e. VDPQ score of at least 0.37 (scale = 0‐1))

Condition: Low back injury

Interventions

1.Distribution of educational materials + reminders to GPs (letters regarding the specific patient with advice on how to limit work loss)               

2. Control

Outcomes

3‐month work absence rate

VDPQ (disability) score

Satisfaction with health care

Impact of health care on return to work

Days of work loss

Days until first return to work

Notes

Conflicts of interest and sources of funding as reported by the authors: Supported, in part, by The National Institute on Disability
and Rehabilitation Research, Washington, D.C. (grant H133E30014‐95).

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

"Each high risk worker was assigned to the physician intervention group or to the control group, according to a predetermined allotment list developed from a table of random digits balancing the assignments with every six workers".

Allocation concealment (selection bias)

High risk

“Physician was sent a letter identifying the patient and the patient's high risk for work absence 3 months after injury."

"The workers themselves knew whether they were in the intervention or control groups".

Protection against contamination

Unclear risk

Physicians could be working in the same practice.

Baseline outcomes similar

Unclear risk

Not specified

Baseline characteristics similar

Unclear risk

Not specified

Blinding of outcome assessment (detection bias)
All outcomes

High risk

"The follow up interviewer was not blinded to the VDPQ scores or groups assignments".

Incomplete outcome data (attrition bias)
All outcomes

Low risk

No missing outcome data

Selective reporting (reporting bias)

Unclear risk

No published protocol of the study

Was knowledge of the allocated interventions adequately prevented during the study?

High risk

Other bias

Unclear risk

Participants were not blinded

Methods

Study: ITS

Participants

Setting: Primary care

Country: UK

Number of practices and GPs not reported. Analysed 2100 x‐ray referrals

The mean age of the 2100 patients whose radiography reports were selected for review was 53.6 years (range = 7 to 94 years), 57.9% were female.

Interventions

Distribution of guidelines (by Royal College of General Practitioners and Royal College of Radiologists)

Outcomes

Number of primary care referrals for radiography of the lumbar spine

Notes

Conflicts of interest and sources of funding as declared by the authors: The lead author is sponsored by a MRC training fellowship in Health
Services Research. This study was funded in part by task‐linked NHS R&D support funding.
The work was undertaken at University of Cambridge, which received funding from the NHS Executive eastern region. The views expressed in
this publication are those of the authors and not necessarily those of the NHS Executive eastern region.

Risk of bias

Bias

Authors' judgement

Support for judgement

Was the intervention independent of other changes?

Unclear risk

No other changes were reported at the time of the guideline dissemination. However, such changes (such as waiting times, funding arrangement changes or changes in the prevalence of low back pain were possible)

Was the shape of the intervention effect pre‐specified?

Unclear risk

This was not specified in the study

Was the intervention unlikely to affect data collection?

Low risk

The intervention was independent of the data collection method

Was knowledge of the allocated interventions adequately prevented during the study?

Low risk

Participants were not aware of the study

Were incomplete outcome data adequately addressed?

Unclear risk

The study does not give sufficient information on this

Other bias

Low risk

No other risks identified

Methods

Study design: Cluster‐RCT

Participants

Setting: Primary care

Country: France

155 GPs (randomisation was stratified by University; 20 different Universities), 772 participants

Condition: Chronic musculoskeletal pain

All included patients were over 18 years of age, had been suffering for at least 3 months from sustained daily chronic pain, of musculoskeletal origin

affecting the locomotor system, and were regularly taking painkillers.

Interventions

1.Training in the use of VAS and HAD scales for pain (educational meeting + patient‐mediated intervention)    

2. Control: usual care                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                              

Outcomes

GP outcomes: changes in prescription of painkilling modalities

Patient outcomes: Level of relief obtained (numerical relief scale) (self reported by participant)

Notes

Conflicts of interest and sources of funding as declared by the authors: The Fondation de la Caisse Nationale de Prevoyance funded the study and the Nukleus company provided material support.

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Unclear risk

“Randomisation was stratified by University.” No further information provided

Allocation concealment (selection bias)

Unclear risk

No information provided

Protection against contamination

Low risk

"In order to avoid contamination bias, physicians from both groups never met during the course of the study".

Baseline outcomes similar

Low risk

"Painkilling treatment prescribed at inclusion was comparable in both groups ‐ the main difference being that a larger number of patients in the scale group were taking level 3 analgesics, although the number of patients concerned was very small".

Baseline characteristics similar

Low risk

"The characteristics of the physicians were comparable for both groups", "The patient groups in the treatment and control groups were of similar nature", "pain location was comparable in the 2 groups except for back pain"

Blinding of outcome assessment (detection bias)
All outcomes

Unclear risk

Participants probably unaware of GPs' training. Not clear how secondary outcomes were assessed and by whom

Incomplete outcome data (attrition bias)
All outcomes

Unclear risk

Not reported

Selective reporting (reporting bias)

Unclear risk

No published protocol of the study

Other bias

Low risk

No other risks identified

Methods

Study design: cluster‐RCT

Participants

Setting: Primary care

Country: UK
69 practices (practice is the level of randomisation), 175 GPs, 43,778 radiological requests
Condition: People potentially requiring an x‐ray of chest, spine or limbs and joints

Interventions

1. Distribution of guidelines + individual feedback on referral rates + graph of the average radiation dose for different examinations (educational material and audit/feedback)
2. Control: standard care

Outcomes

Professional practice: number of x‐rays requested (chest, limbs and joints, spine) within 12 months

Notes

Conflict of interests and sources of funding as declared by the authors: This study was funded by The South Thames Project
Grant Scheme.

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

“Practices were randomly allocated to an intervention or a control group using a stratified randomisation.”

Allocation concealment (selection bias)

Unclear risk

Not reported

Protection against contamination

Unclear risk

Randomisation was at practice level but unclear if practices of different groups could communicate

Baseline outcomes similar

Unclear risk

Not reported

Baseline characteristics similar

Unclear risk

Not reported. Although randomisation happened using 10 strata depending on "numbers of partners, referral rates, fund‐holding status, and having received guidelines in a previous study".

Blinding of outcome assessment (detection bias)
All outcomes

Low risk

Outcome objectively collected (routine data, collected electronically)

Incomplete outcome data (attrition bias)
All outcomes

Low risk

All outcomes were reported

Selective reporting (reporting bias)

High risk

Results (referrals for x‐rays) not reported per 1000 patients. Protocol of the study not published

Other bias

Low risk

No other risks identified

Methods

Study design: cluster‐RCT

Participants

Setting: Primary care

Country: USA

15 practices randomised after stratification (randomisation at practice level), 123 primary care physicians, 10,354 patients

Condition: women 65 to 89 years of age on 3/31/2003 and high likelihood of osteoporosis with a visit between 4/1/2001 and 3/31/2003 to
a primary care physician.

Interventions

1. Patient‐directed component (educational material on osteoporosis)

2. Patient‐directed (educational material on osteoporosis) + physician prompt/reminder (reminder on electronic medical record and biweekly letter to physician listing patients needing treatment)

3. Control: standard care

Outcomes

Professional practice: proportion of patients receiving BMD testing within 12 months; prescription of osteoporosis medication

Notes

Possible risk of contamination

Conflicts of interest and sources of funding as declared by the authors: Dr. Weiss and Dr. Chen are employees of Merck & Co.

The study did not provide sufficient information to allow the re‐calculation of adjusted for clustering effect sizes.

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

"Within stratum, clinics were allocated to the three arms using a random numbers table"

Allocation concealment (selection bias)

Unclear risk

Not reported

Protection against contamination

Unclear risk

Randomisation was at practice level but unclear if practices of different groups could communicate

Baseline outcomes similar

Low risk

Women with previous BMD screening or on osteoporosis medication were excluded from the evaluation

Baseline characteristics similar

Unclear risk

"Although statistically significant baseline differences were found for most of the patient characteristics assessed, only a handful meaningful differences existed".

"Women in the patient mailed reminder arm were less likely to be African American",

"there was variation in health plan enrolment".

There was no assessment of GP baseline characteristics

Blinding of outcome assessment (detection bias)
All outcomes

Low risk

Used "automated clinical and pharmacy data"

Incomplete outcome data (attrition bias)
All outcomes

Low risk

Missing outcome data balanced in numbers across intervention groups

Selective reporting (reporting bias)

Unclear risk

Insufficient information provided. No published protocol of the study

Other bias

Low risk

Methods

Study design: RCT

Participants

Setting: Primary care

Country: Canada
4264 patients were randomised; patients included men and women 50 years of age and older who had suffered a previous fracture and had not received a BMD or osteoporosis medication
Condition: osteoporosis

Interventions

1. Notification letter to primary care physician (reminder) about the patient's fracture accompanied by educational material

2. Notification letter to primary care physician accompanied by educational material as above plus patient‐directed intervention (educational material and reminder)

3. Control group: usual care

Outcomes

BMD and osteoporosis medication

Notes

Conflicts of interest and sources of funding as declared by the authors: Wiliam Leslie has received speaker fees from Merck Frosst and Amgen;

he has unrestricted research grants from Merck Frosst, Sanofi‐Aventis, Procter and Gamble, Novartis, Amgen and Genzyme; he is on the advisory boards

for Genzyme, Novartis, and Amgen. Patricia A Caetano has received unrestricted research grant from Amgen. No other competing interests were declared.

The article was funded by the Manitoba Patient Access Network whose mandate is to identify, advocate, support and guide health system change and process improvement

initiatives. The network is financially supported by the Wait Times Reduction Fund and receives secretariat services from Manitoba Health`s Wait Times Task Force.

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

"The randomisation was done using a centralised computer‐based algorithm"

Allocation concealment (selection bias)

Low risk

"The computer‐based algorithm concealed the allocation process from the clinical investigators"

Protection against contamination

Unclear risk

It is not clear if primary care physicians in the control group were also physicians of patients under the intervention groups

Baseline outcomes similar

Low risk

"The groups were well balanced in terms of age, sex and site of fracture"

Baseline characteristics similar

Low risk

As above. The patients included had not received previous BMD test or osteoporosis medication

Blinding of outcome assessment (detection bias)
All outcomes

Low risk

The outcomes were taken from a centralised database

Incomplete outcome data (attrition bias)
All outcomes

Low risk

The analysis followed an intention‐to‐treat methodology and all participants were included

Selective reporting (reporting bias)

Low risk

All outcomes as per study protocol were reported

Other bias

Low risk

No other risks identified

Methods

Study design: RCT

Participants

Setting: Primary care

Country: Canada
266 GPs, 272 patients (unit of randomisation the patients)

Condition: 50 years or older and distal forearm fracture (high likelihood of osteoporosis) 

Patients were excluded if they were already receiving treatment for osteoporosis with a bisphosphonate,
were unable or unwilling to provide informed consent, had no fixed address, were residing outside the Capital Health region
or were residing in a long‐term care facility.

Interventions

1. Distribution of guidelines endorsed by five local leaders (educational material) + physician reminder (patient‐specific letter to GP) + patient‐directed component (education and counselling via telephone)
2. Control group, usual care

Outcomes

Proportion of participants who had received BMD test

Prescription of osteoporosis medication

Composite measure of quality of guideline‐concordant or “appropriate” care

Patient Outcomes:  Health status (SF‐12)

Osteoporosis‐related quality of life

Wrist‐related functional outcomes

Osteoporosis‐related knowledge

Satisfaction with care

Costs: intervention cost per patient (this outcome was reported in a different publication, Majumdar 2007)

Notes

Conflicts of interest and sources of funding as declared by the authors: None declared for Sumit Majumdar, Jeffrey Johnson,
Finlay McAlister, Debbie Bellerose, Anthony Russell, Don Morrish, Walter Maksymowych or Brian Rowe. David Hanley has been a clinical investigator
in phase III clinical trials of bisphosphonates manufactured by Proctor & Gamble, Merck and Novartis; in addition, he has received speaker fees from
and has been a paid consultant on advisory boards of these companies.Sumit Majumdar, Jeffrey Johnson, Finlay McAlister and
Walter Maksymowych receive salary support awards from the Alberta Heritage Foundation for Medical Research; Sumit Majumdar and Finlay McAlister
receive salary support awards from the Canadian Institutes of Health Research; Jeffrey Johnson and Brian Rowe hold Canada Research Chairs; and
Finley McAlister holds the Aventis/Merck‐Frosst Chair in Patient Health Management. The study was supported by peer‐reviewed grants from the Canadian
Institutes of Health Research.

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

 “Allocation was concealed by application of various block sizes and by use of a secure,centralised, Internet based computer‐generated randomisation system.”

Allocation concealment (selection bias)

Low risk

As above

Protection against contamination

High risk

The randomisation unit was the patient and therefore there could be contamination at physician level

Baseline outcomes similar

Unclear risk

Not reported

Baseline characteristics similar

Unclear risk

"Intervention and control patients were comparable" and "all multivariable analysis adjusted" for any differences. There was no assessment of physician characteristics

Blinding of outcome assessment (detection bias)
All outcomes

Low risk

"Research nurses collected outcome data without knowledge of allocation status".

Incomplete outcome data (attrition bias)
All outcomes

Low risk

All participants were accounted for

Selective reporting (reporting bias)

Unclear risk

All outcomes listed in Methods were reported. However, there was no published study protocol

Was knowledge of the allocated interventions adequately prevented during the study?

Low risk

Other bias

Low risk

"Neither physicians nor patients were aware of the study outcomes"

Methods

Study design: Cluster‐RCT

Participants

Setting: Primary care

Country: Canada

249 providers, patients; before intervention = 3280, and post intervention = 2883

Condition: Osteoarthritis

All NSAID, COX‐2 inhibitor, or acetaminophen prescriptions dispensed to patients with osteoarthritis were identified.
Patients with osteoarthritis were those who had at least one diagnosis for osteoarthritis (ICD‐9 code 715) in the
previous 1215 days.

Interventions

1. Distribution of educational material (decision‐tree laminated sheet) without face‐to‐face discussion

2. 90‐minute workshops on osteoarthritis without distribution of the decision‐tree laminated sheet

3. 90‐minute workshop on osteoarthritis + distribution of the decision‐tree laminated sheet      

4. Controls: standard care, no educational intervention

Outcomes

Professional practice: prescription of medications for elderly patients suffering from osteoarthritis

Notes

Conflicts of interest and sources of funding as declared by the authors: Drs. LeLorier, Choquette, Bessette and Rahme have served as consultants
and paid speakers for Merck & Co. Inc. and for Pfizer Inc. Ms. Beaulieu is an employee at Merck Frosst Canada Ltd. Dr. Rahme is a
research scholar funded by The Arthritis Society.

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Unclear risk

“Each town was randomly allocated 1 of 4 intervention options”. No further details of the randomisation method are given

Allocation concealment (selection bias)

Unclear risk

Not reported

Protection against contamination

Low risk

"The towns were geographically distant to minimise cross‐contamination"

Baseline outcomes similar

Unclear risk

Not reported

Baseline characteristics similar

Low risk

"Patient and physician characteristics were on average similar among the four groups"

Blinding of outcome assessment (detection bias)
All outcomes

Low risk

Outcomes data were obtained from electronic databases (Provincial Health Care Fund database)

Incomplete outcome data (attrition bias)
All outcomes

Low risk

Intention‐to‐treat analysis performed.

Selective reporting (reporting bias)

Unclear risk

No study protocol was published

Other bias

Unclear risk

Not clear if participants were blinded with regards to the study outcomes

Methods

Study design: cluster‐RCT

Participants

Setting: Primary care

Country: UK
30 practices, 182 MRI requests
Condition: people who potentially require MRI for knee or lumbar problems

Interventions

1. Distribution of local guidelines + practice‐based seminar during which a 15‐minute video was shown (outreach visit)
2. Distribution of local guidelines + feedback on practice‐specific MRI use and comparative data on orthopaedic and neurosurgical referrals (audit and feedback)
3. (1 + 2) Distribution of educational material plus outreach visits plus audit and feedback
4. Control group: distribution of local guidelines by post

Outcomes

Professional practice: proportion of MRI requests that are in concordance with guideline (length of follow‐up not clear)
Cost outcome: intervention cost

Notes

Conflicts of interest and sources of funding: The study was funded by the NHS Research and Development Programme on the

Primary Secondary Care Interface.

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

“Randomisation was performed using a random numbers table.”

Allocation concealment (selection bias)

Unclear risk

 Not reported

Protection against contamination

Unclear risk

Randomisation was at practice level but it is unclear if practices of different groups could communicate

Baseline outcomes similar

Unclear risk

Not reported

Baseline characteristics similar

Unclear risk

Not reported

Blinding of outcome assessment (detection bias)
All outcomes

Low risk

 "Anonymised interview data were assessed by a study panel"

"Panel members were blinded to study randomisation"

Incomplete outcome data (attrition bias)
All outcomes

Unclear risk

Not reported

Selective reporting (reporting bias)

Unclear risk

Insufficient information, no study protocol was published.

Other bias

Unclear risk

It is unclear if participants were blinded on the outcome measures

Methods

Study design: Cluster‐RCT

Participants

Setting: Primary care

Country: Germany

75 practices, 75 GPs, 1021 patients. The GPs were the unit of randomisation.

Condition: Osteoarthritis  

To be eligible for inclusion, patients had to be age 18 years and diagnosed with OA in
the knee or the hip according to the American College of Rheumatology criteria   

Interventions

1. Educational meeting/workshop (2 interactive quality circle meetings of 8 hours each on management of osteoarthritis and motivational skills) plus educational material (written educational material + patient educational material including leaflets, booklets and audio CDs)

2. 1 + practice nurse training to call participants and complete questionnaire on osteoarthritis management

3. Control

Outcomes

Patient outcomes: Change in quality of life, assessed by the German version of the Arthritis Impact Measurement Scales Short Form (AIMS2‐SF),

Health service utilisation

Prescriptions

Physical activity.

Notes

Conflicts of interest and sources of funding as declared by the authors: None reported

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

GPs randomised by SPSS

Allocation concealment (selection bias)

Unclear risk

Not reported

Protection against contamination

High risk

As the randomisation was at GP level, it may have been possible that communication between intervention and control professionals could have occurred.

Baseline outcomes similar

Low risk

No significant differences between participant groups were identified

Baseline characteristics similar

Low risk

"No statistically significant differences in the outcome measures" at baseline were found

Blinding of outcome assessment (detection bias)
All outcomes

Unclear risk

Participant answers were cross‐checked by a research assistant but it is not clear if the assistants were blinded

Incomplete outcome data (attrition bias)
All outcomes

Low risk

No missing outcome data."No practice dropped out during the study"

Selective reporting (reporting bias)

Low risk

Published study protocol

Other bias

Unclear risk

The participants were not blinded and this may have affected the results

Methods

Study design: RCT

Participants

Setting: Primary care
Country: Canada (Quebec, Sherbrooke)
Women and men aged 50 years or older with a fracture confirmed on radiograph were screened by the study co‐ordinators for circumstances suggestive of a fragility fracture when they attended the orthopedic outpatient clinics.

Patients unable to speak French or English fluently, as well as those with known severe psychiatric problems, delirium, or dementia were not approached because of their inability to provide valid informed consent. 881 patients were randomised

Interventions

Group 1: Verbal and written information on osteoporosis to patient (patient‐directed component) and letter with specific management plan sent to their treating physician (GP reminder). Patient reminders at 6 and 12 months. Reminder to physician if patient untreated at 6 months

Group 2: Verbal and written information on osteoporosis to patient (patient‐directed component) and letter with specific management plan sent to their treating physician (GP reminder). Blood tests and BMD test ordered for patient and results sent to the physician (patient‐mediated intervention). Patient reminders at 4,8 and 12 months and physician reminders at 4 and 8 months if patient remained untreated

Control group: Telephone interviews at 6 and 12 months to assess treatment scores.

Outcomes

Osteoporosis‐related drug treatment at 12 months was the main outcome.

Notes

Conflicts of interest and sources of funding as reported by the authors: Supported by unrestricted research grants from Merck Canada, The
Alliance for Better Bone Health at Procter & Gamble (now Warner Chilcott) and Sanofi‐Aventis, Amgen Canada, Novartis Pharmaceuticals
Canada Inc., and Servier Canada; and by the Centre de Recherche Clinique Étienne‐LeBel (CRC), Centre Hospitalier Universitaire de
Sherbrooke (CHUS), which received a team grant from the Fonds de la Recherche en Santé du Québec.

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

Patients were recruited concurrently by a research co‐ordinate from consecutive fracture clinics. Attending surgeons did not play an active role in recruitment

Recruitment to the control group was random but not randomised relative to recruitment to the intervention groups in order to avoid contamination between participants

Allocation concealment (selection bias)

Low risk

The consent form outlined all 3 interventions but did not suggest that any of the 3 was more effective. Primary care physicians were blinded to which group their patients were assigned to

Protection against contamination

Unclear risk

Participants were protected from contamination by separating the control and intervention groups but the possibility of physician contamination was not explored

Baseline outcomes similar

Low risk

There were no significant differences between groups

Baseline characteristics similar

High risk

The participants in the first intervention group were older (median age 67 while for the control group this was 64 and for the second intervention group was 63)

Blinding of outcome assessment (detection bias)
All outcomes

High risk

The assessors were not blinded

Incomplete outcome data (attrition bias)
All outcomes

Low risk

The data were analysed by intention‐to‐treat methodology

Selective reporting (reporting bias)

Unclear risk

The main outcome was reported but we could not find a published study protocol

Other bias

Unclear risk

Unclear if the participants were blinded

Methods

Study design: RT (no control group)

Participants

Setting: Primary care

Country: USA

Aim 1: 240/298 patients: retrospective review

Aim 2: RT 50 patients were randomised to 1 of 2 interventions

Condition: Osteoporosis

The inclusion criteria included an age of over fifty years (for women) or over sixty‐five years (for men), a fragility fracture
of the distal part of the radius, no evaluation with a bone mineral density examination within two years before the fracture,
and no current treatment with antiresorptive medication or hormone replacement therapy.

Interventions

1. Orthopaedic surgeon orders BMD and BMD results are forwarded to primary care physician (patient‐mediated)

2. Letter from orthopaedic surgeon to primary care physician outlining guidelines for osteoporosis screening (educational material and reminder)

No control group

Outcomes

Professional practice: the rates of evaluation (BMD testing) within 6 months and treatment (discussion and initiation) for osteoporosis

Notes

Conflicts of interest and sources of funding as declared by the authors: In support of their research for or preparation of this work, one or more of the authors received, in any one year, outside funding or grants of
less than $10,000 (a Procter and Gamble Development Grant). Neither they nor a member of their immediate families received payments or other
benefits or a commitment or agreement to provide such benefits from a commercial entity. No commercial entity paid or directed, or agreed to pay or
direct, any benefits to any research fund, foundation, division, center, clinical practice, or other charitable or nonprofit organization with which the
authors, or a member of their immediate families, are affiliated or associated.

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Unclear risk

No information on how the randomisation took place

Allocation concealment (selection bias)

Unclear risk

No information is provided on allocation concealment

Protection against contamination

High risk

Randomisation happened at patient level

Baseline outcomes similar

Unclear risk

These were not assessed at baseline

Baseline characteristics similar

Low risk

Baseline characteristics between participant groups seemed similar

Blinding of outcome assessment (detection bias)
All outcomes

Unclear risk

Not specified in the paper.Not clear if the assessor was blinded

Incomplete outcome data (attrition bias)
All outcomes

Low risk

All participants were accounted for

Selective reporting (reporting bias)

Unclear risk

No study protocol was published

Other bias

Unclear risk

It is unclear if participants were blinded with regards to the outcomes

Methods

Study design: Cluster‐RCT

Participants

Setting: Primary care

Country: USA

85 physicians, 2020 patients, 14 group practice sites, randomisation at practice level

Condition: Acute low back pain 

Patients were eligible for study inclusion if they met all three of the following criteria: (1) presence of low back pain; (2) duration
of current symptoms less than 6 weeks; and (3) no episodes of pain reported or office visits for low back pain in the preceding year.

Interventions

1. Distribution of guideline on the management of acute low back pain + educational meeting + feedback on back pain encounters + individual follow‐up visit by investigator 6 months afterwards and another feedback on back encounters + educational material for patients including a videotape (educational material + meeting + audit + outreach)
2. Education materials for patients: pamphlet and video and 2 reminders within the first 3 months to clinicians to use these materials (educational material)
3. 1 + 2
4. Control group

Outcomes

Professional practice: 

Proportion of lumbar plain x‐rays

CT or MRI consistent with guideline within 12 months

Subspecialty referral

Physiotherapy referral
Patient outcomes:

Beliefs about care

Satisfaction with care

Clinical outcome measures using validated instruments

Notes

Conflicts of interest and sources of funding as reported by the authors: Agency for Health Care Policy and Research,
Public Health Service, Department of Health and Human Services, Grant #: RO1 HS07069.

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Unclear risk

“Clinician practices were stratified by affiliation and then, using sealed envelopes, randomised by an investigator to 4 groups in a 2 × 2 factorial design.”

Allocation concealment (selection bias)

Low risk

Sealed envelopes used

Protection against contamination

Unclear risk

Randomisation was at practice level and also stratification by affiliation was used which can reduce the risk of contamination. However, there may have been contamination at patient level

Baseline outcomes similar

High risk

"The intervention group had substantially higher utilization of radiologic and specialty services during the baseline period".

"Similar baseline differences were found for utilization of services inconsistent with the guideline".

"These differences remained, though were diminished, after adjustment for patient characteristics that were strongly associated with utilisation".

Baseline characteristics similar

High risk

Patient and clinician characteristics between the groups were not similar

Blinding of outcome assessment (detection bias)
All outcomes

Unclear risk

Not reported if assessors were blinded

Incomplete outcome data (attrition bias)
All outcomes

Unclear risk

Not recorded if all charts audited

Selective reporting (reporting bias)

Unclear risk

Insufficient information provided; no study protocol was published

Was knowledge of the allocated interventions adequately prevented during the study?

Unclear risk

It is unclear if the participants were blinded with regards to the outcomes

Other bias

High risk

"The four intervention groups were collapsed into two for analysis and reporting purposes" after the patient education intervention revealed no effect. This was not in accordance with the study protocol. Potential unit of analysis error, potential contamination between groups.                                                                        

Methods

Study design: cluster‐RCT

Participants

Setting: Primary care

Country: USA

828 providers, 13,455 patients
Condition: people with high likelihood of osteoporosis and high risk of future fracture

The at‐risk patients were women 65 years of age, men and women 65 years of age with a prior fracture, and men and women 65 yr of age who
used oral glucocorticoids.

Interventions

1. Physician education by trained pharmacists or nurses (academic‐detailing approach via outreach visits) + educational material and handouts for patients
2. Patient‐directed component: 3 mailed letters with educational material and questions to ask the physician
3. 1 + 2
4. Control: standard care

Outcomes

Professional practice (primary outcome): number of patients who began osteoporosis medication or had BMD test within 12 months

Patient outcomes (secondary outcomes): fracture of wrist, humerus or hip

Notes

Conflicts of interest and sources of funding: Dr Solomon possessed research grants in the past from Merck and Proctor & Gamble. Dr Gauthier is an employee of the Arthritis
Foundation, which partially funded this study. All other authors state that they have no conflicts of interest.This study was supported by NIH Grant AR48616 and
the Arthritis Foundation.

The study did not provide sufficient information to allow the re‐calculation of adjusted for clustering effect sizes.

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

Random‐number generator. The randomisation was at doctor level

Allocation concealment (selection bias)

Unclear risk

Not reported        

Protection against contamination

Low risk

"All patients in a given physician's practice were randomised as a group (cluster randomisation) to avoid contamination within a given physician's practice".

Baseline outcomes similar

Unclear risk

Not reported

Baseline characteristics similar

Low risk

The baseline characteristics of patients and physicians were similar across the groups

Blinding of outcome assessment (detection bias)
All outcomes

Low risk

Data used were electronic from outside sources (from Medicare, PACE, inpatient and outpatient coding)

Incomplete outcome data (attrition bias)
All outcomes

Low risk

No missing outcome data; intention‐to‐treat analysis

Selective reporting (reporting bias)

Low risk

Primary and secondary outcomes reported. There is published study protocol.

Other bias

Unclear risk

Unclear if the participants were blinded

Methods

Study: Cluster‐RCT

Participants

Setting: Primary care

Country: USA

Participants: 6 communities in the state of Michigan: 3 were randomly selected to be controls and 3 were designated as intervention communities. 6 educationally influential physicians (EIs) recruited (1 in each community)

Interventions

1. Local opinion leaders' education: self‐study programme including textbook, audiovisual materials and recent articles on osteoarthritis (distribution of educational material and local opinion leaders)

The aim was to improve the management of patients with OA by focusing on the role of intra‐articular corticosteroids, physical therapy and joint replacement

2. Standard care: no educational package

Outcomes

Total hip arthroplasties, use of intra‐articular corticosteroids, use of physical therapy

Notes

Conflicts of interest and sources of funding as declared by the authors: Supported by Multipurpose Arthritis Center grant 2P 60
AM20557 from the National Institute of Arthritis, Diabetes, Digestive and Kidney Diseases.

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Unclear risk

Unclear how the randomisation took place

Allocation concealment (selection bias)

Unclear risk

Not reported

Protection against contamination

Low risk

Contamination is less likely due to the randomisation at large cluster level

Baseline outcomes similar

Low risk

There were no significant baseline differences between the groups

Baseline characteristics similar

Unclear risk

Not reported

Blinding of outcome assessment (detection bias)
All outcomes

Unclear risk

Not reported, although data were obtained from hospital records

Incomplete outcome data (attrition bias)
All outcomes

Unclear risk

Not reported

Selective reporting (reporting bias)

Unclear risk

No published protocol of the study

Other bias

Unclear risk

It is not clear if the participants were blinded with regards to the outcomes

Methods

Study design: Cluster‐RCT

Participants

Setting: Primary care

Country: UK

160 providers, 91 practices were randomised to training or not training, 155 patients participated in the first part of the trial

Condition: Acute shoulder pain

Patients were eligible if:
1. they were presenting to GPs with pain in one or both shoulders for ≤12 months who would otherwise have received a steroid injection from primary care.
2. had a clinical diagnosis of rotator cuff tendonitis based on both history of pain in the deltoid area and pain during resisted active movement. Some mild restriction of passive
movement was acceptable.
3. were ≥ 16 yrs.
4. were able and willing to give informed consent.

Interventions

1. 60‐minute lecture on shoulder disorders, handouts, training in injection techniques (educational material + meeting)

2. Control (no educational intervention)

Outcomes

Patient Outcomes: Shoulder pain assessed by 4 instruments: score on the British Shoulder Disability Questionnaire (BSDQ), SF‐36, EuroQol and three VAS (night, rest, movement)

Notes

The study had a second part testing whether cortisone injections were better than anaesthetic injections for rotator cuff problems (patients were randomised into 2 groups).

The study included a cost‐effectiveness analysis

Conflicts of interest and sources of funding as declared by the authors: V.M. and J.W. received salary from the
MRC research grant. J.D. has received travel grants from Pfizer, Wyeth, Novartis and Napp and honoraria for tutorials from
Pfizer and Novartis. He has served on advisory boards for pharmaceutical companies including GlaxoSmithKline, Wyeth,
Novartis and IDEA. All other authors have declared no conflicts of interest.This trial was funded by the Medical Research Council
(grant number G0001147) and received support for the education seminars and training events from Merck, Sharp and Dohme. The
MRC established a trial steering committee to advise the grant holders and trial team on trial design, the collection, analysis,
interpretation and writing up of data and publication policy.

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

“Computer generated sequence” used. Practices "were stratified by area"

Allocation concealment (selection bias)

High risk

“Patients were not informed of the allocation.”  But the researchers were not blinded.                        

Protection against contamination

Unclear risk

Randomisation was at practice level but it is not clear if practices of different groups could communicate between themselves

Baseline outcomes similar

Unclear risk

Not reported

Baseline characteristics similar

Low risk

The participant baseline characteristics were reported and were similar

Blinding of outcome assessment (detection bias)
All outcomes

Unclear risk

Not reported; participants completed the pain‐assessment questionnaires

Incomplete outcome data (attrition bias)
All outcomes

Low risk

Intention‐to‐treat analysis was performed          

Selective reporting (reporting bias)

Unclear risk

Not enough information is provided. The study protocol has not been published

Other bias

Unclear risk

  Unclear if participants were blinded with regards to the outcomes.