Study design and data sources
This cross-sectional data stemmed from a Southwest China cohort study conducted by West China Hospital, Sichuan University. The natural population cohort study of West China Hospital used cluster sampling, which considered economic development status and geographical region, in the cities and counties where the medical alliance hospital is located to enroll residents aged 20 years or older in Sichuan Province. Baseline health information was collected through a series of questionnaires, physical examinations, biological sample collection and clinical examinations, and then 10 years of follow-up would be conducted, to establish a multidimensional, dynamic and quantitative health big data platform for health prevention in Southwest China.
To date, the cohort has completed the collection of baseline information in the cities of Chengdu, Mianzhu, and Ganzi from 2019 to 2021. Our cross-sectional study included baseline participants who were 45 years old or older and had completed the assessment for cognitive function. The study was approved by the ethical community of West China Hospital, Sichuan University, and all participants signed informed consent forms.
Measurement of leisure-time physical activity
Self-reported LTPA was measured using a standard structured questionnaire, with questions including regularly participating in LTPA during the past 6 months or not, type of LTPA (walking, square dancing , ball games, etc.), frequency of LTPA, and amount of time spent on LTPA. We used metabolic equivalents (METs) to assess the LTPA level of each activity according to the 2011 Compendium of physical activities  (supplementary Table 1), and the product of METs and duration (hours) yields the amount of physical activity. For the total amount of LTPA, we summed the MET-hours per week across all activity types engaged in. For the total duration of LTPA, we summed the hours per week across all activity types.
Light, moderate, and vigorous-intensity activities correspond to 1.1 ~ 2.9 METs, 3.0 ~ 5.9 METs, and ≥ 6 METs, respectively . According to the WHO recommendation, physical activity was defined as at least 150 min/week of moderate-intensity LTPA, 75 min/week of vigorous-intensity LTPA, or an equivalent combination of the two. For convenience of calculation, we assigned moderate intensity LTPA to be 4.5 METs; thus, the recommendation level was at least 11.25 MET-hours/week.
Assessment of cognitive function
The Mini-Mental State Examination (MMSE) [15, 16] was used to evaluate the cognitive function of participants, which comprised 5 domains, including orientation to time and place (10 points), registration (3 points), attention and calculation (5 points), recall (3 points), and language (9 points). The total score ranged from 0 to 30, and a higher score indicated better cognitive function. An MMSE score < 24 was defined as MCI .
Measurement of covariates
Covariates including age, marital status, gender, education level, smoking status, drinking status, diet, and status of diabetes, depression, anxiety, sleep quality, stroke, cancer, and traumatic brain injuries were collected by self-report questionnaires. Fasting blood was collected for routine biochemical detection to evaluate health status, such as diabetes. The status of hypertension, body mass index (BMI) and waist-to-hip ratio (WHR) were assessed by objective measurements. We used the Patient Health Questionnaire-9 (PHQ-9) to evaluate depression status  and the 7-item Generalized Anxiety Disorder scale (GAD-7) to measure anxiety status , with each of the total scores ≥5 as the cutoff point for the diagnosis of minor depression or anxiety . When measuring sleep quality, the Pittsburgh Sleep Quality Index (PSQI) was used , and participants were graded into three groups according to their PSQI scores: good (0–5), not bad (6–10), and general/poor (11–21). BMI was defined as weight in kilograms divided by height squared in meters, and BMI ≥ 24 was regarded as overweight . WHR was defined as waist circumference divided by hip circumference, and WHR ≥ 0.9 (for men) and ≥ 0.85 (for women) were regarded as central obesity .
Descriptive statistics were used to present the characteristics of the participants. Categorical variables were presented as frequencies (composition ratio). Continuous variables were presented as the mean and standard deviation (SD). The correlation between LTPA and cognitive function was examined using linear regression with a robust standard error model. First, we used a univariate model to select variables, with p < 0.1 as a selection criterion. Then, we performed multiple analyses to show the relationship between LTPA and cognitive function after adjusting for potentially confounding variables. Further subgroup analyses were conducted according to sex, age and WHR.
All the analyses in this study were carried out by R 4.1.2 software, and p < 0.05 was considered statistically significant.