Abstract / Summary
Malnutrition is prevalent among older adults, yet comparative prognostic performance of nutritional indices for mortality in Chinese elderly remains unclear. To compare Geriatric Nutritional Risk Index (GNRI), Prognostic Nutritional Index (PNI), Hemoglobin-Albumin-Lymphocyte-Platelet (HALP), and Controlling Nutritional Status (CONUT) for predicting all-cause and cardiovascular mortality in older Chinese adults. This prospective cohort study included 1,807 participants aged 65 years and older from the Chinese Longitudinal Healthy Longevity Survey (CLHLS) 2014–2018. Cox proportional hazards models, restricted cubic splines (RCS), time-dependent receiver operating characteristic (ROC) curves, subgroup analyses, and sensitivity analyses were performed. Cox models were progressively adjusted (Model 0, unadjusted; Model 1, age and sex; Model 2, plus socioeconomic and lifestyle factors; Model 3, plus clinical covariates), and the proportional-hazards assumption was verified using Schoenfeld residuals. During 2.88-year follow-up, 627 all-cause and 172 cardiovascular deaths occurred. PNI was the most robust predictor (all-cause HR 0.948, cardiovascular HR 0.945, both p < 0.001). GNRI had the highest 3-year AUC (0.684). Each 1-unit higher ln(CONUT + 1) was associated with higher all-cause (HR 1.350) and cardiovascular (HR 1.423) mortality, whereas HALP showed only weak, adjustment-sensitive associations. Sensitivity analyses confirmed PNI superiority. The four indices should be interpreted as integrated markers of overall health and physiological reserve rather than isolated measures of malnutrition. PNI showed the most consistent associations, but discrimination was only moderate; these indices are not standalone prognostic tools and may serve only as adjuncts to multidimensional geriatric assessment pending external validation.