Building a predictive model for depression risk in fracture patients: insights from cross-sectional NHANES 2005–2020 data and an external hospital-based dataset

Abstract

Abstract Background Depression represents a frequent mental health challenge in individuals with fractures, negatively impacting their recuperation and overall well-being. The purpose of this research was to formulate and corroborate a prognostic framework for pinpointing depression risk among fracture sufferers by utilizing data from the National Health and Nutrition Examination Survey (NHANES) from 2005 to 2020 and a separate hospital-based group. Methods We analyzed records from 1,748 individuals with fractures documented in the NHANES database spanning 2005 to 2020, of which 362 were diagnosed with depression, as indicated by a Patient Health Questionnaire-9 (PHQ-9) score of 10 or higher. An additional validation group comprised 360 fracture patients sourced from a medical center. Considered variables for prediction encompassed demographic details, lifestyle habits, past medical conditions, and laboratory results. The method of least absolute shrinkage and selection operator (LASSO) regression facilitated the narrowing down of variables, while multivariate logistic regression was employed to pinpoint significant predictors. To assist in prediction, a nomogram was designed and subsequently validated. Results Five independent predictors were identified: drinking, insomnia, poverty-to-income ratio, education level, and white blood cell count. The nomogram showed good discrimination in the NHANES cohorts (training area under the curve (AUC) 0.734, validation AUC 0.740) and hospital-based external validation (AUC 0.711). Calibration curves and decision analysis supported its predictive accuracy and clinical value. Conclusion The constructed nomogram offers a precise and clinically relevant instrument for forecasting depression risk in patients with fractures, facilitating the early detection of individuals at high risk and enabling prompt intervention.

Keywords

• Depression
• Fracture
• Prediction model
• Nomogram