Treating low back pain in athletes: a systematic review with meta-analysis

Key points

1. 01Across 14 RCTs, exercise was the most studied treatment and generally reduced pain and disability, but no exercise trial reported return-to-sport data.
2. 02Pooled exercise reduced pain on a 0 to 10 visual analogue scale by 1.65 points versus control, though heterogeneity was very high (I2 = 91%).
3. 03There was insufficient evidence to support manual therapy (massage, spinal manipulation) or biomechanical modifications as stand-alone treatments.
4. 04No RCTs evaluated spinal injections or surgery, and first-line general-population strategies like education, self-management, and cognitive-behavioural therapy were not tested in any trial.
5. 05The trials had biases across performance, attrition, and reporting domains, with mostly short follow-up, so effect estimates should be read cautiously.

How it was conducted

Design

Systematic review with random-effects meta-analysis of RCTs (PROSPERO CRD42018087229)

Search

EMBASE, Medline, CINAHL, Web of Science, Scopus from inception to September 2020; 14 RCTs included

Participants

541 athletes (67% men) aged 18 and over, mean age 30.1 years, across 12 sports and recreational to international levels

Intervention

Exercise (7 trials), manual therapy (5 trials), and biomechanical modifications (2 trials) versus usual care or other treatments

Outcomes

Pain (VAS), disability (ODI, RMDQ), and return to sport; follow-up ranged from 24 hours to 6 months

Analysis

Pooled mean and standardised mean differences with 95% CI; risk of bias by Cochrane tool; no formal certainty grading applied

What they found

• Exercise vs control reduced VAS pain by a mean of 1.65 points (95% CI -2.74 to -0.55, p=0.003, I2=91%); removing one outlier trial dropped the difference to 1.04 points and I2 to 10%.
• Exercise reduced disability (ODI and RMDQ pooled) with SMD -2.6 (95% CI -5.14 to -0.04), with very high heterogeneity (I2=96%) that sensitivity analysis did not resolve.
• No exercise trial reported return-to-sport outcomes, and no RCTs evaluated spinal injections or surgery.
• Biomechanical modifications: an unstable golf shoe reduced lab-test pain (21.97 mm vs 37.83 mm on a 100 mm VAS, p=0.05) but not on the course; cyclists adjusting saddle angle had 72% pain-free at 6 months with no between-group difference reported.
• Manual therapy showed only short-term benefits from massage and spinal manipulation, but most trials included co-interventions and could not be pooled.
• Nine of 14 trials were at high risk of performance bias and 2 at high risk of attrition bias, reflecting generally poor study quality.

Limitations

• The body of evidence is scarce and low quality, with several very small trials prone to sparse-data bias and no trials free of bias.
• Low back pain was inadequately defined across studies, episode staging was unclear, and confounders such as prior low back pain were often ignored.
• Follow-up was generally short with little data on recurrence or time to return to sport, making overestimation of positive outcomes likely.
• Most athletes were male (at least 65%), and the authors judged formal certainty grading not feasible given the heterogeneity.

Why it matters

For patients

If you are an athlete with low back pain, exercise is likely to ease your pain and improve function, but evidence cannot yet say which exercise is best or how fast you will return to sport.

For clinicians

Favour active, exercise-based, biopsychosocial management and shared decision-making; the evidence does not support stand-alone manual therapy, biomechanical tweaks, injections, or surgery for athletes.

For readers

There is a striking shortage of high-quality trials on treating low back pain in athletes, especially any measuring return to sport.