The association between nutrition risk status assessment and hospital mortality in Chinese older inpatients: a retrospective study.

Purpose: The association between nutritional risk status assessment and hospital mortality in older patients remains controversial. The aim of this study was to assess the relationship between nutritional risk on admission and in-hospital mortality, and explore the best Nutritional Risk Status Screening 2002 (NRS2002) threshold for predicting in-hospital mortality of older inpatients in China.

Method: The elderly inpatients were recruited from a hospital in Hunan Province, China. Nutritional risk was screened and assessed using the NRS2002. Logistic regression was used to analyze whether NRS2002 scores were independently associated with hospital mortality, and the results were expressed as odds ratios (OR) and 95% confidence intervals (CIs). Receiver operating characteristic curve (ROC) was used to determine the best NRS2002 threshold for predicting in-hospital mortality in elderly inpatients. And 500 bootstrap re-samplings were performed for ROC analysis.

Result: In total, 464 elderly inpatients completed the survey (15 of whom died, 205 males and 259 females, mean age = 72.284 ± 5.626 years). Multifactorial analysis revealed that age, the NRS2002 score, and length of hospital stay significantly influenced in-hospital mortality among older inpatients (P < 0.05). The results also showed that higher NRS2002 scores were associated with an increased risk of in-hospital mortality in both the unadjusted (OR = 1.731,95%CI = 1.362-2.20, P < 0.0001), adjusted model I (OR = 1.736, 95% CI = 1.354-2.206, P < 0.0001) and model II (OR = 1.602, 95% CI = 1.734-2.488, P = 0.0005). The optimal NRS2002 threshold for predicting in-hospital mortality in older inpatients was 3.5, with the largest ROC area of 0.84.

Conclusion: Our findings indicated that nutritional risk was an independent predictor of in-hospital mortality, with a cut-off value of 3.50 for the NRS2002 nutritional risk assessment being more appropriate than a cut-off value of 3.0.