Models of care for managing non-specific low back pain

Alternative care models seek to improve the quality or efficiency of care, or both, and thus optimise patient health outcomes. They provide the same health care but change how, when, where, or by whom health care is delivered and co-ordinated. Examples include care delivered via telemedicine versus in-person care or care delivered to groups versus individual patients.

To assess the effects of alternative models of evidenced-based care for people with non-specific low back pain on the quality of care and patient self-reported outcomes and to summarise the availability and principal findings of economic evaluations of these alternative models.

We searched the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, Embase, and trial registries up to 14 June 2024, unrestricted by language.

We included randomised controlled trials comparing alternative care models to usual care or other care models. Eligible trials had to investigate care models that changed at least one domain of the Cochrane EPOC delivery arrangement taxonomy and provide the same care as the comparator arm. Participants were individuals with non-specific low back pain, regardless of symptom duration. Main outcomes were quality of care (referral for/receipt of lumbar spine imaging, prescription/use of opioids, referral to a surgeon/lumbar spine surgery, admission to hospital for back pain), patient health outcomes (pain, back-related function), and adverse events.

Two review authors independently selected studies for inclusion, extracted data, and assessed risk of bias and the certainty of evidence using GRADE. The primary comparison was alternative models of care versus usual care at closest follow-up to 12 months.

Fifty-seven trials (29,578 participants) met our inclusion criteria. Trials were primarily set within primary care (18 trials) or physiotherapy practices (15 trials) in high-income countries (51 trials). Forty-eight trials compared alternative models of care to usual care. There was substantial clinical diversity across alternative care models. Alternative care models most commonly differed from usual care by altering the co-ordination/management of care processes (18 trials), or by utilising information and communication technology (10 trials). Moderate-certainty evidence indicates that alternative care models probably result in little difference in referral for or receipt of any lumbar spine imaging at follow-up closest to 12 months compared to usual care (risk ratio (RR) 0.92, 95% confidence interval (CI) 0.86 to 0.98; I² = 2%; 18 trials, 16,157 participants). In usual care, 232/1000 people received lumbar spine imaging compared to 213/1000 people who received alternative care models. We downgraded the certainty of the evidence by one level due to serious indirectness (diversity in outcome measurement). Moderate-certainty evidence suggests that alternative care models probably result in little or no difference in the prescription or use of opioid medication at follow-up closest to 12 months compared to usual care (RR 0.95, 95% CI 0.89 to 1.03; I² = 0%; 15 trials, 13,185 participants). In usual care, 349 out of 1000 people used opioid medication compared to 332 out of 1000 people in alternative care models. We downgraded the certainty of the evidence by one level due to serious indirectness (diversity in outcome measurement). We are uncertain if alternative care models alter referral for or use of lumbar spine surgery at follow-up closest to 12 months compared to usual care as the certainty of the evidence was very low (odds ratio (OR) 1.04, 95% CI 0.79 to 1.37; I² = 0%; 10 trials, 4189 participants). We downgraded the certainty of the evidence by three levels due to very serious imprecision (wide CIs) and serious indirectness (diversity in outcome measurement). We are uncertain if alternative care models alter hospital admissions for non-specific low back pain at follow-up closest to 12 months compared to usual care as the certainty of evidence was very low (OR 0.86, 95% CI 0.67 to 1.11; I² = 8%; 12 trials, 10,485 participants). We downgraded the certainty of the evidence by three levels due to serious indirectness (diversity in outcome measurement), serious publication bias (asymmetry of results), minor imprecision (wide CIs), and minor risk of bias (blinding of participants/personnel). High-certainty evidence indicates that alternative care models result in a small but clinically unimportant improvement in pain on a 0 to 10 scale (mean difference -0.24, 95% CI -0.43 to -0.05; I² = 68%; 36 trials, 9403 participants). Mean pain at follow-up closest to 12 months was 2.4 points on a 0 to 10 rating scale (lower score indicates less pain) with usual care compared to 2.2 points with alternative care models, a difference of 0.2 points better (95% CI 0.4 better to 0.0 better; minimal clinically important difference (MCID) 0.5 to 1.5 points). High-certainty evidence indicates that alternative care models result in a small, clinically unimportant improvement in back-related function compared with usual care (standardised mean difference -0.12, 95% CI -0.20 to -0.04; I² = 66%; 44 trials, 13,688 participants). Mean back-related function at follow-up closest to 12 months was 6.4 points on a 0 to 24 rating scale (lower score indicates less disability) with usual care compared to 5.7 points with alternative care models, a difference of 0.7 points better (95% CI 1.2 better to 0.2 better; MCID 1.5 to 2.5 points). We are uncertain of the effect of alternative care models on adverse events compared to usual care as the certainty of the evidence was very low (OR 0.81, 95% CI 0.45 to 1.45; I² = 43%; 10 trials, 2880 participants). We downgraded the certainty of the evidence by three levels due to serious risk of bias (blinding of participants/personnel), serious indirectness (variation in assumed risk), and serious inconsistency (substantial between-study heterogeneity).

Compared to usual care, alternative care models for non-specific low back pain probably lead to little or no difference in the quality of care and result in small but clinically unimportant improvements in pain and back-related function. Whether alternative care models result in a difference in total adverse events compared to usual care remains unresolved.