This is the authors' abstract. We don't add key points for this paper.
Background Malnutrition, as reflected by a low Geriatric Nutritional Risk Index (GNRI), has been associated with poor outcomes after surgery; however, its relationship with early postoperative intensive care unit (ICU) admission in patients undergoing orthopedic surgery remains unclear. Methods This retrospective study included two independent cohorts: a derivation cohort of 4,030 operations in 3,699 orthopedic surgery patients from the INSPIRE database and an external validation cohort of 493 patients. GNRI was calculated as 1.489 × serum albumin (g/L) + 41.7 × (actual body weight/ideal body weight), and categorized into major (<82), moderate (82–92), low (92–98), and no (>98) nutritional risk. Logistic regression with progressive adjustment, restricted cubic splines, threshold (segmented) regression, subgroup and sensitivity analyses (propensity-score matching, inverse-probability and overlap weighting, restriction to patients aged ≥65 years, exclusion of intraoperative covariates, and cluster-robust standard errors), incremental predictive value [hierarchical models, C-statistic, net reclassification improvement (NRI), integrated discrimination improvement (IDI)], and external validation of the prespecified prediction model were used. The primary outcome was early postoperative ICU admission. Results The incidence of ICU admission increased stepwise across decreasing GNRI categories in the derivation cohort (8.8%, 18.3%, 22.9%, and 29.1% for GNRI >98, 92–98, 82–92, and <82, respectively; p < 0.001). After full adjustment, each 1-unit decrease in GNRI was associated with a 3% higher odds of ICU admission [odds ratio (OR) 1.03, 95% confidence interval (CI) 1.01–1.05; p < 0.001]. A non-linear dose–response relationship was observed ( P for non-linearity = 0.020), with an exploratory turning point at GNRI = 96.7 (95% bootstrap CI 88.5–102.0), above which each unit increase in GNRI was associated with lower odds of ICU admission (OR 0.933, 95% CI 0.904–0.964) and below which the association plateaued. Sensitivity analyses showed directionally consistent results, although the association was attenuated and statistically inconclusive after propensity-score matching (OR 1.20, 95% CI 0.87–1.65; p = 0.260). GNRI added minimal discriminative value beyond its own components: the C-statistic was 0.796 for the baseline model, 0.798 after adding albumin and BMI, and 0.800 after further adding GNRI (both increments p > 0.2). In the external validation cohort, GNRI <82 remained associated with a higher risk of ICU admission (OR 3.84, 95% CI 1.23–12.03; p = 0.021); transported prediction models showed an AUC of 0.640 with a calibration slope of 0.61. Conclusion Preoperative low GNRI is independently associated with early ICU admission after orthopedic surgery, with a non-linear dose–response relationship. Because planned and unplanned admissions could not be distinguished, and because the incremental discriminative value of GNRI beyond routinely available data (including albumin and BMI) was minimal, GNRI should be regarded as an accessible risk-stratification aid rather than a stand-alone predictor of postoperative deterioration.
Answers are grounded in this paper and cite their sources.
No takes yet. Share an insight, caveat, or question.
Zhang et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: