The Meta-Analysis Ecosystem Model: evidence velocity, effect decay, and a Future-Proof Index across 501 Cochrane reviews

Abstract

Does the magnitude of early meta-analytic effects systematically decay as evidence accumulates across the Cochrane Library? The Meta-Analysis Ecosystem Model analysed 3,651 trajectories from 501 Cochrane reviews (≥5 studies each) over a median 16-year span; 2,817 of 3,062 computable trajectories gave a finite, in-range decay ratio (834 excluded as undefined or >10). It applies cumulative fixed-effect meta-analysis to binary outcomes, which understates heterogeneity, and integrates velocity, decay, and stability with publication-bias scores into a Future-Proof Index (0–100). Among the 2,817 the median decay ratio was 0.66 (IQR 0.32–1.23), 67.3% shrinking; of 3,062, 25.1% reversed sign but only 3.4% meaningfully. Decay was independent of velocity (r=−0.002) and bias (r=+0.001)—not a scaling artefact—though velocity and bias correlated weakly (r=−0.19). Two-thirds of pooled effects shrank as evidence matured, so early meta-analytic results should not be treated as definitive. The decay ratio is sensitive to the fixed-effect choice, and the Future-Proof Index is an expert-weighted heuristic, not a validated instrument.