A study on the identification of factors related to depression in a population with an increasing number of chronic diseases in the short term in China based on a health ecology model
Weimin Guan,Wenyu Su,Huaiju Ge,Shihong Dong,Huiyu Jia,Yan Liu,Qing Yu,Yuantao Qi,Huiqing Zhang,Guifeng Ma
DOI: https://doi.org/10.1016/j.jad.2024.09.090
2024-09-16
Abstract:Background: The rapid increase in the number of patients with chronic diseases and depression, as well as the rapid spread of their effects, have led to these two health problems gradually developing into major public health issues in China and around the world. Currently, many individuals with chronic diseases are experiencing depressive symptoms one after another. Therefore, it is imperative to conduct research on how to prevent depression in this growing population of individuals with chronic diseases in a timely manner. Methods: Based on the data of the 2015 and 2018 national follow-up surveys of the China Health and Retirement Longitudinal Study, a total of 7641 patients with short-term increase in the number of chronic diseases were selected as the study objects, and a binary logistic regression model was constructed according to the five dimensions of the health ecology model. The neural network model was used to explore the main (first two) factors affecting the increase in the number of chronic diseases in China in the short term, and the random forest and extreme value gradient lifting algorithm were used to verify them, and effective suggestions were put forward. Results: The detection rate of depression in the population with increasing number of chronic diseases from 2015 to 2018 was 42.13 %. The model was established based on five dimensions of the health ecology model: Model 1 (Personal trait layer), Model 2 (Personal trait layer plus Behavioral feature layer), Model 3 (Personal trait layer plus Behavioral feature layer plus Living and working conditions layer), Model 4 (Personal trait layer plus Behavioral feature layer plus Living and working conditions layer plus Networking layer) and Model 5 (Personal trait layer plus Behavioral feature layer plus Living and working conditions layer plus Networking layer plus Policy environment layer).The prediction accuracy of the five models was 66.4 %, 68.3 %, 70.7 %, 71.6 % and 71.6 %, respectively, and Model 5 showed that the P values of gender, self-rated health, night's sleep time (h), disability, life satisfaction, child satisfaction, place of residence and highest level of education were all <0.05, life satisfaction and self-rated health importance were 0.249 (100 %) and 0.226 (90.8 %). Conclusion: Gender, self-rated health, night sleep duration, disability, satisfaction with life, satisfaction with children, place of residence and highest level of education were the main influencing factors for the increase of depressive symptoms in the population with chronic diseases in the short term, among which life satisfaction and self-rated health have the greatest impact on depressive symptoms, and there is an interaction between the two.