Leisure activity and risk of cognitive impairment: The Chongqing aging study

Subjects.

The study was conducted from 1999 to 2004. Residents aged 55 years and older with normal baseline Mini-Mental State Examination (MMSE) scores in the nine randomly selected communities at Chongqing were enrolled into the present study (see inclusion and exclusion criteria on the Neurology Web site at www.neurology.org).

At baseline screening, the following data were collected: 1) demographic data: age, sex, education (lower education [illiteracy and ≤6 years] vs higher education [>6 years]), and occupation (blue collar vs white collar); 2) smoking (current smoking vs no current smoking) and drinking (daily drinking vs no daily drinking); 3) medical conditions: regular use of nonsteroidal anti-inflammatory drugs (NSAIDs) and comorbidities (hypothyroidism, coronary heart disease, chronic obstructive pulmonary disease, chronic hepatitis, chronic renal insufficiency, hypertension, diabetes mellitus, hypercholesterolemia, stroke, and Parkinson disease); 4) neuropsychological evaluation: the cognitive status was assessed using the Chinese version of MMSE with the cutpoint previously defined as 17 (illiteracy), 20 (≤6 years of education), and 24 (>6 years of education) and functional ability and depressive symptomatology were assessed using modified Activities of Daily Living (ADL) and Zung Depression Inventory; 5) leisure activities: the activities were 13 common leisure activities of the elderly Chinese urban people including playing board games (mahjong, chess, or poker), reading (newspapers, magazines, and books), writing, calligraphy or painting, playing music, drama or dancing, watching television, listening to the radio, visiting friends or relatives, traveling, gardening, fishing, and walking and running for exercise. The frequency of participation in a leisure activity was coded as 7 for daily, 1 for weekly, 1/4 for monthly, and 1/52 for annual participation. The score of participation in an activity was calculated as hours spent per week by multiplying participation frequency with average hours spent per time. Composite measures were generated as total hours spent per week in cognitive (playing board games, reading, writing, calligraphy or painting, playing music, drama or dancing, and listening to the radio), physical (walking, fishing, running, and gardening), and social activities (visiting friends or relatives and traveling).

Subjects were followed up annually. Cognitive status was assessed using the same procedure as baseline screening.

Statistical analysis.

In univariate analysis, baseline variables between subjects who developed cognitive impairment and those who did not were compared using Pearson χ² test, Fisher exact test, a t test, or the Mann–Whitney U test as deemed appropriate. In multivariate analysis, onset of cognitive impairment was assigned as the time of follow-up visit at which cognitive impairment was diagnosed. The association between cognitive impairment and the individual activity or composite measure was calculated using Cox proportional hazards regression in three subject groups: Model A (all subjects), Model B (subjects who had a full baseline MMSE score and did not develop cognitive impairment during the first year of follow-up), and Model C (subjects without comorbidity). All the models were adjusted with age, sex, education, occupation, medical conditions, smoking, drinking, depressive symptoms, baseline MMSE, and ADL scores, and participation in other activities.

Follow-up and baseline characteristics.

There was a total of 9,341 residents aged 55 years and older in the communities at baseline, of whom 5,437 (58.2%) were enrolled into the present study. The mean duration of follow-up was 4.7 ± 0.5 years, during which 273 (5.0%) died, 493 (9.1%) dropped out, and 593 (10.9%) developed cognitive impairment. The mean incidence of cognitive impairment was 2.3% per year.

The baseline characteristics of subjects who developed cognitive impairment and those who did not are presented in table 1. Compared with subjects who did not, those who developed cognitive impairment were more frequent in women, lower education, blue-collar occupation, current smoking, daily drinking, and had older age, more comorbidities, lower MMSE and higher ADL scores.

Univariate analysis.

Subjects who developed cognitive impairment had lower scores of playing board games and reading, and a higher score of watching television than those who did not. When activities were grouped into cognitive, physical, and social activity, the composite score of cognitive activity was lower in subjects who developed cognitive impairment than in those who did not; while no statistical differences were found in composite scores of physical and social activities between the two groups (table 2).

Multivariate analysis.

Among individual activities, playing board games and reading were associated with the reduced risk of cognitive impairment, while watching television was associated with an increased risk of cognitive impairment. On the analysis of composite measures, only cognitive activity was associated with the reduced risk of cognitive impairment (Model A in table 3).

Comorbidity and preclinical cognitive impairment have potential impacts on activity participation and the studied association. To control the bias, we verified the studied association in subjects who had full baseline MMSE scores and did not develop cognitive impairment during the first year of follow-up (Model B in table 3) and in subjects who had no comorbidity at baseline (Model C in table 3). The findings were essentially unchanged.