Establishment of a regression equation for risk factors of refeeding syndrome in critically ill elderly patients and analysis of intervention measures

Objective  To establish a regression equation for risk factors of refeeding syndrome (RFS) in critically ill elderly patients and analyze the relevant intervention measures.

Methods  Clinical materials of 154 critically ill elderly patients treated in Intensive Care Unit (ICU) from January 2021 to March 2023 were retrospectively analyzed, and they were divided into RFS group (n=51) and non-RFS group (n=103) according to incidence condition of RFS. The influencing factors of RFS were analyzed by Logistic regression model; the predictive values of predictors for RFS were analyzed by receiver operating characteristic (ROC) curve; the relevant Logistic regression equations were constructed and verified, and relevant nursing interventions were formulated.

Results  The incidence of RFS in critically ill elderly patients was correlated with the Acute Physiology and Chronic Health Evaluation Ⅱ (APACHE Ⅱ) score, the Nutritional Risk Screening 2002 (NRS2002) score, invasive mechanical ventilation, fasting time before feeding, D-dimer level, nutritional intake methods, and pre-feeding blood phosphorus, potassium and magnesium levels (P < 0.05). Logistic regression analysis showed that APACHE Ⅱ score, NRS2002 score, nutrient intake methods and pre-feeding blood phosphorus and potassium levels were the independent risk factors of RFS in critically ill elderly patients (P < 0.05). The ROC curve results showed that the values of area under the curve (AUC) of APACHE Ⅱ score, NRS2002 score, nutritional intake methods, pre-feeding blood phosphorus and potassium levels, and the combined predictor in predicting the incidence of RFS in critically ill elderly patients were 0.754, 0.723, 0.707, 0.783, 0.774 and 0.859, respectively (P < 0.05). The Logistic regression equation for RFS was as follow: L=0.085×APACHE Ⅱ score-0.337×NRS 2002 score+0.537×nutrient intake methods-0.776×pre-feeding blood phosphorus level-0.207×pre-feeding blood potassium level+0.942. The predictive value of this equation was good, and targeted nursing interventions could be formulated based on this equation.

Conclusion  The regression equation of risk factors can be used for clinical prediction of the RFS risk in critically ill elderly patients, and clinical nursing interventions can be formulated based on the regression equation to prevent the occurrence of RFS.