BMC Psychology (Aug 2023)

Development and validation of a predictive model for depression risk in the U.S. adult population: Evidence from the 2007–2014 NHANES

  • Wei Tian,
  • Yafeng Zhang,
  • Xinhao Han,
  • Yan LI,
  • Junping Liu,
  • Hongying Wang,
  • Qiuju Zhang,
  • Yujie Ma,
  • Guangcan Yan

DOI
https://doi.org/10.1186/s40359-023-01278-0
Journal volume & issue
Vol. 11, no. 1
pp. 1 – 9

Abstract

Read online

Abstract Background Depression is a prevalent mental health disorder with a complex etiology and substantial public health implications. Early identification of individuals at risk for depression is crucial for effective intervention and prevention efforts. This study aimed to develop a predictive model for depression by integrating demographic factors (age, race, marital status, income), lifestyle factors (sleep duration, physical activity), and physiological measures (hypertension, blood lead levels). A key objective was to explore the role of physical activity and blood lead levels as predictors of current depression risk. Methods Data were extracted from the 2007–2014 National Health and Nutrition Examination Survey (NHANES). We applied a logistic regression analysis to these data to assess the predictive value of the above eight factors for depression to create the predictive model. Results The predictive model had bootstrap-corrected c-indexes of 0.68 (95% CI, 0.67–0.70) and 0.66 (95% CI, 0.64–0.68) for the training and validation cohorts, respectively, and well-calibrated curves. As the risk of depression increased, the proportion of participants with 1.76 ~ 68.90 µg/L blood lead gradually increased, and the proportion of participants with 0.05 ~ 0.66 µg/L blood lead gradually decreased. In addition, the proportion of sedentary participants increased as the risk of depression increased. Conclusions This study developed a depression risk assessment model that incorporates physical activity and blood lead factors. This model is a promising tool for screening, assessing, and treating depression in the general population. However, because the corrected c-indices of the predictive model have not yet reached an acceptable threshold of 0.70, caution should be exercised when drawing conclusions. Further research is required to improve the performance of this model.

Keywords