Examining the effects of light versus moderate to vigorous physical activity on cognitive function in African American adults.

Physical activity (PA) recommendations for older adults often endorse participation in moderate to vigorous intensity (MVPA). However, health disparities are evident such that African Americans engage in lower levels of MVPA, have a higher prevalence of chronic health and cognitive impairments. The purpose of this cross-sectional study was to examine the role of light PA in addition to MVPA and their associations with measures of executive function among African American older adults. One hundred and ten participants (mean age = 64.78 ± 5.7, males = 14) completed measures of cognitive functioning, including the Trail making, Flanker and the N-back tasks. Additionally, participants completed a 6-minute walk test to estimate their cardiovascular fitness and were given an Actigraph accelerometer for 7-days to objectively assess their light and MVPA. Regression analyses controlling for age, fitness and education showed that higher levels of light PA but not MVPA predicted better cognitive performance on the incongruent flanker reaction time (β=-.24), trails B (β=-.24) and 1-back accuracy (β=.28). Both light PA and MVPA predicted faster reaction times on the 1-back and 2-back conditions of the n-back (light PA: β's=-.22-.23; MVPA: β's=-.28). Light PA demonstrated similar or better associations than MVPA with cognitive functions which are known to decline with age. Designing and promoting light PA interventions in African American older adults maybe more feasible given the prevalence of disability and functional health disparities. Intervention studies testing the efficacy and effectiveness of light PA are needed and could have a significant public health impact among aging African Americans.

Keywords: Cognition; exercise; physical activity; minority aging; executive function

African Americans continue to exhibit among the highest rates of chronic disease and premature death in the U.S (Fineberg, [15]). Ample evidence now exists validating that regular physical activity (PA) is key to preventing and managing a number of chronic diseases common to older adults (Committee 2018 Physical Activity Guidelines Advisory, [11]). However, the PA guidelines for older adults continue to emphasize the importance of participation in moderate to vigorous intensity physical activity (MVPA). MVPA, defined as any waking behavior with an energy expenditure ≥ 3 METS (World Health Organization, [40]), has emerged as a promising low-cost method for improving neurocognitive function in older adults. Several meta-analyses of randomized clinical trials RCTs suggest that PA is effective at improving cognitive performance in older adults (Angevaren et al., [2]; Colcombe & Kramer, [10]; Smith et al., [35]) with small-to-moderate effect sizes ranging from.10 to.70 depending on the outcome and duration of PA. However, MVPA can be challenging for older adults, especially for African Americans who experience a disproportionately higher prevalence of obesity, diabetes and cardiovascular disease which can limit physical functioning.

Accumulating evidence suggests that engaging in light PA can confer health benefits after accounting for MVPA. Light intensity activities are defined as >1 and <3 metabolic equivalent task hours. These are particularly important for older adults as they tend to spend a greater portion of their day engaging in light PA than any other age-group (Berkemeyer et al., [4]). Participation in light PA has been associated with reduced risk of all-cause mortality (Saint-Maurice et al., [33]; Amagasa et al., [1]), specifically metabolic syndrome (Chastin et al., [9]), attenuation of arterial stiffening in old age (Gando et al., [16]), increase in mobility for people with chronic disease (Taylor, [37]) and delay in the onset of frailty (Peterson et al., [29]). Although evidence for physical health benefits have been documented, little is known about the role of light PA in improving mental health or cognitive functioning. A cohort study from Australia (Johnson et al., [21]) examined associations between accelerometer based PA and performance on the trail making test among community dwelling cognitively healthy older adults. Their results showed that light PA but not MVPA significantly predicted performance on the trail making part B that assess mental set-shifting. Another cross-sectional study (Kerr et al., [22]) conducted with cognitively healthy older adults residing in continuing care retirement communities examined the effects of low light-intensity PA (< 1040 counts/min or 1.5 to 2.24 metabolic equivalent METs) and high light-intensity PA (1040 – 1951 counts/minute or 2.25 to 2.9 METs) on the trail making test. High light-intensity PA was significantly related to performance on trail making part A, B and B-A in the unadjusted models and high intensity PA, specifically 30 min of MVPA was associated with better cognitive function as assessed by the trail making task. It remains to be determined whether the effect of light PA is domain specific or extends to other dimensions of the cognition, other than the trail making test which mainly requires visuospatial ability (part A) and reflects working memory and task switching abilities (part B) (Sánchez-Cubillo et al., [34]). Additionally, both these studies primarily comprised of Caucasian samples limiting generalizability of the findings to minority populations.

The purpose of this study was to examine the relationship between objectively measured light PA and MVPA and cognitive functioning in a sample of African American middle aged and older adults. In addition to the trail making test, we also assessed performance on two other measures: i) Flanker task of attention and inhibitory control, and ii) N-back working memory task, both of which have been extensively studied in the PA and cognition literature. We hypothesized to observe a significant association of MVPA with the cognitive outcomes. In addition to MVPA we wanted to explore the role of light PA and its association with cognitive function in a sample of African American older adults.

One hundred and ten (96 females, mean age = 64.8 ± 5.7 years) African American adults and older adults were recruited through the Wayne State University's list-serves and the Institute of Gerontology's Healthy Black Elders Center between December 2014 and December 2016. Flyers were advertised at community centers and urban residential facilities around the Detroit-Metro area to recruit participants. Participants were eligible for the study as long as they met the following inclusion criteria: age range 55-75 years, African American, English speaking, ambulatory and willing to visit the university campus to complete assessments. Interested and eligible participants were phone screened and scheduled for a 2-hour study visit to complete the study assessments. Parking was provided and participants were given a $20 CVS Pharmacy gift card for study participation. Prior to data collection, all participants read and signed the informed consent approved by the University's Institutional Review Board. Trained research assistants administered the consent, questionnaires and cognitive assessments.

Participants were then asked to complete a demographic questionnaire documenting their age, income, education, employment, marital status and self-reported health status. The Seca scale and digital stadiometer (Model: Seca 763) was used to record every participant's height and weight. The Telephone Interview for Cognitive Status (de Jager et al., [12]), modeled on the Mini Mental State Exam, was also administered to rule out cognitive impairments using a cut off score of < 21 out of possible 39 points (de Jager et al., [12]; Knopman, [24]).

PA was assessed objectively for all participants. Participants were given an ActiGraph accelerometer (Model: wGT3X-BT, ActiGraph, LLC, Pensacola, FL) and were instructed to wear the monitor during the day (waking hours), on their non-dominant hip for a period of 7 consecutive days. Participants were provided with a log to record the times they wore and took off the device. In line with the existing norms, a minimum of 3 valid days and 10 h of wear time/day was required for a day to be considered valid (Mailey et al., [26]). The National Health and Nutrition Examination Survey (NHANES) cut-points for older adults were used to score the raw data and they were categorized as sedentary (<100 counts/minute), light (101-2019 counts/minute) and moderate to vigorous (>2020 counts/minute) PA (Troiano et al., [38]).

Cardiovascular fitness was estimated using a 6-minute walk test (Enright, [13]). Subjects were instructed to lap around a square hallway at their own pace, while attempting to cover as much ground as possible in the allotted 6 min. Research assistants encouraged subjects with the standardized statements "You're doing well" or "Keep up the good work," but were asked not to use other phrases. Participants were allowed to stop and rest during the test but were instructed to resume walking as soon as they felt able to do so. Research assistants walked a step behind the participants, so as to not set the pace for the test and used a mechanical wheel recording the distance in feet. Resting heart rate and post-test (immediately after the 6-minutes) exercise heart rate was recorded along with the distance covered by the participants. Cardiovascular fitness was estimated based on the distance covered, participant's weight, sex, age and resting heart rate (Burr et al., [8]).

A neurocognitive battery of tests assessing attention and processing speed, executive function and working memory was administered at the study appointment. The tests are described below:

The Trail Making Test (Reitan, [32]) is composed of two parts, A and B. Part A requires mainly visuo-perceptual abilities, whereas part B reflects working memory working memory and task-switching ability, and time B minus time A (B–A) provides a relatively pure indicator of executive control abilities (Sánchez-Cubillo et al., [34]). Part A consisted of 25 circles printed on a sheet of paper. Each circle contained a number from 1 to 25. The participant's task was to connect the circles with a pencil line as quickly as possible, beginning with the number 1 and proceeding in numeric sequence. Part B consisted of 25 circles numbered from 1 to 13 and lettered from A to L. The task in part B was to connect the circles, in sequence, alternating between numbers and letters. The assessors immediately drew attention to any errors, which the participant corrected before proceeding with the test. The scores represent the time taken in seconds to complete each part of the test and the difference between parts A and B, referred to as trail cost or interference. The part B/part A ratio was also calculated and reported.

A modified flanker task (Eriksen & Eriksen, [14]; Pontifex & Hillman, [31]; Gothe et al., [19]) which incorporated arrays of arrows was used to manipulate inhibitory control and selective attention. This was administered on a computer using the E-prime (version 2) software. The congruent trials consisted of the target arrow being flanked by other arrows that faced the same direction (e.g.⋘≪ or ⋙≫). The incongruent trials consisted of the target arrow being flanked by other arrows that faced the opposite directions (e.g.≪>≪ or ≫<≫). Participants were instructed to respond as quickly and as accurately as possible with a left button press when the target arrow (regardless of condition) faced to the left (e.g.'<') and a right button press when the target arrow faced to the right (e.g.'>'). During each of the 3 visits, 1 block of 200 trials, randomized across task conditions was presented. The block consisted of 100 congruent and 100 incongruent trials with left and right target arrows occurring with equal probability. The stimuli were 3 cm tall white arrows, which were presented focally for 80 ms on a black background with a variable inter-trial interval of 1100, 1300, or 1500 ms. Reaction time (average time in milliseconds) and accuracy (% correct responses) on the congruent and incongruent conditions of the task were examined.

Working memory was assessed by a computer administered a modified serial n-back task (Nystrom et al., [28]; Kirchner, [23]) that involved two consecutive phases: 1-back and 2-back. Each phase required the participants to discriminate between a sequence of letters that served as stimuli. In the 1-back condition, the participant was instructed to respond as quickly and accurately as possible if the current letter was the same as the previous trial for the 1-back condition, and two trials previous for the 2-back condition. A practice block of 13 stimuli and five experimental blocks of 20 stimuli each were presented for each of the two conditions. All stimuli letters were in white, small caps, Arial size 72, approximately 3 cm tall, presented one at a time on a computer screen with a black background for a duration of 2000 ms with a fixed 1,500 ms inter-trial interval. Reaction time (average time in milliseconds) and accuracy (% correct responses) on the 1- and 2-back conditions were examined.

Data were analyzed using SPSS 24.0 (IBM Corp., Armonk, NY). Continuous variables are reported as means ± standard deviations and categorical data are presented as observations (n) and percentages in Tables 1 and 2. Prior to testing the regression models, the assumption of normality was assessed by examining residuals scatter plots for the dependent variables. Pearson's product moment correlations were computed for all study variables. Hierarchical linear regressions were conducted to determine the role of sedentary time, light PA, MVPA and cardiovascular fitness in predicting performance on the cognitive function measures [reaction time (RT) and/or accuracy (AC)]. Age and education were included as covariates in all models. Sedentary time did not significantly predict performance on any of the cognitive measures; we therefore report regression models with MVPA and light PA in this paper. All participants completed the trail making test. Two participants reported the accelerometers as misplaced and lost. Four and two participants did not complete the battery of computer based cognitive tests rendering missing data for the n-back task and flanker task, respectively.

Table 1. Participant characteristics for the 110 African American middle-aged and older adults.

1 Note: *One participant chose not to disclose this information.

Table 2. Physical activity and cognitive performance of the 110 participants.

One hundred and ten African Americans completed the study measures. Descriptive data for demographics, PA and cognitive variables are reported in Tables 1 and 2. The mean age was 64.77 years and only 22.72% reported their current marital status as being married. Majority of the participants had an annual income between $25,000 to $50,000 (40%), had attended between 1-3 years of college or technical school (41.8%), and were retired (52.7%). More than half the participants reported being in good health (56.4%), despite the average BMI of 30.87 which is categorized as obese. On average, participants wore the accelerometer for 6.56 days with time spent in light PA being 252.24 min/day and MVPA being 12.25 min/day.

Correlations between age, education, fitness, PA variables and the cognitive task outcomes are presented in Table 3. Both age and fitness showed correlations with the cognitive variables in the expected direction. Older age was associated with poor performance on cognitive variables, i.e.higher RTs and/or lower AC, whereas higher fitness was associated with shorter RTs and/or higher AC on the tasks. Age was significantly negatively correlated with performance on trails A and B time and performance on the 1-back AC and 2-back RT and AC. Fitness on the other hand showed a significant correlation with only the RT variables for flanker congruent condition and 1-back and was marginally significant for the flanker incongruent RT. Education level significantly correlated with RT on the incongruent flanker condition, AC on the n-back conditions and was marginally significant with time taken on trails part B. These correlations were also in the expected direction such that higher education was correlated with higher AC and shorter RT.

Table 3. Correlations of cognitive functions with age, education level, fitness and measures of physical activity.

• 2 Correlation is significant at the 0.01 level (2-tailed).
• 3 Correlation is significant at the 0.05 level (2-tailed).
• 4 Correlation is marginally significant at 0.06 to 0.08 level (2-tailed).
• 5 RT = Reaction Time; AC = Accuracy; Education coded as 1 (partial high school) through 6 (PhD or equivalent).

Correlations between the PA categories and cognitive variables are also reported in Table 3. Sedentary time showed a significant positive correlation with 1-back RT and was marginally significant for trail B time. The correlation was in the expected direction with higher sedentary times correlating with longer RTs on the cognitive tasks. Light and MVPA also showed significant correlations across the three cognitive task outcomes. Both were significantly negatively correlated with trail B time and flanker RTs on congruent and incongruent conditions, such that participants who engaged in more light and MVPA demonstrated shorter RTs. For the working memory task, both light and MVPA showed significant or marginally significant correlations with 1-back and 2-back performance including RT and AC. Overall, both light and MVPA showed consistent significant correlations across the three cognitive task outcomes.

Standardized beta coefficients, R2 and R2 change values are reported in Table 4. For each cognitive outcome, we tested two models: Model 1 with age, education, fitness and MVPA; and Model 2 with age, education, fitness, MVPA and light PA to determine whether addition of light PA significantly explained any additional variance over and above MVPA and other predictors for the cognitive outcomes.

Table 4. Standardized Beta Coefficients, R-square and R-square change values from the multiple linear regressions examining the role of age, fitness, MVPA and light PA in predicting cognitive outcomes.

• 6 Note: Model 1 (without light physical activity); Model 2 (with light physical activity); PA = Physical Activity; RT = Reaction Time; Values in bold indicate significance at p <.05.
• 7 Indicates marginal significance (p =.06 -.08).
• 8 No significance for Flanker Congruent Accuracy, and Flanker Incongruent Accuracy was observed.

Age was a significant predictor of performance on the trail A and B, such as older participants required longer times to complete the A and B parts. Fitness did not predict performance on the trail A and B parts, however MVPA was a marginally significant (β = −.21, p =.05) predictor. Addition of light PA in the model rendered MVPA insignificant. Light PA now significantly predicted (β = −.24, p =.02) performance on the trail making part B indicating that participants with higher levels of light PA were faster to complete part B of the trail making test.

Age and fitness did not significantly predict accuracy or reaction time on the congruent or incongruent conditions of the flanker task. MVPA and light PA also did not explain any variance in the flanker congruent reaction time and accuracy. However, for the flanker incongruent RT, MVPA significantly predicted RT (β = −.22, p =.04) such that individuals reporting higher MVPA time were faster to respond on the flanker incongruent trials. Addition of light PA in the model rendered MVPA insignificant and light PA now significantly predicted performance on the flanker incongruent trials (β = −.24, p =.02) such that higher light PA time was predictive of faster RT. There were no significant predictors across the models for flanker accuracy on the congruent and incongruent conditions.

Age significantly predicted accuracy on the 1-back trials and both accuracy and reaction time on the 2-back trails of the n-back task. Fitness only significantly predicted accuracy on the 1-back trials, such that higher fitness levels were associated with higher accuracy on the trials. MVPA was a significant predictor of reaction time on the 1-back (β = −.35, p =.001) as well as 2-back (β = −.34, p =.001) trials such that higher MVPA levels were associated with shorter reaction times. Addition of light PA significantly explained more variance in the models for 1-back and 2-back RT, such that light PA significantly predicted the reaction times on 1-back (β = −.23, p =.02) and 2-back (β = −.22, p =.02) trials, over and above MVPA. For accuracy, MVPA only marginally predicted accuracy on the 1-back (β =.19, p =.06) but not the 2-back condition. Addition of light PA to the model rendered MVPA insignificant. Light PA now significantly predicted accuracy on the 1-back (β =.28, p =.005). Education was marginally significant in predicting accuracy on the 2-back across the two models (β =.18 and.16, p =.06 and.08 respectively). All the standardized beta values were in the expected direction, such that higher light PA and MVPA was associated with faster reaction times and/or better accuracy on the n-back trials.

To our knowledge, this is the first study to examine the associations between accelerometer measured light intensity PA and MVPA with cognition in a sample of aging African Americans. In our analyses, age was negatively associated with performance on the trail making and n-back tasks of attention-task switching and working memory respectively. Fitness levels showed a positive association with the 1-back condition of the n-back working memory task. Consistent with the literature, MVPA was positively associated with performance on the flanker and n-back tasks (Peven et al., [30]; Bherer et al., [5]). Interestingly, after controlling for age, fitness and MVPA, our regression models identified light PA to be positively associated with each of the measures of cognitive function.

Our findings that objectively measured light PA was significantly associated with measures of cognitive function adds to the previous evidence that demonstrates positive associations between light PA and word fluency (Wilbur et al., [39]) as well as set-shifting performance assessed using the trail making test (Johnson et al., [21]). Additionally, light PA based interventions such as yoga have also been found to improve cognitive function including attention and processing speed (Gothe et al., [18]) as well as executive functions (Gothe et al., [17]). Our findings of MVPA association with the cognitive function measures are also consistent with the literature where MVPA was associated with set-shifting in a large sample of older women (Barnes et al., [3]) and has been associated with cognitive benefits following exercise interventions (Northey et al., [27]). These studies have been conducted with healthy community dwelling older adults (without cognitive impairment) and often use accelerometers as an objective measure of PA. Self-reported measures are often largely inaccurate for assessing the quantity of light and moderate intensity unstructured lifestyle PA which are frequently performed by older adults (Brach et al., [6]). The increasing use of objective measurements in PA research is key to analyze the nuances in PA intensities, especially light PA and their relationship to cognitive performance among aging populations.

Although sizable evidence for MVPA-cognition exists for older adults (Colcombe & Kramer, [10]; Smith et al., [35]), most of these studies are site or lab based structured exercise interventions. These structured programs minimize several barriers to exercise for individuals that volunteer for such studies who are mostly Caucasian. It is also known that many older adults find it difficult to independently adopt and maintain MVPA (Brawley et al., [7]). Given the results from our study and other cross-sectional studies examining light PA and cognition (Johnson et al., [21]; Kerr et al., [22]), light PA can be recommended to older adults to displace and reduce sedentary time to gain cognitive health benefits. Rigorous trials that test different doses of light PA and its effects on cognitive health and brain health are needed and may be especially effective for the disproportionate numbers of African American older adults with mobility limitations.

In addition to MVPA, cardiorespiratory fitness has also been associated with improved cognition as well as brain health in older adults (Kramer et al., [25]). A cross sectional study with middle-aged and older African Americans demonstrated that cardiovascular risk factors were important predictors of cognitive function (Grodstein, [20]). Specifically, peak oxygen consumption during exercise (VO2peak) independently predicted 7% to 14% of the variance in verbal memory and executive function measures. Past cross-sectional studies that have examined the role of light PA and cognition did not test or control for cardiovascular fitness (Johnson et al., [21]; Wilbur et al., [39]). Our results only demonstrate a significant effect of fitness on accuracy on the 1-back condition of the n-back working memory task. Although this result was in the expected direction, we did not observe a significant consistent effect of fitness on other cognitive variables. This could be attributed to methodological differences in fitness measurement. In our study, we utilized the 6-minute walk test, a field test to estimate cardiovascular fitness. Most of the literature that has examined the role of cardiovascular fitness and cognition has assessed peak oxygen consumption during exercise, also known as the VO2max or VO2peak which is considered the gold standard measure of fitness. While the 6-minute walk test is a good predictor of oxygen uptake in individuals with cardiovascular disease, it appears to be more precise than VO2max as a measure of mobility and functional fitness among primarily healthy older adults (Steffen et al., [36]). Future studies should use validated and standardized measures to account for cardiovascular fitness before testing the effects of PA intensities on cognitive variables.

To our knowledge, this is the first study to examine the association of light PA and cognition in a relatively large sample of African American older adults. We used well established measures of objective PA measurement and cognitive functioning. Although we used three tests of cognitive function, future studies should examine a more comprehensive battery of neurocognitive measures and possibly some neuroimaging measures to determine neural correlates of PA intensities. The study is cross-sectional and comes with the limitations of assessing measures at a single timepoint. Longitudinal and intervention studies are needed to determine possible associations and causality among PA intensities and cognition. Given the limited resources, we administered only a field test to estimate fitness and gold standard measures such as assessing the VO2max should be used in future studies. It remains to be determined if these associations are unique to African American older adults or apply to all aging adults regardless of racial or ethnic differences. As with many aging studies, our sample was skewed with a large percentage of female participants and a small number of male volunteers.

This is the first study to examine the relationship between objectively measured light and moderate intensity PA with multiple measures of cognitive function in a sample of African American older adults. Light PA appears to have modest associations with cognitive function including domains of attention, set shifting, inhibitory control, and working memory. Exercise neuroscientists need to examine the potential and extent of cognitive benefits conferred by light PA in contrast with MVPA, especially among minority populations that are known to be disproportionally affected by cognitive decline and dementias.

Thank you to Dr. Peter Lichtenberg and Dr. James Jackson for their mentorship as part of the Michigan Center for Urban African American Aging Research. Thank you to all research participants who volunteered for this study and the research staff who contributed to the data collection process.

No potential conflict of interest was reported by the author(s).