Manual therapy with exercise for neck pain

Abstract

RATIONALE: Manual therapy and exercise are supported by evidence of effectiveness as single modal interventions for neck pain; however, their combined effect remains unclear. OBJECTIVES: To assess the benefits and harms of manual therapy with exercise versus placebo or no treatment for acute to chronic neck pain with or without radicular symptoms or cervicogenic headache in adults. SEARCH METHODS: We searched multiple databases (CENTRAL, MEDLINE, Embase, CINAHL, Index to Chiropractic Literature, trial registries) together with reference checking and handsearching up to 5 March 2025. ELIGIBILITY CRITERIA: We included parallel, cross-over, or cluster-randomised controlled trials (RCTs) in adults with neck pain, which compared manual therapy and exercise with placebo or no treatment. We excluded studies in people with myelopathy or headaches not of cervical origin. OUTCOMES: Our outcomes were pain intensity, function or disability, health-related quality of life, participant-reported treatment success, and serious or non-serious adverse events, measured at short-term and long-term follow-up. RISK OF BIAS: We used RoB 1 plus an additional six items to assess bias in the included studies. SYNTHESIS METHODS: We synthesised results for each outcome using meta-analysis or, when not possible, SWiM methods. We used random-effects models to calculate mean differences (MD) or standardised mean differences (SMD) and 95% confidence intervals (CI) for continuous outcomes, and risk ratios (RR) with 95% CI for adverse events. We used GRADE to assess the certainty of evidence. INCLUDED STUDIES: We included nine RCTs (seven parallel, two cross-over) with a total of 694 participants. Studies were conducted in outpatient settings across North America, Europe, Central Asia, East Asia, and the Pacific. Manual therapy with exercise was compared with placebo (two studies) or no treatment (seven studies). Participants were 76% female, with a mean age of 46 years and mean pain severity of 4.75 on a 0 to 10 scale. Six studies (67%) reported receiving institutional or government funding. The disorder classification included chronic (n = 8) and subacute (n = 1) neck pain. SYNTHESIS OF RESULTS: The main biases affecting our findings were selection (44%), performance (100%), detection (100%), and reporting bias (78%). Performance bias is an inherent limitation in manual therapy and exercise RCTs, while detection bias was unavoidable due to reliance on self-reported outcomes. All data reflect short-term follow-up (closest to four weeks). Manual therapy with exercise versus placebo (short-term) Manual therapy with exercise may result in: 1) little or no difference in pain, which was a mean of 3.35 with placebo and showed little to no improvement of 0.91 points (95% CI 1.85 better to 0.04 worse) on a 0 to 10 scale, where a lower score indicates less pain (I² = 0%; 2 studies, 114 participants; low-certainty evidence due to imprecision, performance and detection bias); 2) moderate increase in function, which was a mean of 21.50 with placebo and improved by 10.20 points (95% CI 16.84 better to 3.56 better) on a 0 to 100 scale where 0 indicates best function (I² = 0%; 2 studies, 115 participants); this represents transformed data (SMD 0.77 better, 95% CI 1.15 better to 0.39 better; low-certainty evidence due to imprecision, performance and detection bias); 3) little or no improvement in health-related quality of life, which was a mean of 52.10 with placebo and slightly improved by 2.00 points (95% CI 5.78 better to 1.78 worse) on the Short Form-12 0 to 100 scale, where 0 indicates better quality of life (1 study, 64 participants; low-certainty evidence due to imprecision, performance and detection bias). Data on participant-reported treatment success and adverse effects were not available. Manual therapy with exercise versus no treatment (short-term) Manual therapy with exercise may result in: 1) large reduction in pain, which was a mean of 4.01 with no treatment and had a large reduction of 2.44 points (95% CI 3.23 better to 1.65 better) on a 0 to 10 scale where a lower score indicates less pain (I² = 66%; 7 studies, 360 participants; low-certainty evidence due to imprecision, performance and detection bias); 2) moderate improvement in function, which was a mean of 22.38 with no treatment and improved by 13.84 points (95% CI 25.24 better to 2.44 better) on a 0 to 100 scale where 0 indicates best function (I² = 92%; 5 studies, 303 participants; low-certainty evidence due to imprecision, performance and detection bias); 3) moderate improvement in health-related quality of life, which was a mean of 53.60 with no treatment and improved by 24.80 points (95% CI 31.38 better to 18.22 better) on the Short Form-36 0 to 100 scale, where 0 indicates better quality of life (1 study, 65 participants; low-certainty evidence due to imprecision, performance and detection bias); 4) very uncertain evidence for participant-reported treatment success (SMD 2.57 bet