Are There Interindividual Differences in Anxiety as a Result of Aerobic Exercise Training in Adults With Fibromyalgia? An Ancillary Meta-analysis of Randomized Controlled Trials.

Only five trials and 333 participants, with four of the five enrolling women only, so the findings may not hold for men or for people with different fibromyalgia severity. Every included study was rated at high risk of bias, largely because participants and the people delivering the exercise knew which group they were in and anxiety is self-reported. The authors accept that this could push the exercise group toward reporting bigger improvements. Major asymmetry suggested publication bias, so the anxiety benefit seen here may be larger than what would actually happen. The prediction interval for a future trial's average effect included zero, meaning a new study could easily find no difference at all. Certainty of the evidence was graded low, driven by risk of bias, plausible confounding, and publication bias. Dropout ranged from 0% to 50% across the exercise groups. No dose-response analysis was possible and no meta-regression was run, so the paper cannot say what dose or type of exercise is best. No agreed minimum clinically important difference exists for anxiety in fibromyalgia, so the authors assumed one at 0.45 points; a different value would have changed those conclusions. The number-needed-to-treat of 7 was calculated using an assumed 30% control-group risk and an effect size converted from this study's own scale, not measured directly in these trials. Interventions lasted at most 24 weeks, so nothing is known about longer-term effects.

The authors declare no conflicts of interest. No funding source or funder role is stated in the article text; the journal indexes the paper as receiving extramural NIH research support.

Do not read the main finding, that people's responses varied no more than chance, as meaning aerobic exercise does not help anxiety in fibromyalgia. Exercise did reduce anxiety more than control in these trials. What the study failed to find was evidence that people genuinely differ in how much they respond.