Correlation between malnutrition and mortality in older patients aged ≥90 years with multimorbidity

Malnutrition is prevalent in geriatric patients and associated with poor prognosis. This study aimed to evaluate the incremental prognostic value of different nutritional assessment tools in patients (90 years and older) with multimorbidity in China.

Patients aged ≥90 years with multimorbidity from the Geriatric Research Center in Nanjing Jinling Hospital from January 1, 2008 to December 31, 2018 were analyzed. Patients were followed until December 31,2022. The nutrition status was assessed according to the mini nutritional assessment (MNA), the geriatric nutritional risk index (GNRI), and the prognostic nutritional index score (PNI), respectively. The outcome was all-cause death. Kaplan-Meier method was used to draw the survival curve, and univariate and multivariate Cox proportional risk regression analysis was used to explore the influencing factors of all-cause death in the patients.

160 participants aged 90(90,91) years were included. During a median follow-up of 5.41(3.12-7.64) years, 106(66.25 %) patients died. Deceased patients had lower MNA [20.75(16.75,23.00) vs. 26.00(24.00,26.00); p < 0.001], lower GNRI [96.21±8.75 vs. 100.94±6.80; p = 0.001] and lower PNI [46.16(40.77,49.57) vs. 47.75(45.36,51.53); p = 0.010] than did survivors. According to MNA, GNRI, and PNI score, 93(58.1 %), 72(45.0 %) and 41(25.6 %) of participants were at risk of malnutrition. Multivariate analysis revealed that malnutrition was independently associated with increased risk for mortality by MNA score (adjusted hazard ratio [HR] 2.502, 95 % confidence interval [CI] 1.561-4.009, p<0.001), GNRI score (adjusted HR 1.650, 95 % CI, 1.117-2.438, p = 0.012), and PNI score (adjusted HR 2.894, 95 % CI, 1.891-4.431, p<0.001). Furthermore, the inclusion of malnutrition indicators in the survival prediction model significantly improved the predictive power of mortality.

The risk of malnutrition, as assessed by MNA, GNRI and PNI, in long-lived patients with multimorbidity is a strong independent predictor of mortality and adds significant prognostic information to the survival models.