Background: To identify correlates of the incidence of cognitive impairment among older Chinese populations through the use of logistic regression analysis-based decision tree approaches.
Methods: Correlates of cognitive impairment among older Chinese adults were identified through logistic regression analyses, with significant variables subsequently being incorporated into a decision tree analysis, with the CHAID method being employed for pre-pruning.
Results: The risk score derived from the combination of logistic regression and decision tree analyses (0.237) was lower than that derived from a decision tree analysis alone (0.389). The primary factors related cognitive impairment in this patient population included age, gender, residence status, physical health status, and caring for grandchildren.
Conclusion: A combination of logistic regression and decision tree analyses can lower predicted risk scores, enabling the subdivision of populations with different characteristics and providing intuitive and specific insight regarding the effects of individual variables on predictive analyses. Overall, these results suggest that older adults in rural areas of China should be the focus of further cognitive impairment screening and interventions, particularly for older women.
Keywords: CHARLS; China; cognitive impairment; decision tree; the elderly.
Copyright © 2022 Wang, Zhang, Wang, Liu and Ding.