Development and validation of a risk prediction model for demoralization syndrome in patients with type 2 diabetes.
Demoralization syndrome (DS), characterized by helplessness, hopelessness, and diminished self-worth, is a common but underrecognized psychological distress in chronic diseases. Type 2 diabetes mellitus (T2DM) affects over 537 million adults worldwide, yet DS remains underexplored in this population.
A cross-sectional observational study using a convenience sampling method was conducted to recruit 852 patients with T2DM from a hospital in China between October 2023 and October 2024. Participants were divided into a training set (n = 598) and a temporal validation set (n = 254) based on the chronological order of the survey. Feature selection was conducted using LASSO regression. A risk prediction model was developed using multivariable logistic regression, and a nomogram was subsequently constructed. Bootstrap resampling and Decision Curve Analysis (DCA) were utilized for internal evaluation and clinical utility assessment.
LASSO and multivariable logistic regression identified eight independent influential factors for DS: educational level, two-hour postprandial plasma glucose, glycated hemoglobin A1c, diabetes self-management, the confrontation and avoidance dimensions of the medical coping modes questionnaire, resilience, and sleep quality. The model demonstrated excellent predictive accuracy, with an AUC of 0.898 in the training set (optimism-corrected AUC of 0.892 via Bootstrap) and 0.868 in the temporal validation set. DCA indicated a positive net clinical benefit across a wide range of threshold probabilities.
This well-calibrated and highly discriminative model, validated temporally, offers healthcare professionals a robust and practical tool for the early identification of DS and the optimization of clinical nursing decisions in T2DM patients.
A cross-sectional observational study using a convenience sampling method was conducted to recruit 852 patients with T2DM from a hospital in China between October 2023 and October 2024. Participants were divided into a training set (n = 598) and a temporal validation set (n = 254) based on the chronological order of the survey. Feature selection was conducted using LASSO regression. A risk prediction model was developed using multivariable logistic regression, and a nomogram was subsequently constructed. Bootstrap resampling and Decision Curve Analysis (DCA) were utilized for internal evaluation and clinical utility assessment.
LASSO and multivariable logistic regression identified eight independent influential factors for DS: educational level, two-hour postprandial plasma glucose, glycated hemoglobin A1c, diabetes self-management, the confrontation and avoidance dimensions of the medical coping modes questionnaire, resilience, and sleep quality. The model demonstrated excellent predictive accuracy, with an AUC of 0.898 in the training set (optimism-corrected AUC of 0.892 via Bootstrap) and 0.868 in the temporal validation set. DCA indicated a positive net clinical benefit across a wide range of threshold probabilities.
This well-calibrated and highly discriminative model, validated temporally, offers healthcare professionals a robust and practical tool for the early identification of DS and the optimization of clinical nursing decisions in T2DM patients.