Modified frailty index versus risk analysis index for 30-day mortality after noncardiac surgery: a systematic review and meta-analysis

Background mFI and RAI are widely used in noncardiac surgical risk stratification, but they differ in content and data requirements. Because construct validity and outcome discrimination are distinct, we evaluated their prognostic discrimination for 30-day all-cause mortality without treating either instrument as a criterion-standard multidimensional frailty measure. Methods This PRISMA-aligned, prospectively registered systematic review used two complementary evidence levels: tool-specific random-effects meta-analyses across independent cohorts and direct within-study comparison. Standalone AUCs were pooled on the logit scale using REML with Hartung–Knapp inference and 95% prediction intervals. Risk of bias was assessed with QUIPS. CFS was analyzed post hoc as secondary exploratory evidence. Results Seventy-seven studies were included. For mFI, 39 independent cohorts yielded a pooled AUC of 0.634 (95% CI, 0.618–0.649; 95% prediction interval, 0.551–0.709). For RAI, 30 independent cohorts yielded 0.767 (95% CI, 0.737–0.795; 95% prediction interval, 0.579–0.888). Among 29 dual-tool studies, 21 had source-valid numeric comparisons in which RAI was higher, none had a source-valid numeric comparison in which mFI was higher, 2 supported RAI only qualitatively, and 6 could not be numerically classified; no paired meta-analysis was conducted because within-study covariance or delta-AUC standard errors were unavailable. Fourteen CFS studies were identified; 6 source-confirmed 30-day AUC cohorts yielded a post hoc exploratory AUC of 0.757 (95% CI, 0.618–0.857). Conclusions RAI showed higher average prognostic discrimination than mFI, a pattern also observed in most directly comparative studies. Separate pooled AUCs are not a formal superiority test because case mix differs across studies. The analysis does not establish construct validity, and CFS evidence remains exploratory. Systematic Review Registration https://www.crd.york.ac.uk/PROSPERO/view/CRD420261378778 , identifier PROSPERO, CRD420261378778.