Abstract: Objective To construct and validate a prediction risk model for sleep disorders in patients transferred from ICU. Methods With convenient sampling, 518 patients hospitalized in the ICU of two tertiary grade-A hospitals in Bengbu were selected and divided into training set (n=363) and verification set (n=155). General Information Questionnaire and Clinical Data Questionnaire were used to collect risk factors, and Pittsburgh Sleep Quality Index (PSQI) was used for one-month follow-up after transfer from the ICU. The patients were further divided into non-sleep disorder group and sleep disorder group according to the results of follow-up. Logistic regression was used to determine the influencing factors; R software to establish a nomogram to predict sleep disorder risk in patients transferred from ICU; C-index and calibration curve were used to evaluate the predictive ability of the model. Results Nine predictors were included in the model including total length of stay in ICU, APACHEII score, Charlson comorbidity index score, tracheotomy, anxiety, depression, post-traumatic stress disorder, ICU infection, and the number of invasive tubes when transferred from ICU. The model was internally validated with a C-index of 0.926 (95%CI:0.898~0.953) and externally validated with a C-index of 0.930 (95%CI:0.889~0.972), and the calibration plot showed good results. Conclusion The nomogram to predict sleep disorder risk for patients transferred from ICU can be used to screen the high risk group with sleep disorder.

Key words: ICU, intensive care, sleep disorder, nomogram, predictive model