To analyze the influencing factors of cognitive impairment and construct a risk assessment model for illiterate elderly individuals.

Based on the 2020 China Health and Retirement Longitudinal Study (CHARLS) database, illiterate elderly individuals aged≥60 years were taken as the research subjects. Ten-fold cross-validated LASSO regression was used to identify the optimal predictor variables. Logistic regression was further applied to explore the risk factors of cognitive impairment in this population, and an assessment model was further constructed. The performance of the model was evaluated using the receiver operating characteristic (ROC) curve, area under the curve (AUC), sensitivity, specificity, Hosmer-Lemeshow test, calibration curve, decision curve analysis (DCA), categorical/continuous net reclassification improvement (NRI), and integrated discrimination improvement (IDI).

A total of 1 575 illiterate elderly individuals were included in this study, with a prevalence of cognitive impairment of 39.75%. Lasso regression screened potential predictor variables, including age, marital status, residence type, alcohol consumption, physical activity, social engagement, and internet use. Multivariable logistic regression analysis showed that advanced age (70-79 years: OR=1.509, 95% CI: 1.120-2.032; ≥80 years: OR=2.178, 95% CI: 1.341-3.537), rural residence (OR=1.631, 95% CI: 1.159-2.295) were the risk factors. In contrast, being married (OR=0.607, 95% CI: 0.450-0.817), alcohol consumption (OR=0.647, 95% CI: 0.471-0.888), social engagement (OR=0.683, 95% CI: 0.524-0.891), moderate-to-high intensity physical activity (moderate intensity: OR=0.633, 95% CI: 0.428-0.937; high intensity: OR=0.628, 95% CI: 0.439-0.898), internet use (OR=0.327, 95% CI: 0.184-0.581) were protective factors. An assessment model was established based on the seven factors. The areas under the ROC curves of the assessment model in the training and validation sets were 0.670 and 0.655, respectively. The Hosmer-Lemeshow χ²=11.977, P=0.152, and χ²=7.977, P=0.436. Brier scores were 0.216 and 0.227, respectively. The calibration curve revealed good consistency between the predicted values and the actual values, and the DCA indicated that the model had good net benefit and assessment accuracy. Additionally, the model showed a significant incremental improvement in terms of risk reclassification.

The risk assessment model for cognitive impairment in illiterate elderly individuals, based on seven influencing factors including age and residence type, can provide support for the early identification of cognitive impairment in this population.