Choose a random-effects model when the included winter-health studies differ in populations, exposures, settings, or methods enough that assuming one common effect is implausible; record that reasoning alongside the estimate. Cochrane Handbook Chapter 10 treats fixed and random effects as different assumptions about the studies, so the decision rests on whether a shared true effect is credible for your evidence base.
Compare diversity against the common-effect assumption
List the clinical and methodological features that could shift the effect: who was studied, how winter exposure or the outcome was defined, and how each study was designed and analyzed. If those features vary in ways that plausibly change the effect, a common-effect assumption is hard to defend and a random-effects model fits the question better. If the studies are close replications, a common-effect model may be reasonable.
Build the justification paragraph
- State the assumed effect structure and the diversity you observed across included studies, citing the specific features that vary.
- Name the chosen model and explain why that diversity supports it, then report the pooled estimate with its confidence interval and a heterogeneity measure.
- Add a sensitivity analysis using the alternative model and state how the estimate and its interval change, so readers can see whether the choice alters the conclusion.
Cochrane Handbook Chapter 15 stresses careful interpretation of pooled results, including confidence intervals and applicability, so describe what the estimate means for the winter-health question you posed. Keep the paragraph focused on the model choice and its effect on the estimate; route full heterogeneity methods to a methods reference.