Association between preoperative Geriatric Nutritional Risk Index and 90-day all-cause mortality after non-cardiac surgery in older adults: a retrospective cohort study

Background Older adults undergoing non-cardiac surgery remain vulnerable to postoperative death, yet a simple preoperative measure that identifies this risk across diverse procedures is needed. We evaluated whether the Geriatric Nutritional Risk Index (GNRI) defines a high-risk mortality phenotype in this population. Methods This retrospective cohort study used the INSPIRE v1.4.2 perioperative database. We included patients aged ≥65 years undergoing non-cardiac surgery and compared GNRI ≤98 with GNRI >98. The primary outcome was 90-day all-cause mortality; secondary outcomes were 30-day, 1-year and in-hospital mortality, with 90-day major adverse cardiovascular and cerebrovascular events (MACCE) assessed exploratorily. The primary Cox model adjusted for preoperative characteristics without penalization; the expanded penalized perioperative model was secondary. Estimates were pooled using Rubin rules across 20 multiply imputed datasets. Results The analytic cohort included 25,126 patients, including 6,314 with GNRI ≤98 and 18,812 with GNRI >98. During 90 days of follow-up, 349 patients died. GNRI ≤98 was associated with substantially higher 90-day mortality before adjustment (HR, 15.28; 95% CI, 11.53–20.25; p < 0.001) and after preoperative adjustment (HR, 3.77; 95% CI, 2.71–5.25; p < 0.001). The expanded penalized perioperative estimate was 1.87 (95% CI, 1.59–2.19). The preoperative-adjusted association extended to 30-day mortality (HR, 3.50; 95% CI, 1.79–6.86), 1-year mortality (HR, 3.22; 95% CI, 2.81–3.69) and in-hospital mortality (HR, 2.87; 95% CI, 2.09–3.94). Continuous GNRI showed a significant nonlinear association with 90-day mortality. The preoperative-adjusted association with exploratory 90-day MACCE was not evident (HR, 0.93; 95% CI, 0.82–1.07). Conclusion Preoperative GNRI ≤98 was associated with substantially higher postoperative mortality in older adults undergoing non-cardiac surgery. These findings support GNRI as a marker of nutritional risk and physiological vulnerability and motivate prospective evaluation of its role in perioperative assessment.