Geriatric Nutritional Risk Index is Associated with Hospital Death in Elderly Patients with Multiple Organ Dysfunction Syndrome: A Retrospective Study Based on the MIMIC-III Database

Zhi Mao; Tao Wen; Xiaoli Liu; Jinsong Chen; Pan Hu; Chao Liu; Hui Liu; Hongjun Kang; Zhengbo Zhang; Feihu Zhou

doi:10.3389/fnut.2022.834256

Geriatric Nutritional Risk Index is Associated with Hospital Death in Elderly Patients with Multiple Organ Dysfunction Syndrome: A Retrospective Study Based on the MIMIC-III Database

Front Nutr. 2022 Jun 3:9:834256. doi: 10.3389/fnut.2022.834256. eCollection 2022.

Authors

Zhi Mao¹, Tao Wen¹, Xiaoli Liu², Jinsong Chen³, Pan Hu¹, Chao Liu¹, Hui Liu¹, Hongjun Kang¹, Zhengbo Zhang^{2

4

5}, Feihu Zhou¹

Affiliations

¹ Department of Critical Care Medicine, The First Medical Center, Chinese People's Liberation Army General Hospital, Beijing, China.
² School of Biological Science and Medical Engineering, Beihang University, Beijing, China.
³ Department of Gerentology, Zhangzhou Zhengxing Geriatrics Hospital, Zhangzhou, China.
⁴ Center for Artificial Intelligence in Medicine, Chinese Peoples Liberation Army General Hospital, Beijing, China.
⁵ Department of Biomedical Engineering, Chinese People's Liberation Army General Hospital, Beijing, China.

Abstract

Purpose: Elderly patients with multiple organ dysfunction syndrome (MODS) have a higher mortality during hospitalization in the intensive care unit (ICU). Elderly patients often suffer from malnutrition. On the basis of the MIMIC-III database, this study analyzed the effect of the baseline nutritional status on the death of elderly patients with MODS during hospitalization.

Materials and methods: Elderly patients with MODS were screened out from MIMIC-III 1.4 database. The geriatric nutritional risk index (GNRI) was calculated and used to group patients into: normal nutrition (GNRI > 98) and malnutrition (GNRI ≤ 98) groups. The malnutrition group was divided into mild (92-98), moderate (82-91), and severe (≤81) groups. The differences in the baseline data and the incidence of adverse events between groups were compared. The GAM model was used to determine whether a curve relationship was present between the hospital death of elderly patients with MODS and GNRI and analyze the threshold saturation effect. The multivariate logistic regression was used to calculate the odds ratio (OR) of in-hospital deaths in different GNRI groups. The interaction test was performed to find subgroups with differences.

Results: A total of 2456 elderly patients with MODS were enrolled. A total of 1,273 (51.8%) and 1183 (48.2%) patients were in the normal nutrition and malnutrition groups, respectively. The mortality rate of patients in the normal nutrition group during hospitalization was lower than that in the malnutrition group (206/1273 vs. 292/1183, X2 = 27.410, P < 0.001; OR = 0.59, 95% CI: 0.48-0.72). The GAM model fitting analysis showed a threshold saturation effect at GNRI = 92. Adjusted OR values with GNRI ≥ 92 began to change to 1, and GNRI and death had no association. At GNRI < 92, high GNRI related to low risk of death. Subgroup analysis of patients with GNRI < 92 showed that the risk of death in elderly male patients was lower than that of female patients.

Conclusion: GNRI is related to the severity of illness in elderly patients with MODS. At GNRI < 92, moderate to severe malnutrition increases the risk of death in elderly patients with MODS during hospitalization.

Keywords: MIMIC-III; elderly; geriatric nutritional risk index; intensive care unit; multiple organ dysfunction syndrome.