Models were adjusted for age, race, sex, educational level, smoking, alcohol use, body mass index, and hypertension. The most active (reference) group had a long-term pattern of moderate to high physical activity and low to moderate television viewing time; the intermediate group had a long-term pattern of moderate to high physical activity and high television viewing time or long-term pattern of low physical activity and low to moderate television viewing time; and the least active group had a long-term pattern of low physical activity and high television viewing time. DSST indicates Digit Symbol Substitution Test; RAVLT, Rey Auditory Verbal Learning Test; circles, odds ratios; and error bars, 95% CI.
Hoang TD, Reis J, Zhu N, Jacobs DR, Launer LJ, Whitmer RA, Sidney S, Yaffe K. Effect of Early Adult Patterns of Physical Activity and Television Viewing on Midlife Cognitive Function. JAMA Psychiatry. 2016;73(1):73-79. doi:10.1001/jamapsychiatry.2015.2468
Sedentary behaviors and physical inactivity are not only increasing worldwide but also are critical risk factors for adverse health outcomes. Yet, few studies have examined the effects of sedentary behavior on cognition or the long-term role of either behavior in early to middle adulthood.
To investigate the association between 25-year patterns of television viewing and physical activity and midlife cognition.
Design, Setting, and Participants
Prospective study of 3247 adults (black and white races; aged 18-30 years) enrolled in the Coronary Artery Risk Development in Young Adults (CARDIA) Study (March 25, 1985, to August 31, 2011). Data analysis was performed June 1, 2014, through April 15, 2015.
Main Outcomes and Measures
We assessed television viewing and physical activity at repeated visits (≥3 assessments) over 25 years using a validated questionnaire. A 25-year pattern of high television viewing was defined as watching TV above the upper baseline quartile (>3 hours/d) for more than two-thirds of the visits, and a 25-year pattern of low physical activity was defined as activity levels below the lower, sex-specific baseline quartile for more than two-thirds of the of the visits. We evaluated cognitive function at year 25 using the Digit Symbol Substitution Test (DSST), Stroop test, and Rey Auditory Verbal Learning Test.
At baseline, the mean (SD) age of the 3247 study participants was 25.1 (3.6) years, 1836 (56.5%) were female, 1771 (54.5%) were white, and 3015 (92.9%) had completed at least high school. Compared with participants with low television viewing, those with high television viewing during 25 years (353 of 3247 [10.9%]) were more likely to have poor cognitive performance (<1 SD below the race-specific mean) on the DSST and Stroop test, with findings reported as adjusted odds ratio (95% CI): DSST, 1.64 (1.21-2.23) and Stroop test, 1.56 (1.13-2.14), but not the Rey Auditory Verbal Learning Test, adjusted for age, race, sex, educational level, smoking, alcohol use, body mass index, and hypertension. Low physical activity during 25 years in 528 of 3247 participants (16.3%) was significantly associated with poor performance on the DSST, 1.47 (1.14-1.90). Compared with participants with low television viewing and high physical activity, the odds of poor performance were almost 2 times higher for adults with both high television viewing and low physical activity in 107 of 3247 (3.3%) (DSST, 1.95 [1.19-3.22], and Stroop test, 2.20 [1.36-3.56]).
Conclusions and Relevance
High television viewing and low physical activity in early adulthood were associated with worse midlife executive function and processing speed. This is one of the first studies to demonstrate that these risk behaviors may be critical targets for prevention of cognitive aging even before middle age.
Increasing observational evidence and results from randomized clinical trials provide support for the beneficial effects of physical activity on cognitive function.1- 3 However, global surveillance data suggest that overall levels of physical activity are declining,4 and the percentage of adolescents meeting recommended guidelines for physical activity is low, with even more failing to maintain these levels in adulthood.5,6 Over the life course, levels of physical activity tend to peak in childhood with large declines occurring during early adulthood in conjunction with major life transitions, such as entrance into the workforce and parenthood.6- 9 Although early adulthood is a critical period for establishing life-long physical activity habits,10 few studies have investigated the association between physical activity in early adulthood and cognitive function later in life.11- 13
Coupled with the increasing prevalence of sedentary or screen-based activities, such as watching television, these trends are of particular concern for upcoming generations of young people.14,15 In middle-aged and older adults, low levels of physical activity and high sedentary behavior are associated with cognitive impairment and dementia,16- 18 but the long-term risks associated with these lifestyle behaviors require further investigation. In particular, the longitudinal association of television viewing with cognitive function is not well defined. Understanding the effects of these modifiable risk factors across the life course could be crucial for the development of effective population-based strategies for optimal cognitive aging.
The objective of this study was to examine the association between long-term patterns of physical activity and television viewing time during young adulthood with cognitive function in midlife. We hypothesized that patterns of low physical activity and high television viewing time will be associated with worse cognitive impairment at midlife.
Young adults (aged 18-30 years) were recruited from population-based samples of 4 US cities (Birmingham, Alabama; Chicago, Illinois; Minneapolis, Minnesota; and Oakland, California) and enrolled in the Coronary Artery Risk Development in Young Adults (CARDIA) Study. The study was conducted from March 25,1985, to August 31, 2011. Between 1985 and 1986, the baseline examination (year 0) was completed for 5115 participants. Follow-up examinations were completed every 2 to 5 years over 25 years: 1987-1988 (year 2), 1990-1991 (year 5), 1992-1993 (year 7), 1995-1996 (year 10), 2000-2001 (year 15), 2005-2006 (year 20), and 2010-2011 (year 25). Study protocols were reviewed by institutional review boards at each study site as well as the CARDIA coordinating center at the University of Alabama, Birmingham, and at the University of California, San Francisco. At each examination, participants provided written informed consent. Participants received reimbursement for travel and parking and nominal compensation for participation in the in-person examinations (only). Further details of study recruitment and design are available elsewhere.19,20
Of the 5115 participants, 3499 completed the year 25 visit. To assess long-term patterns of physical activity and television viewing time, we included participants with at least 3 assessments of physical activity, 3 assessments of television viewing time, and a cognitive assessment at year 25. The final analytic cohort included 3247 participants. Participants not included in the analytic cohort had a lower educational level and were more likely to be black, male, and smokers and have diabetes mellitus (P < .05).
At baseline and each follow-up visit, the Physical Activity History Questionnaire, which provides a reliable measure of habitual physical activity,21,22 was used to assess participation in 13 types of vigorous-intensity (running or jogging; racquet sports; bicycling; swimming; exercise or dance class; job lifting, carrying, or digging; shoveling or lifting during leisure; and strenuous sports) and moderate-intensity (nonstrenuous sports, walking and hiking, golfing and bowling, home exercises or calisthenics, and home maintenance or gardening) physical activities in the past 12 months. Based on the duration of participation (2-5 hours/week) and intensity level (3-8 metabolic equivalents) of each activity, a total activity score was calculated as measured in exercise units. Further details on the questionnaire and scoring have been published.22- 24 Previous CARDIA analysis24 indicated that a cut point of 250 exercise units has 97.1% specificity and 70.4% sensitivity for meeting recommended guidelines of 150 minutes of moderate-intensity activity per week. A long-term pattern of low physical activity over 25 years was defined as reporting levels below the bottom, sex-specific quartile (males, 280 exercise units, and females, 148 exercise units) of baseline physical activity levels for more than two-thirds of the visits. Participants not meeting these criteria were categorized as having long-term patterns of moderate to high physical activity. A long-term pattern of very low physical activity over 25 years was defined as reporting activity levels of less than 50 exercise units for more than two-thirds of the visits.
At years 5, 10, 15, 20, and 25, participants were asked the average number of hours per day spent watching television in the past 12 months. A long-term pattern of high television viewing time over 25 years was defined as reporting activity levels within the top quartile of the year 5 (baseline) visit (>3 hours/day) for more than two-thirds of the visits. All other participants were categorized as having a low to moderate pattern of television viewing time.
Based on patterns of physical activity and television viewing time, a categorical variable was created to examine the combined effects of both behaviors. Participants were categorized into 3 groups: (1) most active (reference), including moderate to high physical activity and low to moderate television viewing time; (2) intermediate, including moderate to high physical activity and high television viewing time or low physical activity and low to moderate television viewing time; and (3) least active, including low physical activity and high television viewing time.
At year 25, trained interviewers administered a battery of 3 cognitive tests: the Digit Symbol Substitution Test (DSST), which assesses processing speed and executive function (higher scores indicating better cognitive function)25; the Stroop test, which assesses executive function (an interference score was calculated with lower scores indicating better function)26,27; and the Rey Auditory Verbal Learning Test (RAVLT), which assesses verbal memory (the delayed score was used with higher scores indicating better function).28,29 We defined low cognitive performance as a race-specific score that was 1 SD worse than the mean.
We assessed baseline demographic characteristics, cigarette smoking, and alcohol use using self-reports. At baseline, height and weight were measured, and body mass index (BMI) was calculated as weight in kilograms divided by height in meters squared. Hypertension at baseline was defined as systolic blood pressure of 140 mm Hg or higher, diastolic blood pressure of 90 mm Hg or higher, or receiving antihypertensive medication. Diabetes mellitus at baseline was defined as fasting plasma glucose levels of 126 mg/dL or higher (to convert to millimoles per liter, multiply by 0.0555) or use of diabetes medications. Isoelectric focusing and immunoblotting techniques were used to determine the apolipoprotein E4 (ApoE4) phenotype from year 7 blood samples.30,31
Baseline characteristics were compared by patterns of physical activity and television viewing using unpaired, 2-tailed t tests, Kruskal-Wallis tests, and χ2 tests as appropriate. We used logistic regression to determine the association between patterns of physical activity, television viewing, and a combination of the 2 during young adulthood with poor cognitive performance in middle age. Multivariable models were adjusted for age, race, sex, educational level, smoking, BMI, and hypertension. In addition, we assessed interactions with race, sex, obesity, and ApoE4 phenotype, as well as between physical activity and television viewing. We also evaluated associations between patterns of activity and cognitive performance in linear models. The level of significance was set at P < .05. SAS, version 9.4 (SAS Institute Inc) was used for all analyses. Data analysis was performed June 1, 2014, through April 15, 2015.
At baseline (year 0), the mean (SD) age of the CARDIA participants was 25.1 (3.6) years, 1836 (56.5%) were female, 1771 (54.5%) were white, and 3015 (92.9%) had completed at least high school. A total of 528 (16.3%) participants met the criteria for a long-term pattern of low physical activity and were more likely to be black (314 [59.5%]) compared with those reporting moderate to high physical activity (1162 [42.7%]; P < .001). Participants with low physical activity were more likely to have hypertension (25 [4.7%]; P < .001) and had a slightly lower educational level (mean [SD], 13.7 [2.2]; P < .001) and drank less alcohol (10.7 [21.8]; P = .001) (Table 1).
Compared with participants with moderate to high long-term patterns of physical activity, those with low physical activity were more likely to have poor cognitive performance at midlife on the DSST (low, 21.6% vs moderate to high, 14.5%; odds ratio [OR], 1.62; 95% CI, 1.28-2.04) and Stroop test (low, 16.2% vs moderate to high, 12.3%; OR, 1.37; 95% CI, 1.06-1.78) but not on the RAVLT (low, 22.1% vs moderate to high, 19.5%; OR, 1.17; 95% CI, 0.93-1.47) (Table 2). After adjusting for age, race, sex, educational level, smoking, alcohol use, BMI, and hypertension, the association between low physical activity and poor cognitive performance remained significant for the DSST (OR, 1.47; 95% CI, 1.14-1.90). The results of linear models were consistent with cognitive impairment models.
In a sensitivity analysis, we further adjusted for ApoE4, which did not significantly alter the results. In addition, there were no significant interactions with race, sex, obesity, or ApoE4 for the association between physical activity patterns and poor cognitive performance (P > .05 for all). We also investigated patterns of very low or almost no physical activity. Only 53 of the participants (1.6%) reported very low patterns of physical activity, and after multivariable adjustment (reported as OR [95% CI]), associations with the DSST (2.43 [1.23-4.79]), Stroop test (2.02 [1.05-3.89]), and RAVLT (2.01 [1.05-3.83]) were significant.
A total of 353 participants (10.9%) met the criteria for a long-term pattern of high television viewing. This high level was associated with poor cognitive performance at year 25 on all cognitive tests (Table 3). In unadjusted models, the odds of poor cognitive performance were between 1.5 and 2 times higher for participants reporting high levels of television viewing over time compared with the low to moderate viewers, reported as OR (95% CI) (DSST: high, 27.4% vs low to moderate, 14.3%; OR, 2.26; 95% CI, 1.75-2.93; Stroop test: high, 21.4% vs low to moderate, 12.0%; OR, 2.01; 95% CI, 1.52-2.66; and RAVLT: high, 27.1% vs low to moderate, 19.1%; OR, 1.58; 95% CI, 1.23-2.04). After multivariable adjustment for age, race, sex, educational level, smoking, alcohol use, BMI, and hypertension, the association was attenuated but significant for poor performance on the DSST and Stroop test (DSST: OR, 1.64; 95% CI, 1.21-2.23; Stroop test: OR, 1.56; 95% CI, 1.13-2.14). However, high television viewing time was no longer associated with poor cognitive performance on the RAVLT (OR, 1.14; 95% CI, 0.86-1.53). Further adjustment for ApoE4 did not significantly alter the association with the DSST, but the effect size with the Stroop test increased slightly (OR, 1.81; 95% CI, 1.29-2.53). There were no consistently significant interactions with race, sex, obesity, or ApoE4 for the association between television viewing time and cognitive performance. In models that also adjusted for long-term physical activity patterns, effect sizes for both physical activity and television viewing time were only slightly attenuated, and similar patterns were observed. The interaction between television viewing and physical activity was not significant.
When physical activity patterns and television viewing time were combined, 2473 participants (76.2%) reported the most active patterns of moderate to high physical activity and low to moderate television viewing, 667 individuals (20.5%) reported intermediate patterns with low physical activity and low to moderate television viewing or moderate to high physical activity and high television viewing time, and 107 participants (3.3%) reported the least active patterns of low physical activity and high television viewing time. In adjusted models (results reported as OR [95% CI]), participants with the least active patterns of physical activity and television viewing were more likely to have poor cognitive performance on both the DSST and Stroop test (DSST, 1.95 [1.19-3.22]; Stroop test, 2.20 [1.36-3.56]) (Figure) compared with those who reported the most active patterns. The association with the RAVLT was elevated but not significant (1.39 [0.87-2.22]). Participants reporting intermediate patterns also had elevated odds of poor cognitive performance on the DSST (1.57 [1.23-2.00]).
In this biracial cohort followed for 25 years, we found that low levels of physical activity and high levels of television viewing time during young to mid-adulthood were associated with worse cognitive performance in midlife. In particular, these behaviors were associated with slower processing speed and worse executive function but not with verbal memory. Participants with the least active patterns of behavior (ie, both low physical activity and high television viewing time) were the most likely to have poor cognitive function.
Previous longitudinal investigations of physical activity in midlife and late life support our findings.13,32,33 These earlier studies indicated that low physical activity levels are associated with poor cognitive performance and cognitive decline in older adults. Results from the slightly older British 1946 birth cohort34 suggest that greater leisure time physical activity in adults during their mid-30s contributes to less cognitive decline during middle age (43-53 years). A few other studies11,12 have also investigated the effects of physical activity participation at earlier ages, but these relied on retrospective measures. In one investigation,11 regular early life physical activity was associated with better information processing speed in men but not women; in the second study,12 older women who reported more physical activity during adolescence were also less likely to develop cognitive impairment based on a global test of cognition. Similar results were reported in a small retrospective study35 of postmenopausal women in whom long-term, moderate physical activity beginning in early adulthood was associated with better cognitive function in late life. In CARDIA,13 change in cardiorespiratory fitness in young adulthood was also associated with psychomotor speed at midlife, and a more recent prospective study36 examined the association between leisure time physical activity throughout life and midlife cognitive function. Although the measure of leisure time activity was not standardized and patterns of activity over time were not assessed, the study found that participation in leisure time physical activity (defined as participation in sports as a child or adolescent and regular sport or exercise as an adult) for 4 or more days each week was associated with better memory and executive function compared with no leisure time physical activity.
Studies of physical activity and cognitive function among adolescents and young adults have focused mainly on the acute benefits for cognitive function.37 In small trials38- 40 that have been limited to short time periods, physical activity among young adults improved visuospatial memory, executive function, and processing speed. In cross-sectional, observational studies,41- 43 regular physical activity was also associated with better cognitive function in young adults, but the long-term effects of physical activity on cognitive function during this life stage are unclear.37
Physical activity during young adulthood may preserve cognitive function and contribute to cognitive reserve by increasing neurogenesis as well as synaptic plasticity, particularly in regions associated with executive function and processing speed, but physical activity may also affect other risk factors for cognitive impairment, including cardiovascular risk factors, inflammatory factors, and depressive symptoms.44,45 Observational studies46- 49 in middle and late age have also reported correlations between physical activity and higher measures of total brain and gray matter volume as well as lower β-amyloid levels in the brain. Although some studies50 have reported differences in the association of physical activity with Alzheimer disease risk and amyloid deposition by ApoE4 status, we did not find an interaction between physical activity patterns and ApoE4 phenotype. It is possible that physical activity affects cognition through non–ApoE4-associated pathways, or there may be differences in this association by age.
To our knowledge, this study is also one of the first to report an adverse association between increased television viewing time in early adulthood and midlife cognitive performance. Increasing evidence suggests that sedentary behavior, such as television viewing, is associated with cardiovascular disease, obesity, and mortality in adults,51,52 and television viewing in young adulthood has also been associated with worse cardiometabolic risk factors in middle age.53 In addition, television viewing in middle-aged and older adults has been identified as a risk factor for cognitive impairment,54- 56 but the content and degree of cognitive engagement may be an important aspect of this behavior to consider.18,57
The effects of television watching on cognitive function may involve several complex pathways. Physiologic studies58- 60 suggest that sedentary behaviors, such as television viewing, adversely affect metabolic function by increasing blood pressure as well as lipid and glucose levels. Television viewing may also be associated with different cognitive and social patterns, depression, and poor dietary patterns.61- 63 The association of physical activity and sedentary behaviors with cognitive function may also be bidirectional,64 and reverse causation could contribute to the observed associations.65 Even though we did not perform baseline cognitive testing, given the young age of the participants at study enrollment and the fact that they remained in the study for 25 years, it is unlikely that they had clinically significant cognitive deficits.
The CARDIA Study is a well-characterized, large, diverse cohort with follow-up data on measures of physical activity from more than 20 years. It is one of the first studies to examine the longitudinal association between physical activity and sedentary behaviors and cognitive function in a younger cohort with carefully repeated measures. However, there are a few limitations to consider. Although the retention of CARDIA participants was high over the 25 years of the study, it is possible that there was some selection bias owing to loss to follow-up. Both physical activity and television viewing were self-reported. In addition, we were limited to examining the effects of television viewing and were unable to consider cognitively stimulating sedentary activities, such as video games. Although we assessed the domains of executive function, processing speed, and verbal memory, not all cognitive domains were evaluated.
The findings in this study suggest the need for additional investigation in several areas. In particular, early adulthood may be a critical period to promote physical activity for healthy cognitive aging, especially as physical activity levels during this life stage track with activity levels in later life.10,66 More research is also needed to understand the association between screen-based sedentary behaviors and cognitive function, especially clarification of the mechanisms of this association, exploration of the full range of sedentary behaviors, and possible effects independent of physical activity. Regarding population-based health, the effect of sedentary behavior may be especially consequential because the use of screen-based technologies for work and leisure has increased in the past several decades.4,14
Our results indicate that the lifestyle behaviors in early adulthood that were evaluated in this study could have an effect on the risk of cognitive impairment in midlife and support a potential role for both physical activity and sedentary behavior as modifiable risk factors for prevention. Individuals with both low physical activity and high sedentary behavior may represent a critical target group.
Corresponding Author: Tina D. Hoang, MSPH, Northern California Institute for Research and Education, 4150 Clement St, Veterans Affairs Medical Center, 116-H, San Francisco, CA 94121 (email@example.com).
Submitted for Publication: August 14, 2015; final revision received October 8, 2015; accepted October 11, 2015.
Published Online: December 2, 2015. doi:10.1001/jamapsychiatry.2015.2468.
Author Contributions: Dr Yaffe had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Hoang, Jacobs, Launer, Whitmer, Sidney, Yaffe.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Hoang, Yaffe.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Hoang, Zhu.
Obtained funding: Reis, Jacobs, Launer, Sidney, Yaffe.
Administrative, technical, or material support: Reis.
Study supervision: Reis, Whitmer, Yaffe.
Conflict of Interest Disclosures: Dr Yaffe serves on data safety and monitoring boards for Takeda, Inc and a National Institutes of Health–sponsored study and is a member of the Alzheimer’s Association Medical and Scientific Advisory Council. No other disclosures are reported.
Funding/Support: The Coronary Artery Risk Development in Young Adults (CARDIA) Study is conducted and supported by grants from the National Heart, Lung, and Blood Institute (NHLBI) in collaboration with the University of Alabama at Birmingham (HHSN268201300025C and HHSN268201300026C), Northwestern University (HHSN268201300027C), University of Minnesota (HHSN268201300028C), Kaiser Foundation Research Institute (HHSN268201300029C), and the Johns Hopkins University School of Medicine (HHSN268200900041C). CARDIA is also partially supported by the Intramural Research Program of the National Institute on Aging (NIA) and an intra-agency agreement between the NIA and NHLBI (AG0005). The CARDIA Cognitive Function Ancillary Study is supported by grant HL122658 from the NHLBI. Dr Yaffe is also supported by grant K24 AG031155 from the NIA.
Role of Funder/Sponsor: The NHLBI participated in the design and conduct of the study as well as collection, management, analysis, and interpretation of the data. The manuscript was reviewed by the CARDIA Publications and Presentations Committee for scientific content.