← All sources
Identifying patients with chronic pain who respond to acupuncture: results from an individual patient data meta-analysis
Skewness meta-analysis across 39 acupuncture RCTs (20,827 patients) finds control arms slightly more right-skewed than acupuncture arms for pain change scores, but sensitivity analyses excluding three outlying trials remove significance for sham-controlled comparisons.
Design
- IPD skewness analysis of 39 acupuncture RCTs (20,827 patients) published before Dec 2015
- Controls: 25 sham-controlled trials (n = 7,097) and 25 non-acupuncture-controlled trials (n = 16,041)
Findings
- Skew difference (acupuncture − control): +0.124 for non-acupuncture controls (p = 0.047) and +0.141 for sham controls (p = 0.029) — very small asymmetry shifts
- Sensitivity: excluding 3 prespecified outlier trials rendered the sham skew contrast non-significant (p = 0.2)
Interpretation
Provides limited evidence that exceptional responders alone explain acupuncture–control gaps; distribution shifts are subtle—pair with Vickers 2018 primary effect estimates.
Publication
Foster NE, Vertosick EA, Lewith G, et al. Acupunct Med. 2021 Apr;39(2):83-90. PMID 32571096.
Outcomes
- Meta-analysis of skewness differences: +0.124 (p=0.047) vs non-acupuncture controls and +0.141 (p=0.029) vs sham for (acupuncture − control) pain-change skew; sham contrast lost significance (p=0.2) after excluding three outlier trials.