The China Health and Retirement Longitudinal Study (CHARLS) is a nationally representative study of Chinese adults aged ≥ 45 years. The CHARLS is designed to describe the dynamics of retirement and its impact on health, health insurance, and economic well-being. The baseline survey was conducted in 2011-12 among 17,708 participants from 150 counties of China’s 28 provinces , and data on socioeconomic status, lifestyles, medications, health status and functioning assessments were collected. Details on the study design, sampling procedure, and data collection have been described in previous publications . Briefly, the CHARLS participants were recruited through a four-stage, stratified, cluster random sampling method. The CHARLS participants were followed biennially to obtain updated information. The CHARLS data are available for the baseline survey in 2011-2012 (wave 1), the first follow-up survey in 2013-2014 (wave 2), and the second follow-up survey in 2015-2016 (wave 3). The Biomedical Ethics Committee of Peking University approved this study, and all participants provided written informed consent.
The current analyses focused on 12338 individuals who were aged 50-75 years and attended the “health status and function” module in the wave 1 survey. Of these, 3375 individuals were excluded for the following reasons: they had self-reported diagnosis of dementia and/or Parkinson’s disease (n = 252), they did not complete all of the cognitive tests (n = 2586) or they had cognitive impairment  (defined as a global cognitive score < 5 [1.5 SD below its mean], n = 537) at baseline. An additional 759 individuals were excluded because they were lost to follow-up from waves 2 to 3. The remaining 8204 participants (4289 males and 3915 females) with complete baseline data and at least one reassessment of cognitive function (waves 2-3), were included in the analyses reported here (Fig. 1).
Social and intellectual activities
In the “health status and function” module of CHARLS, four social activities (interacting with friends; going dancing, exercising, or practicing Qigong; participating in community-related organizations; and doing voluntary charity work or assisting others) and four intellectual activities (playing Mahjong, cards or chess; attending an educational or training course; investing in stock; and surfing the internet) in the past month were assessed. The frequency of each activity was rated as never (score = 0), not regularly (score = 1), almost every week (score = 2), or almost daily (score = 3). These activities were assembled to a sum score based on the frequency level (score 0-3). Thus, the total scores for social and intellectual activities could range from 0 to 12 points and were categorized as 0, 1-2, and ≥ 3.
In accordance with previous studies [20, 21], cognitive function was calculated using two categories: episodic memory and mental intactness. The word recall test evaluated episodic memory. Examiners read a list of 10 random words, and participants were instructed to recall as many words as possible immediately afterward (immediate recall). The number of correctly recalled words was scored and indicated the participant’s immediate recall. Ten minutes later, the participants were asked to recall the same list of words (delayed recall). Episodic memory scores were calculated as the average number of immediate and delayed word recalls and ranged from 0 to 10. The mental intactness based on some components of the mental status questions of the Telephone Interview of Cognitive Status (TICS) battery established to capture intactness or mental status of individuals. In CHALRS, mental status questions included serial subtraction of 7 from 100 (up to five times), the date (month, day, and year), the day of the week, the season of the year, and intersecting pentagon copying test. Answers to these questions are summed into a mental intactness score that ranges from 0 to 11. Global cognitive scores were calculated as the sum of the scores of episodic memory and mental intactness and ranged from 0 to 21.
Baseline measurements of age, sex, education level, marital status, location of residence, household income level, smoking, drinking, self-report of health, physician-diagnosed chronic diseases, restriction, self-reported visual and hearing impairments, depressive symptoms, and body mass index (BMI) were included as covariates in the current analyses. Educational level was categorized as “no formal education”, “primary school”, “middle school”, or “high school or above”. Marital status included “married” and “others”. Location of residence was divided into “rural” and “urban”. Household income was categorized into tertiles and coded as “low”, “medium”, and “high”. Self-perceived health status was reported as “good”, “fair” or “poor”. Current smoking and drinking status were assessed by self-report based on the questions “Do you currently smoke?” and “Do you currently drink alcohol?”. Hypertension was defined as a systolic blood pressure ≥ 140 mm Hg and/or a diastolic blood pressure ≥ 90 mm Hg, use anti-hypertensive drugs, or self-reported history of hypertension. Diabetes mellitus was defined as a fasting blood glucose ≥ 126 mg/dL, HbA1c ≥ 6.5%, or current use of anti-diabetic therapy, or self-reported history of diabetes mellitus. Dyslipidemia was defined as a total cholesterol ≥ 240 mg/dL, or current use of lipid-lowering therapy, or self-reported history of dyslipidemia. Chronic kidney disease was defined as an estimated glomerular filtration rate < 60 mL/min/1.7 m2 or self-reported history of chronic kidney disease. Other physician-diagnosed chronic diseases, including heart diseases, stroke, chronic lung disease, arthritis, and cancer, were self-reported. We defined comorbidity as 0, 1, or at least 2 according to the number of nine chronic diseases that the participant had. Restriction was defined as having limitations in any of the five activities of daily living, including bathing, dressing, eating, getting into/out of bed, and toileting . Depressive symptoms were assessed using the 10-item version of the Center for Epidemiologic Studies Depression Scale, and a score of ≥ 10 indicated the presence of depressive symptoms . BMI was calculated as the weight in kilograms divided by the square of the height in meters and was categorized as follows: <18.5, 18.5-23.9, 24.0-27.9, and ≥ 28.0 kg/m2 .
The primary outcome was the trajectory of global cognitive scores, and the second outcomes were the trajectory of episodic memory and mental intactness scores. We first performed a multiple regression equation adjusting for age, sex, and education to obtain the predicted cognitive scores, and then we used the following equation to calculate the adjusted Z scores: , Y is the raw cognitive score, is the predicted population mean score, and RMSE is the root mean square error of the regression equation . We used this method to transform the global cognitive scores and scores for individual cognition domains. The transformed Z scores were used in analyses.
We applied group-based trajectory model (GBTM) implemented through the “traj” plugin procedure in Stata  to identify distinct trajectories of cognitive scores as a function of current age at each visit. GBTM allowed for all available cognitive scores to be included in model estimates under the assumption that missing cognitive scores measures were missing at random. The successive cognitive Z scores were modeled as censored normal . A maximum of six trajectory groups was set a priori. We fitted the models from one group trajectory to six group trajectories, and age in years was used as a time scale. To identify the model with optimal number of distinct cognitive trajectories, we first modeled longitudinal trajectories of cognitive scores by adapting a polynomial model (up to cubic models) to for each of the cognitive outcomes with age as independent predictor. Then, we compared the Bayesian information criteria (BIC) and Akaike’s information criterion (AIC) value to identify the best fitted model. Furthermore, an average posterior probability of assigning each participant to a group of approximately 70% or higher was indicative of a good ﬁt, and models with greater than 5% membership in each trajectory group were selected.
Subsequently, multinomial logistic regression model was used to estimate the association of social and intellectual activities with the trajectories of the cognitive function measures. Odd ratio (OR) and the corresponding 95% confidence intervals (CI) were reported. Multivariable model adjusted for these following covariates: social and intellectual activities scores (0, 1-2, ≥ 3), age at baseline (continuous), sex (male, female), education (no formal education, primary school, middle or high school, college or above), marital status (married, others), residence (urban, rural), household income (low, medium, high), smoking (yes, no), drinking (yes, no), body mass index (< 18.5, 18.5-23.9, 24.0-27.9, ≥ 28.0 kg/m2), self-report of health (good, fair, poor), comorbidity (0, 1, ≥ 2), depressive symptoms (yes, no), restriction on activities of daily living (yes, no), visual impairment (yes, no), and hearing impairment (yes, no).
Given that trajectory analysis is more stable for participants with 3 or more observations over time, we conducted a sensitivity analyses by included participants with all three waves of cognitive function measures. All analyses were performed with Stata version 15.1 (StataCorp, College Station, TX). A two-sided p-value less than 0.05 was considered statistically significant.