Related Experiment Videos
Prognostic nutritional index versus pragmatically operationalized GLIM criteria for predicting postoperative
YongBo Yang1, Rui Shi1, Jian Zhou1
1Department of Orthopedics, Xinxiang Central Hospital, The Fourth Clinical College of Henan Medical University, Xinxiang, Henan, China.
Background:
Preoperative malnutrition increases surgical complications, yet the Prognostic Nutritional Index (PNI) and Global Leadership Initiative on Malnutrition (GLIM) criteria demonstrate conflicting validity in spine surgery populations.
Methods:
Prospective outcome cohort with retrospectively abstracted baseline data, comprising 1,341 consecutive spine surgery patients (Chinese tertiary center, February 2022-September 2025) stratified by PNI (normal ≥45 vs. low <45) and GLIM criteria. Baseline clinical and demographic variables were extracted from electronic medical records; all postoperative outcomes were ascertained prospectively using predetermined protocols. Outcomes: 30-day major complications (primary), healthcare utilization, and 90-day patient-reported recovery. Analysis: multivariable logistic regression, AUC comparison, net reclassification improvement.
Results:
Among 1,341 consecutive spine surgery patients (mean age 54.8 years, 46.8% female, 84.9% elective), low PNI (6.5%, n = 87) and GLIM malnutrition (6.9%, n = 92) showed near-complete absence of positive agreement (κ = 0.00; positive agreement 5.6%, negative agreement 93.2%), with 94.3% of PNI-high-risk patients classified as GLIM-negative. Low PNI independently predicted the primary endpoint of 30-day major complications (40.2% vs. 22.9%; adjusted OR 2.02, 95% CI 1.28-3.18, p = 0.003). PNI provided statistically significant but numerically modest incremental discrimination (AUC 0.73 vs. baseline 0.68, ΔAUC 0.05, p < 0.001; net reclassification improvement 18%). By contrast, pragmatically operationalised GLIM showed no prognostic association (adjusted OR 1.18, p = 0.50) and added no discriminative value over PNI (combined AUC 0.73, likelihood ratio p = 0.50). In exploratory secondary analyses, low PNI patients showed directionally higher healthcare utilization and persistently impaired 90-day recovery across functional, nutritional, and patient-reported domains; these findings are hypothesis-generating given multiple comparisons without multiplicity adjustment.
Conclusion:
PNI independently predicts perioperative complications and 90-day recovery trajectories in spine surgery patients, whereas the pragmatic GLIM operationalization employed in this study demonstrated no significant prognostic associations. Whether a fully resourced GLIM implementation with imaging-based muscle assessment would perform differently remains to be established.
