Objective To investigate the influencing factors of malnutrition in hospitalized elderly patients with comorbidities of chronic diseases, and to construct a predictive model.
Methods A convenience sampling method was used to select 426 elderly patients with comorbidities of chronic diseases admitted to the Department of Geriatrics of Suzhou Ninth People's Hospital Affiliated to Soochow University from January 2023 to February 2024. Based on a Mini-nutritional Assessment-Short Form (MNA-SF) score < 8 and either an albumin level < 34.0 g/L or a prealbumin level < 200 mg/L as reference of malnutrition, patients were classified into malnutrition group and non-malnutrition group. General characteristics, oral statusassessed using the Oral Health Assessment Tool (OHAT), dietary inflammatory index (DII, evaluated through a food frequency questionnaire), and activities of daily livingassessed using the Barthel Index (BI)were compared between the two groups. Multivariable Logistic regression analysis was employed to explore the influencing factors of malnutrition in elderly patients with comorbidities of chronic diseases and to construct a model formula. A gradient boosting machine (GBM) algorithm was implemented using R software to build a GBM predictive model. Receiver Operating Characteristic (ROC) curves were utilized to analyze the predictive performance of both models, and the Delong test was applied to compare the difference of the area under the curve (AUC).
Results Ninety-two patients were diagnosed with malnutrition (malnutrition group), while 334 patients had no malnutrition (non-malnutrition group). Statistically significant differences were observed between the malnutrition and non-malnutrition groups in terms of age, the number of chronic comorbidities, the number of medication taken, OHAT scores, DII, and BI scores (P < 0.05). Advanced age, a higher number of chronic comorbidities, a greater number of medication taken, higher OHAT scores, higher DII, and lower BI scores were all influencing factors of malnutrition in elderly patients with comorbidities of chronic diseases (P < 0.05). The ROC curve analysis revealed an AUC of GBM model was 0.901 and 0.874 for the Logistic regression model. The Delong test indicated that the predictive performance of the GBM model was superior to that of the Logistic regression model (P < 0.05).
Conclusion Malnutrition in hospitalized elderly patients with chronic multimorbidity is associated with age, the number of chronic comorbidities, the number of medications taken, OHAT scores, DII, and BI scores. The constructed GBM model can effectively assess the risk of malnutrition in these patients.
DownLoad: