Table 2 Description of included studies

 Fine 2017 [34] Canada

x-ray, CT or MRI

ITS

No intervention

Components:

a) Policy change (once)

Provider:

a) Government

Rationale: Yes. Government withdrew funding for uncomplicated LBP imaging

Painter: No theory

 Kullgren 2018 [47] US

x-ray, CT or MRI

Stepped wedge cluster RCT

No intervention

Components:

a) GP commitment (performed once)

b) Reminders (post-it notes (provided for every LBP patient) and emails (sent weekly))

c) Patient resources (handout) (provided for every LBP patient)

Provider:

a) Study team

b) Medical assistants working in the primary care clinics

c) Study team

Rationale: Yes. Intervention ‘pre-commitment’ strategy drawn from behavioural economics

Painter: No theory

 Fenton 2016 [33] US

x-ray, CT or MRI

RCT

Guidelines only

Components:

a) Communication training (10 min, provided once)

Provider:

a) Study team

Rationale: Yes. Patient requests may be one factor driving overuse of diagnostic tests and thus, patient-centred communication may address their concerns and reduce imaging requests.

Painter: No theory

 French 2013 [36] Australia

x-rays

Cluster RCT

Guideline only

Components:

a) Educational workshops (3 h, provided twice)

b) DVD of workshop content (given once)

Provider:

a + b) Study team

Rationale: Yes. They use the TDF and a mapping tool to identity BCTs. Previous interview study identifying barriers to behaviour has been published.

Painter: Yes. testing theory where the theoretical framework was specified (TDF) and more than half of the theoretical constructs applied to the intervention were measured and tested (in their 51-item self-developed questionnaire)

 Lin 2016 [48] Australia

x-ray, MRI, CT

ITS

Components:

a) Educational workshops (3 h, provided twice)

b) Audit and feedback (provided once)

c) Clinical tools (LBP decision making tool and Start Back screening tool) (provided once)

Provider:

a) Study team

b) Study team

c) Study team

Rationale: Yes. Previous study of theirs and literature advocating a tailored and theoretically informed approach.

Painter: Yes. Informed by theory.

 Winkens 1995 [63] NL

x-ray

Cluster RCT

No intervention

Components:

a) Audit and feedback (provided five times)

Provider:

a) Internal medicine specialist

Rationale: Yes. Previously conducted internal surveys suggested a reduction of testing based on routinely provided feedback but a causal relationship has not been found/studied in a trial context.

Painter: No theory

 Jackson 2005 [42] US

x-rays

ITS

No intervention

Components:

a) Educational materials (national guidelines)

Provider:

a) Government

Rationale: No

Painter: No theory

 Graves 2018 [38] US

x-ray, CT or MRI

ITS

No intervention

Components:

a) Policy change

Provider:

a) Government

Rationale: Yes, government introduced a utilisation review policy

Painter: No theory

 Hollingworth 2002 [40] UK

LBP radiology

ITS

No intervention

Components:

a) Educational materials (guidelines)

Provider:

a) Government

Rationale: No

Painter: No theory

 Ip 2014 [41] US

Lumbar spine MRI

ITS

No intervention

Components:

a) Decision support (provided for every lumbar MRI request)

b) Audit and feedback (provided quarterly)

c) Soft stop (peer to peer consultation) (provided every time CDS recommendation to not image was ignored)

Provider:

a) Computerised physician order entry

b) Study team

c) Radiology dept

Rationale: No

Painter: No theory

 Matowe 2002 [50] UK

x-ray

ITS

No intervention

Components:

a) Education materials (guidelines) (provided once)

Provider:

a) Study team

Rationale: Yes, they state that passive dissemination of guidelines has been effective in reducing referrals from primary care and that it would be highly cost-effective (with references).

Painter: No theory

 Robling 2002a [55] UK

MRI

Cluster RCT (two sequential trials)

Guideline only

RCT 1

Components:

a) Change to ordering method (provided throughout study duration)

Provider:

a) Radiology department

Rationale: Previous literature on referral method; they state that a multifaceted approach to education may be the most effective

Painter: No theory

 Robling 2002b [55] UK

MRI

Cluster RCT (two sequential trials)

Guideline only

RCT 2

Components:

a) Educational materials (guidelines) (provided once) plus either

b) Further education (practice-based seminar) (provided once) or

c) Audit and feedback (provided once) or

d) Both b and c or

e) Nothing (apart from the guidelines)

Provider:

a) Study team

b) Academic GP and researcher;

c) Not specified

d) N/A

e) N/A

Rationale: Previous literature on referral method; they state that a multifaceted approach to education may be the most effective

Painter: No theory

 Eccles 2001 [32] UK

x-ray

Cluster RCT

Guideline Only

Components:

a) Educational materials (guidelines) (provided once) plus either

b) Additional education (educational messages) (provided twice) or

c) Audit and feedback (provided for every lumbar x-ray)

Provider:

a) GPs and Radiologists;

b) Radiology dept

c) Study research team

Rationale: Yes—previous evidence, saying that’ ‘Oxman and colleagues have reviewed the effectiveness of interventions. Specific prompts at the time of consultation are powerful strategy and have been shown to alter GPs’ behaviour.

Painter: No theory

 Kerry 2000 [45] UK

x-ray

Cluster RCT

No intervention

Components:

Educational materials (guidelines) (provided twice) +

audit and feedback (provided once)

Provider:

a) Study team

b) Study team

Rationale: No

Painter: No theory

 Oakeshott 1994 [53] UK

x-ray

Cluster RCT

No intervention

Components:

Educational materials (posted guidelines)

Provider:

Study team

Dose:

Provided once

Rationale: No

Painter: No theory

 Morgan 2019 [52] Australia

X-ray, CT

ITS

No intervention

Components:

Audit and feedback (given once) + ongoing access to a prescription pad and online decision support tool

Provider:

a) Study team

b) NPS MedicineWise

c) Study team and the George Institute

Rationale: Yes, previous evidence and literature

Painter: No theory

 Zafar 2019 [64] US

MRI

Cluster RCT

Historic control

Components:

a) Audit and feedback (provided every 4-6 months) or

b) Real-time decision support (provided for every lumbar MRI) or

c) Both a and b

Provider:

a) Unclear

b) Computerised physician order entry system

c) As above

Rationale: Yes, previous evidence and literature

Painter: No theory

 Jenkins 2018 [44] Australia

None

Development of intervention, not looking at imaging outcomes

Components:

a) Educational workshop (provided once)

b) Educational materials provided once)

c) Decision support (provided for every LBP patient)

d) Patient education materials (provided for every LBP patient without indication for imaging)

Provider:

a) Study team

b) Study team

c) Study team (used by GP)

d) Study team (given to patients by GP)

Rationale: Yes, previous evidence and literature

Painter: Yes, informed by theory

 Wang 2020 [62] USA

MRI

UBA

No control

Components:

a) Educational presentations (provided once in person and/or virtual)

Provider:

a) Study team

Rationale: No

Painter: No theory

 Fried 2018 [37] USA

MRI

UBA

No control

Components:

a) Included a simple epidemiologic statement in lumbar MRI imaging reports

Provider:

a) Statement developed by study team

Rationale: Yes, previous research

Painter: No theory

 Klein 2000 [46] USA

CT, MRI

UBA

No control

Components:

a) Educational materials (1-page summary guideline was developed to preface a 16-page detailed guideline)

b) Continuing medical educational presentations (provided once, in-person in a large group session or small group session, or via audiotape)

c) Clinical champion

Provider:

a) A multi-disciplinary team of practitioners

b) A multi-disciplinary team of practitioners

b) Rheumatologist

Rationale: No

Painter: No theory

 Powell 2019 [54] USA

CT, MRI

Non-experimental

Not applicable

Components:

a) Decision support—nondenial prior authorisation

b) Peer-to-peer consultation

Provider:

a) Computerised prior authorisation system

b) Board-certified radiologist

Rationale: No

Painter: No theory

 Solberg 2010 [59] USA

MRI

UBA

No control

Components:

a) Decision support (provided for every lumbar MRI request)

Provider:

a) Electronic medical record

Rationale: Yes, literature and previous research

Painter: No theory

 Suman 2018 [60] NL

X-ray, CT, MRI

CBA

No intervention

Components:

a) Multi-disciplinary continuing medical education training in an evidence-based guideline for LBP was developed in the Netherlands in 2010 including online and offline supplemental educational materials

Provider:

a) Study team

Rationale: Yes, literature and previous research

Painter: No theory

 Chen 2020 [30] USA

X-ray, CT, MRI

UBA

No control

Components:

a) Decision support (provided for every lumbar MRI request)

Provider:

a) Electronic medical record

Rationale: Yes, literature and previous research

Painter: No theory

 Simula 2019 [57] FIN

Unknown

Intervention development

Components:

a) Patient education materials (intended for LBP patients without indication for imaging)

Provider:

a) Study team (to be given to patients by GP)

Rationale: Yes, previous evidence and literature

Painter: Yes, informed by theory

 Slater 2014 [58] AUS

Unknown

Prospective cohort

Non-experimental

Components:

a) Clinical education programme (based on national and international clinical practice guidelines) consisting of 5 modules and detailed case studies/patient vignettes

Provider:

a) Study team (interprofessional team)

Rationale: No

Painter: No theory

 Freeborn 1997 [35] USA

X-ray, CT, MRI

CBA

No intervention

Components:

a) Education (disseminate clinical practice guidelines)

b) Decision support aid (flow diagram)

b) Feedback on performance

Provider:

a) Team of specialist colleagues

b) Study team

c) Study team

Rationale: No

Painter: No theory

 Jarvik 2015 [43] US

Spine related relative value units

Stepped-wedge cluster RCT

No intervention

Components:

a) Passive education (provided throughout study duration)

Provider:

a) Study team (through EHR systems)

Rationale: Yes. Previous literature and pilot study

Painter: No theory

 Min 2017 [51] Canada

x-ray, CT or MRI

UBA

No control

Components:

a) Decision support (provided for every Lumbar image request)

b) Educational materials (provided once)

c) Patient materials (provided for every LBP patient not imaged)

Provider:

a) Computerised physician order entry system

b) Study team

c) Provided by a physician (developed by study team)

Rationale: Yes, they cite that previous evidence has shown CDS to be effective in outpatients to modify clinician behaviour, but efficacy in ED is yet to be established

Painter: No theory

 Chandra 2017 [29] Canada

x-ray

UBA

No control

Components:

a) Educational seminar (provided once)

b) Educational videocast (provided once)

c) Educational materials (posters) (provided twice)

Provider:

a), b), c) and d) Study team

Rationale: No

Painter: No theory

 Tracey 1994 [61] UK

x-ray

UBA

No control

Components:

a) Audit and feedback (provided once)

b) Educational materials (guidelines) (provided once)

c) Educational seminar (provided once)

d) Change to ordering process (provided every time a lumbar image was ordered)

Provider:

a), b) and c) Study team

d) Radiology dept

Rationale: Yes, to develop and introduce their own more detailed guidelines since compliance is more likely when staff are responsible for their development and introduction

Painter: No theory

 Baker 1987 [28] US

x-ray

ITS

No intervention

Components:

a) Change to order form (provided every time a lumbar image was ordered)

Provider:

a) Study team

Rationale: No

Painter: No theory

 Dey 2004 [31] UK

x-ray

Cluster RCT

No intervention

Components:

a) Educational outreach (provided once)

b) Ongoing access to fast track physio and back clinic

Provider:

a) Study team

b) Usual clinical teams

Rationale: Yes. Previous Cochrane review suggests that educational outreach may be effective for modifying professional behaviour. Fast track triage and physio was offered to avoid referral to secondary care with out of date views on LBP management.

Painter: No theory (in the discussion, they state that the educational outreach was based on theoretical models and give a reference. The reference refers to the ‘elaboration likelihood model of persuasion’). There is no other mention of theory and no measures of constructs.

 Schectman 2003 [56] US

x-ray, CT or MRI

Cluster RCT

No intervention

Components:

a) Education session (90 min, provided once)

b) Patient materials (provided for every LBP patient)

c) Audit and feedback (provided twice)

d) Written reminders (provided twice)

Provider:

a) Recognised clinical leaders at each of the respective institutions

b) Developed by physicians, a health services researcher, and an expert in patient education materials

c) Study team

d) Study team

Rationale: No

Painter: No theory

 Haig 2019 [39] US

CT, MRI

UBA

No control

Components:

a) Education (email, provided weekly)

b) Questionnaire (ongoing access)

c) Order sheet (ongoing access)

d) Ongoing access to fast track access to psychiatry and physical therapy

Provider:

a) Principal investigator at site

b) ED triage staff

c) ED triage staff

d) N/A

Rationale: Yes. Previous literature and evidence

Painter: No theory

 Burggraf 2019 [65] Germany

x-ray, MRI, CT

Cluster RCT

No intervention

Components:

a) Educational workshop (one day, provided once)

b) Ongoing access to online educational materials

Provider:

a) Study team

b) Study team

Rationale: Yes. Previous evidence and literature

Painter: No theory

 Machado 2018 [49] Australia

x-ray, MRI, CT

Stepped-wedge cluster RCT

No intervention (all clusters will ultimately end up with the intervention)

Components:

a) Education session (40–60 min, provided once with some booster sessions)

b) Educational outreach (provided once)

b) Audit and feedback (provided monthly)

c) Patient resources (provided for every LBP patient)

Provider:

a) Local opinion leaders

b) Clinical educators

c) Study team

d) The Agency for Clinical Innovation

Rationale: Yes. Large deviations in Australian EDs make them ideal to trial a new model of LBP. The ACI model of care was jointly developed by policy makers, clinicians, consumers and researchers and translated high-quality evidence into key practice messages.

Painter: Yes, informed by theory (the KTA framework)