A meta-analysis of 15 studies on exercise and type 2 diabetes reports a pooled RR of 0.72 (95% CI: 0.68–0.76), I² = 78%, and a funnel plot with visible asymmetry. The most appropriate next step in interpreting this meta-analysis is:
- A Accept the pooled estimate because the confidence interval excludes the null
- B Question the validity of the pooled estimate due to high heterogeneity and likely publication bias ✓
- C Use a fixed-effects model to eliminate the heterogeneity problem
- D Conclude the result is unreliable only because of funnel plot asymmetry
Correct answer: B. Question the validity of the pooled estimate due to high heterogeneity and likely publication bias
Explanation
I² of 78% indicates substantial heterogeneity (thresholds: <25% low, 25–50% moderate, >50% high), meaning studies are measuring different things and a single pooled estimate is misleading. Funnel plot asymmetry additionally suggests publication bias (small negative studies not published). Switching to a fixed-effects model does not resolve true between-study heterogeneity; it only averages results under the assumption of a common effect, which is unjustified here.
Reference: Park's Textbook of Preventive and Social Medicine, 27th ed.
High-yield for: NEET PGINI-CETNExTFMGEUSMLEPLABMRCP
Written and medically reviewed by the StethoPrep medical team.