Study reference

Study characteristics

Patient characteristics

Intervention (I)

Comparison / control (C)

Follow-up

Outcome measures and effect size

Comments

Cherkin, 2018

Matching Appropriate Treatments to Consumer

Healthcare needs (MATCH) trial

Type of study:

RCT [pragmatic

cluster randomized trial]

Setting and country:

Six primary care clinics were randomized to intervention or control; The trial was conducted in Group Health (GH), an integrated healthcare delivery system in Washington State serving over 600,000 members; Washington, USA

Funding and conflicts of interest:

Inclusion criteria:

• 18+ years of age
• identified in

electronic health records as having received primary diagnosis consistent with non-specific LBP (e.g., lumbago, back pain not otherwise specified)

Exclusion criteria:

• specific causes of pain (e.g., pregnancy, disc herniation, vertebral fracture, spinal

stenosis)

• job injuries

N total at baseline:

Intervention: 945

Control: 756

Important prognostic factors²:

Sex (female, No. (%))

I: 512 (54.2)

C: 441 (58.3)

Age (mean (SD), y

I: 55 (17.3)

C: 58 (18.4)

Back-related function (RMDQ;

0–23 scales) mean (SD); higher score indicates greater dysfunction

I: 11.8 (6.3)

C: 11.8 (6.1)

Back pain severity; (0–10 scales;

mean (SD)); higher score indicates greater pain severity 

I: 5.4 (2.5)

C: 5.5 (2.5)

STarT back risk group

Low (No. (%))

I: 392 (41.5)

C: 305 (40.3)

Medium (No. (%))

I: 348 (36.8)

C: 286 (37.8)

High (No. (%))

I: 205 (21.7)

C: 165 (21.8)

Groups comparable at baseline.

Risk-stratification strategy

(STarT back tool)

• Incorporating the original STarT back tool in the electronic health records,
• identifying recommended treatment options available from GH for patients in each risk subgroup, and
• training the primary care teams and physical therapists.
• After a patient’s responses had been entered, the EHR automatically calculated the patient’s risk stratum and displayed the recommended treatment

options for that stratum on a screen visible to both

clinicians and patients.

• Clinician could then discuss the treatment options with the patients.
• To promote routine use

of the STarT Back Tool, we let clinics and clinicians decide how to most consistently and efficiently collect and enter the STarT Back Tool data into the EHR. Although the process varied, most clinicians had their medical assistants

collect the data either on a paper copy and then

enter it into the EHR, or enter it directly into the EHR.

Development and implementation (training of primary care providers en physical therapists) are describe in the protocol paper (Cherkin, 2016)

Care as usual

Control clinicals received no training.

Length of follow-up:

6 months;

“Adults receiving primary care for nonspecific LBP were invited to provide data 2 weeks after their primary care visit and follow-up data 2 and 6months (primary endpoint) later.”

Loss-to-follow-up:

Variable per outcome; numbers included in analysis are provided per result

Treatment effect

Back-related physical function (RMDQ); difference at 6M FU (95% CI)

I (n=245):  −0.05 (−0.83, 0.73)

C (n=297): −0.55 (−1.26, 0.15)

Diff: 0.50 (−0.55, 1.55), p 0.35

  Low risk:

  I (n=109):  −0.01 (−0.94, 0.93)

  C (n=122): −0.56 (−1.42, 0.29)

Diff 0.56 (− 0.71, 1.82), p 0.39

  Moderate risk:

  I (n=88): −0.43 (−1.82, 0.96)

  C (n=108): −0.36 (− 1.61, 0.89)

Diff −0.06 (−1.94, 1.81), p 0.95

  High risk:

  I (n=48): 0.69 (−1.46, 2.84)

  C (n=67): −1.07 (−2.92, 0.79)

Diff 1.76 (−1.10, 4.62), p 0.23

Pain severity

I (n=297): −0.19 (−0.56, 0.18)

C (n=245): −0.32 (−0.66, 0.01)

Diff 0.13 (−0.37, 0.63), p 0.61

  Low risk:

  I (n=109): −0.15 (−0.66, 0.36)

  C (n=122): −0.12 (−0.59, 0.35)

Diff −0.03 (−0.72, 0.66), p 0.93

  Moderate risk:

  I (n=88): −0.33 (−0.97, 0.32)

  C (n=108): −0.34 (−0.91, 0.23)

Diff 0.02 (−0.85, 0.88), p 0.97

  High risk:

  I (n=48): 0.06 (−0.89, 1.01)

  C (n=67): −0.8 (−1.62, 0.03)

Diff 0.86 (−0.41, 2.12), p 0.18

Quality of life

Not reported

Patient satisfaction, 6M;

OR pre-post (95% CI) OR int-contr (95% CI)

Very satisfied with care

I: 1.01 (0.63, 1.61)

C: 1.27 (0.83, 1.94)

OR 0.79 (0.42, 1.49), p 0.471

Very satisfied with treatment

I:  1.15 (0.68, 1.95)

C: 0.91 (0.57, 1.44)

OR 1.26 (0.63, 2.54), p 0.516

Very satisfied with information about cause of pain

I:  1.29 (0.76, 2.20)

C: 1.52 (0.94, 2.46)

OR 0.85 (0.41, 1.740), p 0.65

Duration of treatment

not reported

Cost-effectiveness

Not reported

Authors conclusion:

A resource-intensive intervention to support stratified care for LBP in a US healthcare setting had no effect on patient outcomes or healthcare use.

Foster, 2020

The SCOPiC RCT

Type of study:

RCT

Setting and country:

general practices in three

areas: (1) North Staffordshire, (2) North Shropshire/Wales and (3) Cheshire.

Funding and conflicts of interest:

Inclusion & exclusion criteria:

See Konstantinou, 2020

N total at baseline: 476

Intervention: 238

Control: 238

Important prognostic factors²:

STarT Back Screening Tool risk subgroup, n (%)

Low

I: 56 (23.6)

C: 52 (21.9)

Medium

I: 133 (56.1)

C: 130 (54.9)

High

I: 45 (20.3)

C: 55 (23.2)

Groups comparable at baseline/

Stratified care.

See Konstantinou, 2020

Usual non-stratified care

See Konstantinou, 2020

See Konstantinou, 2020

See Konstantinou, 2020

Cost-effectiveness

Mean costs, £, mean per-participant sciatica-related resource use and costs over 12 months, by intervention arm

=NHS perspective

I: 743.23 (764.05)               

C: 665.55 (978.15)  

MD: 77.67 (–70.49 tot 232.94)

Mean QALYs

I: 0.6599 (0.1731)

C: 0.6713 (0.1685)

MD –0.011 (–0.035 to 0.013)

Definitions:

Remarks:

Authors conclusion:

The stratified care model tested in this trial was not more clinically effective than usual care, and was not likely to be a cost-effective option. The fast-track pathway was felt to be acceptable to both patients and clinicians; however, clinicians expressed reluctance to consider invasive procedures if symptoms were of short duration.

Konstantinou, 2020

The SCOPiC RCT

Type of study:

RCT;

Setting and country:

42 general practices surrounding three centres (North Staffordshire, North Shropshire/ Wales, and Cheshire); enrolment May 28, 2015, and July 18, 2017; UK

Funding and conflicts of interest:

Funding: National Institute for Health Research. Interests reported in paper.

Inclusion criteria:

• aged 18 years or older
• clinical diagnosis of sciatica,
• access to a mobile phone or landline number
• not pregnant,
• not currently

receiving treatment for the same problem

• no previous spinal surgery.

Exclusion criteria:

• clinical suspicion (by their GP or assessing physiotherapist) of serious spinal pathology (eg, cauda equina syndrome, fracture, spondyloarthropathy, malignancy, infection, or foot drop)
• serious physical

or mental co-morbidities

N total at baseline: 476

Intervention: 238

Control: 238

Important prognostic factors²:

age mean (SD):

I: 50.7 (14.5) y

C: 53.3 (13.5) y

Sex:

I: 106/237 (45%)

C: 108/238 (45%)

Symptom duration

<2 weeks

  I: 15 (6%)

  C: 33 (14%)

2–6 weeks

  I: 99 (42%)

  C: 98 (41%)

6–12 weeks

  I: 58 (24%)

  C: 46 (19%)

3–6 months

  I: 31 (13%)

  C: 29 (12%)

6–12 months

  I: 10 (4%)

  C: 10 (4%)

>12 months

  I: 24 (10%)

  C: 22 (9%)

Physical function (RMDQ 0–23), mean (SD)

I: 11.1 (5.2)

C: 11.3 (5.4)

Groups comparable at baseline.

Stratified care

In the stratified care arm, the stratification algorithm

(figure 1) was used to allocate patients to one of three groups.

STarT Back tool score 3 or lower à group 1

STarT Back tool score

higher than 3: assess STarT Back tool psychological sub score (0-5):

• 3 or lower à

assess clinical characteristics (0-4; current leg pain (NRS >6 [0–10]), pain below knee, interference with work and home activities (NRS >6 [0–10]), objective sensory deficit);

3 or lower à group 2;

all 4 à group 3

• STarT Back tool psychological sub score 4 or higher: assess clinical characteristics (see above);
• 2 or lower à group 2
• 3 or higher à group 3

Group 1: brief self-management support (up to 2 sessions)

Group 2: physiotherapy course (up to 6 sessions)

Group 3: referral to MRI and specialist

spinal services

Physiotherapists treating patients in groups 1 and 2 were responsible for providing good clinical governance and could overrule the stratification

algorithm recommendation for matched care pathways if they thought it clinically appropriate.

Care as usual

All participants in

the usual care arm had a consultation with a physiotherapist at the SCOPiC research clinic, their care was planned without the use of any stratification tools, and referrals for further physiotherapy or to other services were made at the discretion of the assessing physiotherapist

and in consultation with the patient.

Length of follow-up:

12 months (also 4 M reported)

Response rate:

Primary outcome:

I: 4688 [88%] of 5310

C: 4779 [90%] of 5291

Overall:

I: 81%

C:  84%

12-month questionnaire:

I: 74%

C:  76%

Loss-to-follow-up:

Intervention:

N (%)

Reasons (describe)

Control:

N (%)

Reasons (describe)

Incomplete outcome data:

Intervention:

N (%)

Reasons (describe)

Control:

N (%)

Reasons (describe)

Treatment effect

Physical function (RMDQ, 0–23)

4 months

I:  n=192, 6.5 (6.3)

C: n=201, 6.2 (6.0)

MD 0.43 (–0.69 to 1.54) p=0.45

12 months

I: n=177, 5.0 (6.2)

C: n=182, 5.5 (6.0)

MD –0.53 (–1.84 to 0.78) p=0.43

Time to resolution of sciatica symptoms

Adjusted HR=1.14 (95% CI 0.89–1.46)

Cumulative proportion of resolved cases by week 48:

All: 0.754 (all)

I: 0.780

C: 0.729

Also reported:

Global perceived change, usual backpain (NRS), usual leg pain (NRS), SBI, S-LANSS, fear of movement

Quality of life

Health outcomes

I: n=238

C: n=238

Difference (CI)(SC-UC)

Base-case (Imputed) analysis

Baseline EQ-5D

I: 0.5168 (0.2109)

C: 0.5201 (0.2300)

MD -0.003 (-0.043, 0.037)

4 month EQ-5D

I: 0.6710 (0.2006)

C: 0.6806 (0.1955)

MD -0.009 (-0.046, 0.026)

12 month EQ-5D

I: 0.7138 (0.2137)

C: 0.7320 (0.1879)

MD -0.018 (-0.054, 0.0176)

Unadjusted QALYs

I: 0.6599 (0.1731)

C: 0.6713 (0.1685)

MD -0.011 (-0.042, 0.019)

Patient satisfaction

Satisfaction with care

At 4 months

Very satisfied

I: 65/153 (43), C: 57/158 (36)

Quite satisfied

I: 42/153 (27), C: 48/158 (30) No opinion

I: 22/153 (14), C: 27/158 (17) Not satisfied

I: 24/153 (16), C: 26/158 (16)

P = 0.210

12 months

Very satisfied

I: 42/123 (34), C: 50/132 (38)

Quite satisfied

I: 32/123 (26), C: 36/132 (27)

No opinion

I: 32/123 (26), C: 28/132 (21)

Not satisfied

I: 17/123 (14), C: 18/132 (14)

P = 0.873

Satisfaction with results of care

At 4 months

Very satisfied

I: 62/153 (41), C: 42/158 (27)

Quite satisfied

I: 43/153 (28), C: 58/158 (37) No opinion

I: 20/153 (13), C: 30/158 (19) Not satisfied

I: 27/153 (18), C: 28/158 (18)

P = 0.030

12 months

Very satisfied

I: 37/123 (30), C: 49/132 (37)

Quite satisfied

I: 36/123 (30), C: 37/132 (28)

No opinion

I: 33/123 (27), C: 27/132 (20)

Not satisfied

I: 16/123 (13), C: 19/132 (14)

P = 0.935

Duration of treatment

Not reported

Cost-effectiveness

Mean costs, £, mean per-participant sciatica-related resource use and costs over 12 months, by intervention arm

=NHS perspective

I: 663.58 (737.14)

C: 617.37 (935.50)

MD 46.21 (–110.60 to 187.06)

Mean QALYs

I: 0.6599 (0.1731)

C: 0.6713 (0.1685)

MD –0.011* (–0.035 to 0.013)

Authors conclusion:

The stratified care model for patients with sciatica consulting in primary care was not better than usual care for either clinical or health economic outcomes. These results do not support a transition to this stratified care model for patients with sciatica.

Hill, 2011

STarT Back

Trial Study Protocol

Type of study:

RCT

Setting and country:

10 general practices, England

Funding and conflicts of interest: funded by Arthritis Research UK; “We declare that we have no conflicts of interest.”

Inclusion criteria:

• ≥ 18 years of age
• seek care for LBP with or without associated leg pain
• speak and understand English
• be willing and able to give full,

informed written consent.

Exclusion criteria:

• potentially serious pathology (e.g. cauda equina compression,

inflammatory arthritis, malignancy etc), serious comorbidity,

psychiatric illness or personality disorder;

spinal surgery in the last 6 months;

• pregnancy;
• already receiving

treatment other than GP care for this episode of back pain;

• inability to attend regular physiotherapy

appointments

N total at baseline: 851

Intervention: 568

Control: 283

Important prognostic factors²:

age (SD):

I: 50.1 (15.0)

C: 49.1 (14.3)

Sex:

I: 330 (58%) F

C: 170 (60%) F

Back pain and function

RMDQ disability score

I: 9.8 (5.6)

C: 9.7 (5.8)

Back pain intensity I: 5.3 (2.2)

C: 5.2 (2.2)

Groups comparable at baseline.

Describe intervention (treatment/procedure/test):

• initial 30-minute protocolised intervention with specially trained study physiotherapist: reassurance about the benign nature of their pain, and simple messages of advice (e.g. pain relief, appropriate activity levels, role of further investigations)
• the physiotherapist is guided by the results of the patient's STarT Back tool score so that specific concerns can be identified and addressed on an individual basis.
• 'Back Book' [35] is given, together with information sheet of local contacts for exercise venues such as swimming pools, gyms and exercise on prescription, and self-help groups.
• 15-minute educational video based on The Back Book, called 'Get Back Active' [36] is shown

Low risk:

• receive no further treatment

Medium & high risk:

• Treatment according to STarT Back assessment and treatment algorithms and STarT Back physiotherapy manual.
• Evidence-based physiotherapy techniques addressing 'signs and symptoms' in non-specific LBP presentations, according to current

recommendations from guidelines and high quality clinical trials.

• Evidence-based advice, education and reassurance for symptoms and management, according to guidelines and current literature within the

biopsychosocial model.

Medium risk [predominantly physical prognostic indicators, without high levels of psychosocial distress]:

• target physical characteristics using a range of physiotherapy techniques delivered over 6 x 30-minute sessions.

High risk [high psychosocial]:  

• Assessing and addressing biopsychosocial risk factors by adopting cognitive behavioural principles to address unhelpful beliefs and behaviours.

All patients are advised that they can access their GP for

ongoing care in the usual way, and that they should contact their GP if their condition worsens.

Best current care

• 30-minute consultation with physiotherapist, who is blinded to STarT Back tool results.
• A management plan is formulated according to normal clinical practice.
• If thought necessary, this may include referral to community physiotherapy

All patients are advised that they can access their GP for

ongoing care in the usual way, and that they should contact their GP if their condition worsens.

Length of follow-up:

4 and 12 months

Loss-to-follow-up & incomplete outcome data:

Intervention: 128/568 (22.5%)

Reasons: withdrawal (n=15; did not respond to questionnaire (n=113)

Control: 74/283 (26.1%)

Reasons withdrawal (n=8; did not respond to questionnaire (n=66)

Treatment effect (effect on pain, but also adherence / drop-out rates)

RMDQ score; mean change

4 months

  I: 4.7 (5.9)

  C: 3.0 (5.9)

  OR 1.85 (1.36 to 2.51)

  12 months

  I:  4.3 (6.4)

  C: 3.3 (6.2)

  OR 1.48 (1.02 to 2.15)

Results also provided per sub groups (risk at baseline)

Back pain intensity, scale Mean change (SD);

4 months,

I: 3.2 (2.5)

C: 2.6 (2.4)

MD 0.55 (0.23 to 0.86)

12 months,

I: 3.0 (2.8)

C: 2.8 (2.6)

MD 0.11 (–0.21 to 0.43)

StarT Back risk group

4 months

Low: I 77%, C 66%

Medium: I 18%, C 25%

High: I 5%, C 10%

OR 2.09 (1.40 to 3.12)

12 months

Low: I 76%, C 69%

Medium: I 20%, C 25%

High: I 5%, C 6%

OR 1.51 (1.01 to 2.25)

Also reported for patients at low/medium/high risk at baseline

Also reported: global change in back pain.

Quality of life

SF12, physical component, mean change (SD)

4 months

I: –7.5 (13.0)

C: –5.2 (13.3)

MD –2.83* (–4.62 to –1.05)

P= 0.0010

12 months

I:  –7.5 (11.3)

C: –5.2 (10.9)

MD –2.93 (–4.31 to –1.56)

P <0.0001

SF12, mental component

4 months

I:  –2.1 (11.3)

C: –2.1 (11.0)

MD –0.16 (–1.63 to 1.32)

P=0.8327

12 months

I: –1.7 (13.0)

C: –1.2 (13.4)

MD –0.69* (–2.39 to 1.01)

P= 0.4050

Patient satisfaction

Satisfaction with care¶

Very satisfied

I: 166/416 (40%)

C: 46/197 (23%)

Quite satisfied

I: 139/416 (33%)

C: 81/197 (41%)

Not satisfied

I: 111/416 (27%)

C: 70/197 (36%)

OR 1.98† (1.45 to 2.70)

Duration of treatment

Not reported

Cost-effectiveness

Primary (imputed) analysis:

I: n=568

C: n=283

Mean costs (SD), £             

I: 240.01 (356.3)

C: 274.40 (350.8)               

MD -34.39 (-80.3 tot 11.5)

QALYs over 12 months (estimates following multiple regression-based adjustment for age, gender, duration of pain at baseline, and baseline scores on the RMDQ and EQ-5D.)

MD 0.039 (0.01 to 0.07)

p-value = 0.01 (2 d.p.)

Definitions:

* - Mean for the intervention group minus mean for the control group (by use of linear regression adjusted for

age, sex, baseline RMDQ, and duration of back pain).

† - Based on observed or available case data.

‡ - Based on imputed datasets with adjustment for clustering by therapist (by use of random-effects linear

regression modelling).

Whitehurst, 2012

Sub-study of Hill et al. 2011

STarT Back

Trial Study Protocol

Type of study:

Within trial study; additional to Hill (2011) RCT; cost analysis

Setting and country;

See Hill (2011); UK

Funding and conflicts of interest:

“Funding: Arthritis Research UK”

“Competing interests None.”

Inclusion & exclusion criteria and baseline characteristics: See Hill 2011

See Hill 2011

See Hill 2011

Length of follow-up:

12 months

Loss-to-follow-up:

None, data was imputed for all 851 participants

Incomplete outcome data:

“Cost estimates were derived for 567 (67%) responders at

12 months. For the EQ-5D, complete data were provided by 458 (54%) participants; time-point specific response rates were 99% (baseline), 73% (4 months) and 62% (12 months).”

Following the multiple imputation procedure, all 851 participants were included in the base-case analysis.

Treatment effect

Not reported

Quality of life

Low-risk; I: n=148, C: n=3

  Baseline EQ-5D

  I: 0.725 (0.19)

  C: 0.733 (0.15)

  MD −0.008 (−0.05 to 0.04)

4-month EQ-5D

  I: 0.799 (0.21)

  C: 0.821 (0.18)

  MD −0.022 (−0.07 to 0.03)

12-month EQ-5D

  I: 0.787 (0.20)

  C: 0.773 (0.24)

  MD 0.014 (−0.05 to 0.08)

Medium-risk; I: n=263, C:n=131

Baseline EQ-5D

  I: 0.540 (0.27)

  C: 0.573 (0.27)

  MD −0.033 (−0.12 to 0.06)

4-month EQ-5D

  I: 0.702 (0.28)

  C: 0.674 (0.28)

  MD 0.028 (−0.03 to 0.09)

12-month EQ-5D

  I: 0.687 (0.32)

  C: 0.635 (0.31)

  MD 0.052 (−0.01 to 0.12)

High-risk; I: n=157, C: n=79

Baseline EQ-5D

  I: 0.325 (0.33),

  C: 0.252 (0.35)

  MD 0.073 (−0.02 to 0.16)

4-month EQ-5D

  I: 0.585 (0.35)

  C: 0.474 (0.38)

  MD 0.111 (0.02 to 0.20)

12-month EQ-5D

  I: 0.541 (0.37)

  C: 0.458 (0.38)

  MD 0.083 (−0.01 to 0.18)

Patient satisfaction

Not reported

Duration of treatment

Not reported

Cost-effectiveness

Estimates for the base-case (imputed) analysis

Total healthcare cost:

Low-risk group (n=221)

   I: 96.15 (191.6)

   C: 160.44 (279.8)

MD (95% CI; p value)* −64.29 (−132.0 to 3.5; 0.06)

Medium-risk group (n=394)

   I: 235.08 (304.0)

   C: 287.8 (338.6)

MD (95% CI; p value)* −52.72 (−119.0 to 13.6; 0.12)

High-risk group (n=236)

   I: 383.88 (456.3)

   C: 357.47 (585.7)

MD (95% CI; p value)* 26.41 (−112.6 to 165.5; 0.71)

Descriptive and incremental health outcomes, over 12 months for the base-case and complete case analyses.

Low-risk; I: n=148, C: n=3

QALYs over 12 months‡

MD −0.001 (−0.04 to 0.03)

p=0.94 (2 dp)

Medium-risk; I: n=263, C:n=131

QALYs over 12 months‡

MD 0.044 (0.00 to 0.09)

p=0.04 (2 dp)

High-risk; I: n=157, C: n=79

QALYs over 12 months‡

MD 0.057 (−0.01 to 0.12)

p=0.08 (2 dp)

Definitions:

‡ Incremental QALY estimates following multiple regression-based adjustment for age, gender, duration of pain at baseline, and baseline scores on the RMDQ and EQ-5D. For the complete case analysis, only the incremental QALY estimate is provided.

Authors conclusion:

Compared with current best practice, stratified primary care management for low back pain provides a highly cost-effective use of resources across all risk-defi ned subgroups

Hill, 2020

Type of study:

RCT

Setting and country:

8 general practices,

West Midlands region of England; recruitment Oct

2016 to May 2017.

Funding and conflicts of interest:

funded by the National Institute for Health

Research (NIHR) under its Programme Grants for Applied Research scheme

(grant number: RP-PG-1211-20010) as well as Centre of Excellence funding

from Versus Arthritis (grant reference: 20202);

Competing interests:

None declared.

Inclusion criteria:

• aged over 18 years
• registered at that general practice,
• consulting for MSK pain in the back, neck, shoulder, knee or multi-site pain

Exclusion criteria:

• clinical indicators

of (suspected) serious ‘red flag’ pathology requiring urgent medical intervention

• known systemic inflammatory

condition

• unable to communicate in

English (reading / speaking)

• vulnerable patients including those on the ‘severe and enduring mental health register’
• diagnosis of dementia or

terminal illness,

• recent trauma or bereavement

N total at baseline: 524

Intervention: 231

Control: 293

Important prognostic factors²:

age (SD):

I: 60.3 (15.1)

C:61.8 (14.5)

Sex, male:

I: 98/231 (42%)

C: 108/293 (37%)

No. of previous MSK pain episodes, median (IQR)

I: 5 (1–15) [n=186]

C: 5 (1–30) [n=229]

MSK Risk status (Keele development version of the STarT MSK Tool – note it was not the final version), mean (SD) [n = 482]

Low risk (0–3 score), n (%)

I: 67 (30.9%)

C: 88 (33.2%)

Medium risk (4–7 score), n (%)

I: 119 (54.8%)

C: 144 (54.3%)

High risk (8–9 score), n (%)

I: 31 (14.3%)

C: 33 (12.5%)

Groups comparable at baseline.

Stratified care

GPs were supported to deliver stratified care, by extending the trial point-of-consultation identification EMR with a prognostic stratification tool (a development version of the Keele STarT MSK tool).

Based on the location of the pain and the risk class, matched treatment options were recommended.

• Low: restricted to supporting self-management and over-the-counter medication, discouraging unnecessary investigations or referral.
• Medium: referral to conservative non-pharmacological treatments (e.g. those offered by physio-therapists) and workplace assessment and advice
• High: referral for corticosteroid injections specialist clinical services (including rheumatology, orthopaedics and pain clinics), and opioids.

GPs received training by an experienced trainer and also received a training-update half-way through their recruitment period (feedback data, peer-to-peer comparisons, discussion)

Usual care

“Usual primary care is known to be variable [43–45]; for example,

some patients may receive advice, prescriptions for medications and nothing more, some may be asked to return to the GP for follow-up assessment or treatment, whereas others may be referred to other services, including for tests and investigations, or treatment services such as physiotherapy, orthopaedics or pain clinics. As part of the trial’s participant identification screen, GPs in control

(and intervention) practices recorded patient’s average

MSK pain intensity (see outcomes section) and primary MSK pain site at the point-of-consultation on the study EMR identification template.”

Length of follow-up:

6 months

Loss-to-follow-up

I: 5 withdrawals (2.2%)

C: 10 withdrawals (3.4%)

Incomplete outcome data:

I: 21/231 (9.1%)

C: 24/293 (8.2%)

N=

I: 200

C:  258

Treatment effect

Overall musculoskeletal health status (MSK-HQ, 0–56, the higher score the better), mean (SD)

I: 37.5 (12.8) [n = 193]

C: 37.3 (11.8) [n = 248]

Knee physical function using KOOS (the higher score the better), mean (SD)

I: 51.7 (24.5) [n = 55]

C: 53.6 (22.9) [n = 72]

Neck physical function using NDI (the higher score the worse), mean (SD)

I: 7.80 (5.83) [n = 27]

C: 11.89 (11.57) [n = 24]

Back physical function using RMDQ (the higher score the worse), mean (SD)

I: 6.90 (6.52) [n = 61]

C: 6.44 (5.80) [n = 75]

Shoulder physical function using SPADI-Function

(the higher score the worse), mean (SD)

I: 30.2 (29.6) [n = 44]

C: 33.4 (27.8) [n = 62]

Quality of life

Quality of life using EQ5D-5 L, mean (SD)

I: 0.65 (0.26) [n = 208]

C: 0.63 (0.25) [n = 258]

Patient satisfaction

Satisfaction with GP care for MSK pain

Very satisfied, n (%)

I: 48 (24.2%); C: 58 (22.8%)

Quite satisfied, n (%)

I: 71 (35.9%); C: 89 (34.9%)

No opinion, n (%)

I: 46 (23.2%); C: 60 (23.5%)

Not very satisfied, n (%)

I: 25 (12.6%); C: 44 (17.3%)

Not at all satisfied, n (%)

I: 8 (4.0%); C: 4 (1.6%)

Duration of treatment

Not reported

Cost-effectiveness

Not reported

Authors conclusion:

This feasibility and pilot trial has successfully demonstrated

the feasibility of the cluster RCT design with high retention rates over 6 months (> 90%) and little evidence of selection bias, although changes to the

main trial sample size were required due to a slower

than expected recruitment rate. GP point-of-consultation

fidelity to the stratified care intervention was mixed with GPs using the tool less often than expected (only when they coded consultations, when they had time and when MSK pain was the primary

reason for the visit). However, there was high fidelity to choosing recommended matched treatment options

(> 80% of cases). The learning from this feasibility and pilot RCT has led to a number of important changes prior to the main STarT MSK trial testing the clinical and cost-effectiveness of stratified primary care for patients with MSK pain.

Jamal, 2019

Type of study:

RCT; cluster

Setting and country:

140 GPs (93 general practices) out of 1145 invited GPs; Recruitment 10 Sept 2014 and 6 Nov 2015; follow-up

period 10 Jan 2015 and 6 March 2016; widely spread over the country, the Netherlands

Funding and conflicts of interest:

‘The author(s) received no specific

funding for this work.’; ‘The authors have declared

that no competing interests exist.’

Inclusion criteria:

• age 18–45 years
• current low back pain (LBP) for more than 12

weeks

registered by means of the ICPC code L03 (LBP without radiation)

Exclusion criteria:

• clear medical explanation for the back pain (e.g. trauma, hernia nuclei pulposi),
• being mentally incompetent
• having insufficient understanding of the Dutch language (written).

N total at baseline: 679

Intervention: 333

Control: 346

Important prognostic factors²:

age ± SD:

I: 36.7 ± 7.1 y

C: 35.8 ± 7.8 y

Sex, male:

I: 115/333 (35%)

C: 130/346 (38%)

chronic low back pain duration, year median (IQR)

I: 10 (4–15)

C: 9 (4–15)

Groups comparable at baseline.

CaFaSpA referral model

Digital CaFaSpA referral strategy consists of four parameters:

• inflammatory back pain (IBP) (questionnaire)
• positive family history of axSpA (question by GP)
• positive response to treatment with NSAID’s (question by GP)
• duration of back pain for more than 5 years (question by GP)

If at least-two out of four referral items are present, referral to a rheumatologist is advised.

Usual care

GPs took care of their patients as usual.

Due to ethical reasons the

score of the CaFaSpA referral strategy was given to both the GP and patient at the end of the 4

months follow-up period.

Length of follow-up:

4 months

Loss-to-follow-up:

Not described

Missing assessment:

Baseline

I: 63/333 (18.9%)

C: 39/346 (11.3%)

4 months

I: 117/333 (35.1%)

C: 78/346 (22.5%)

Treatment effect

RMDQ, mean SD:

Baseline

I: 8.39 (7.59–9.18)

C: 8.61 (7.83–9.39)

4 months

I: 7.65 (6.79–8.50)

C: 8.15 (7.34–8.96)

Frequency of new axSpA diagnosis

I: 192/333 (58%) received  positive referral advice based on the CaFaSpA rule.

103/192 (54%) visited a rheumatologist (of which 75 (73) were verified by hospital records

In total, n=6 (8%)  received diagnosis axSpA from rheumatologist

Quality of life

Not reported

Patient satisfaction

Not reported

Healthcare utilization

Not reported

Cost-effectiveness

Not reported

Authors conclusion:

The CaFaSpA referral strategy for axSpA did not have an effect on disability after four

months caused by CLBP. However, the strategy is able to detect the axSpA patient within the large CLBP population sufficiently.

Study reference

(first author, publication year)

Describe method of randomisation¹

Bias due to inadequate  concealment of allocation?²

(unlikely/likely/unclear)

Bias due to inadequate blinding of participants to treatment allocation?³

(unlikely/likely/unclear)

Bias due to inadequate blinding of care providers to treatment allocation?³

(unlikely/likely/unclear)

Bias due to inadequate blinding of outcome assessors to treatment allocation?³

(unlikely/likely/unclear)

Bias due to selective outcome reporting on basis of the results?⁴

(unlikely/likely/unclear)

Bias due to loss to follow-up?⁵

(unlikely/likely/unclear)

Bias due to violation of

intention to treat analysis?⁶

(unlikely/likely/unclear)

Cherkin, 2018

Web-based

Prior to the intervention, the trial biostatistician randomly assigned one clinic in each of the three geographic and sociodemographic

matched pairs of clinics to the intervention by

computer-generated random number. All eligible patients seen

in the intervention clinics were considered to have received the

intervention.

Unlikely

Adequate, randomized before participation of patients and providing the intervention.

Unlikely

Researchers did not inform patients that their clinics were participating in an intervention.

Likely

Care providers were aware of the allocation as they were trained or not trained. This might have induced an (unaware) difference in contact with the patient, on top of the difference in provided intervention.

Unlikely

Interviewers were

blinded to patients’ clinics.

Likely

Trial registration present:  NCT02286141. Registered November 5, 2014.

Unclear Satisfaction was measured at a 10-item instrument among three dimensions; in this report only 3 dichotomous outcomes are presented

Likely

The numbers of patients that  consented and provided baseline data do not correspond with patients numbers of the results.

Unclear for which patients the baseline characteristics were provided.

Likely

No intention-to-treat analysis, high rate of missing data

Foster, 2020

& Konstantinou, 2020

SCOPiC trial

Computerised

At the research clinic, eligible patients who gave written informed consent were randomly assigned by a computer-generated

code, to either stratified care or usual care.

Randomisation was done with a web-based service from Keele Clinical Trials Unit, and was stratified by centre and sciatica group allocation (sciatica groups 1, 2, and 3), by

use of random permuted blocks of varying size (2, 4, and 6).

Unlikely

Allocation not known before inclusion; computerized strategy by external service

Likely

Patients were told that the trial was comparing

two approaches for the treatment of sciatica, one based on matching patients to treatment by use of a simple tool that helps to decide on the treatment pathway, and one based on treatment needed as discussed and agreed between the patient and physiotherapist.

Unlikely

To further help with

allocation concealment, participants randomised to the UC arm of the trial were seen by physiotherapists

who had not carried out their detailed clinical assessment and eligibility screen for the trial. Therefore,

physiotherapists in the UC arm remained blinded to both the details of the stratification algorithm and the individual patient’s sciatica group status, hence avoiding contamination between the two trial arms.

Unlikely

Statisticians and outcome assessors were masked to treatment allocation.

Unlikely

Outcomes consistent with protocol; relevant sub group reporting

Likely

% of non-responders s 12% and 10% for the primary outcome and 19% and 16% for the secondary outcomes. This is quite a high percentage and could induce a bias in the results.

Costs and EQ-5D data were imputed to minimise bias.

Likely

Not all patients were included in analysis; up to data available for  75% of the patients for the physical function (RMDQ) at 12 months.

Hill, 2011

&

Whitehurst 2012

Computerized

A clinic administrator telephoned a remote clinical trials unit (Keele University), which assigned participants to

intervention and control groups by use of computer-generated

stratified block randomisation (block sizes of three) in a 2:1 ratio to enable future secondary analysis

of targeted treatment mechanisms.

Unlikely

Administrator randomly assigned the patients; clinicians and research nurse no insight in allocation

Likely

Participants, administrator,

or physiotherapists could not be masked to randomisation

 […], however, we ensured that the therapists administering

the control treatment were not made aware of the details of the stratified model of care during the trial, and

that patients were made aware that they would be treated

according to one of two primary care management models.

Likely

Participants, administrator,

or physiotherapists could not be masked to randomisation

because therapists were administering the active intervention;

however, we ensured that the therapists administering

the control treatment were not made aware of the details of the stratified model of care during the trial, and

that patients were made aware that they would be treated

according to one of two primary care management models.

Unlikely

The research nurse who retrieved outcome data was masked to randomisation, and the concealment strategies

were that the nurse was in a separate office during clinic, the administrator randomly assigned the patients, and the use of a masked follow-up database.

Unlikely

Elaborate protocol; announced outcomes reported; no questionable sub-group analyses

Likely

In total, quite high rate of missing data / withdrawal (I: 22.5%, C: 26.1%), plus slight imbalance in attrition between the groups

Unlikely

Analysis performed according to intention-to-treat analysis; missing data was imputed

Hill, 2020

Computerised

Randomisation used stratified block randomisation

based on GP practice list size to allocate the 8 practices

in a ratio of 1:1 (4 intervention, 4 control). Keele Clinical

Trials Unit (CTU) computer-generated the random sequence

and ensured concealment by providing each

practice with an anonymised code.

Unlikely

Keele Clinical

Trials Unit (CTU) computer-generated the random sequence

and ensured concealment by providing each

practice with an anonymised code.

Unlikely

Blinding for participating GPs was obviously not possible,

however, patients were unaware of the RCT and the differences between consultations in intervention and control practices, and instead were informed about, and consented to, providing questionnaire data for a study investigating the Treatment of Aches and Pains (TAPs)

Likely

Blinding for participating GPs was obviously not possible [..]

Allocation (at cluster

and individual level) was shared with the study team (except for the trial statistician and outcome data collectors who were blinded until the analysis was finalised).

Unlikely

Allocation (at cluster

and individual level) was shared with the study team

(except for the trial statistician and outcome data collectors

who were blinded until the analysis was finalised).

Unlikely

Outcomes consistent with trial, registered before study onset.

Unlikely

Numbers provided per outcome; follow-up rate of > 75% with less than 5% missing items in participant questionnaires

Likely

No intention-to-treat analysis, no data imputed; rate of missing data 9.1% (intervention) and 8.2% (control).

Jamal, 2019

Computerised

The block randomization schedule was computer generated and

conducted by an independent person, who was not involved in patient care. Randomization was stratified for the number of GPs working in the general practices (one or two vs. more

than two) to ensure a similar number of patients in both groups.

Unclear

Concealment of allocation not described and unclear at which moment allocation was known.

Likely

Patients and GPs were

unblinded, because of the nature of the intervention.

Likely

Patients and GPs were

unblinded, because of the nature of the intervention.

Unlikely

Online questionnaire, by patient

Likely

Outcomes not consistent or incomplete compared to protocol paper (effect on quality of life and costs not described)

Likely

Percentage of missing data is relatively high and unequal between groups.

Baseline: I: 18.9%, C: 11.3%; 4 months:  I 35.1%, C: 22.5%.

Likely

No intention-to-treat analysis performed;

linear mixed model allows for patients to have a missing outcome at either baseline visit or after 4 months (but not all (both) visits), but yields unbiased estimates in case the data is missing at random.