Core Training for Pain Management and Functional Improvement in Patients With Patellofemoral Pain Syndrome: A Systematic Review and Meta-analysis.

Exercise cannot be blinded, so participants and therapists usually knew which programme was being done. The reviewers judged blinding partly on whether people could guess their group, which is a weaker safeguard than a blinded trial. The trials disagreed markedly with each other on pain, which is why that result was graded very low certainty rather than because the pooled estimate was small. Several of the trials that favoured core training also gave that group more exercise than the knee training group. The reviewers themselves suggest the extra benefit may come from the larger dose rather than from what was trained. Trials were grouped by the muscles the programme was expected to strengthen rather than by exactly what people did, and the reviewers say this may have introduced clinical heterogeneity. The funnel plot for pain was not symmetric and Egger's test suggested possible publication bias, so small positive trials may be over-represented. Six of the 19 trials did not report whether they calculated a sample size in advance, and three scored below 6 on the PEDro scale. The long-term follow-up results come from few trials and were also highly inconsistent.

The authors state that financial disclosure statements were obtained and that no conflicts of interest were reported by the authors or by anyone who controlled the content of the article. No industry funder or sponsor role in designing or writing the paper is described.

Do not read this as proof that core training beats knee training for pain. That comparison rested on very low certainty evidence, the trials disagreed with each other considerably, the funnel plot was not symmetrical and Egger's test suggested possible publication bias, and several of the trials favouring core training also gave that group more exercise than the control group.