Abstract: Objective To construct a nomogram prediction model of the risk of developing hypothermia in patients undergoing continuous renal replacement therapy and validate the model. Methods A retrospective study was conducted, and 450 patients undergoing consecutive renal replacement therapy in a tertiary grade-A hospital in Zhejiang Province from January 2019 to December 2022 were selected by convenience sampling and they were divided into hypothermia group (n=140) and non-hypothermia group (n=310) according to the occurrence of hypothermia or not. One-way analysis of variance and logistic regression analysis were used to analyze the risk factors, construct risk prediction model, and plot a nomogram. A prospective cohort study was used to collect 120 patients undergoing consecutive renal replacement therapy in the intensive care unit of a tertiary grade-A hospital from January to September 2023 as an external validation set to externally validate the model. Results The risk prediction model was constructed by multifactorial analysis incorporating baseline body temperature (OR=0.092), APACHE-II score (OR=2.499), whether or not mechanical ventilation (OR=2.578), and duration of continuous renal replacement therapy (OR=3.483). The areas under the ROC curves of the training and validation set were 0.856 and 0.819, respectively; Bootstrap internal validation of the mean absolute error between the actual and predicted value of its calibration curves was 0.027; Hoer-Lemeshow test of P=0.063 (P>0.05) indicated high goodness of fit of the model; The calibration curve and the decision curve showed that the model had the calibration degree and clinical usefulness. Conclusion The nomogram risk prediction model constructed in this study has the differentiation, calibration and clinical practicability, and can effectively predict the risk of hypothermia in patients undergoing continuous renal replacement therapy.

Key words: continuous renal replacement therapy, hypothermia, risk factor, predictive model, nomogram