Department of Pediatrics, Division of Behavioral Medicine, State University of New York at Buffalo, Farber Hall, Room G56, 3435 Main Street, Building 26, Buffalo, NY 14214-3000. E-mail: email@example.com
Objective: To determine whether interpersonal stress reduces youths’ motivation to exercise in a laboratory setting.
Research Methods and Procedures: Physical activity and sedentary behavior were measured in boys and girls across a control day, after reading children's magazines, and on a stress day, after giving a videotaped speech. For one analysis, children were divided into low (n = 12) and high (n = 13) heart-rate reactivity groups based on changes in heart rate to stress. In a second analysis, children were divided into low and high perceived level of stress based on changes in perceived stress. To determine differences in choice of exercise or sedentary behavior across the control and stress conditions, subjects chose either to exercise for progressively longer periods to earn a monetary reinforcer or to engage in a high-rated sedentary behavior.
Results: The choice to exercise was influenced by stress reactivity differently in the stress and control conditions. Low heart-rate reactive children participated in similar (p > 0.50) amounts of exercise on the stress and control days, but high heart-rate reactive children participated in less (p < 0.01) exercise (22.0 ± 2.5 vs. 26.3 ± 2.2 minutes) on the stress than control days. When grouped by change in perceived stress, there were no group differences, but subjects exercised longer (p < 0.01) on the control day than the stress day.
Discussion: Interpersonal stress decreased exercise in children susceptible to interpersonal stress. Stress-induced alterations in health behaviors may lead to weight gain in children.
Frequent participation in physical activity by youths promotes many positive health benefits, including bone mineral accrual (1), a favorable blood lipid profile (2), aerobic fitness (3), muscular strength (4), and weight control (5, 6). Exercise also reduces stress among youths (7). Moreover, childhood is a critical time for developing lifelong physical activity behaviors. Children who engage in greater amounts of physical activity continue to have greater physical activity in adulthood (8, 9). The many beneficial aspects of exercise in youths and the critical developmental period of childhood for establishing healthy exercise habits highlight the importance of understanding the moderators of physical activity participation in children.
In adults, stress has been found to have a detrimental influence on health behaviors (10, 11), including reduced adherence to programmed (12, 13, 14) and nonprogrammed (15, 16) physical activity participation, although these findings have not been consistent (17, 18). Stress and physical activity are also inversely related in studies of children and adolescents (19, 20), but these studies have been limited by their correlational nature. Interpersonal stress may not only decrease healthy behaviors, but also increase participation in unhealthy behaviors (14). For instance, we have shown in a laboratory study that stress increases snacking (21). The aim of this study was to determine whether interpersonal stress would shift the choice of sedentary and physically active behaviors in children who differed in their stress reactivity.
Research Methods and Procedures
Eight- to 12-year-old boys (n = 14) and girls (n = 11) served as subjects. The subjects were not preselected based on adiposity and had no conditions that would limit or be contraindications to physical activity. Parents gave written informed consent for their children to participate in this study. Children gave their written assent to participate. The University at Buffalo School of Medicine Institutional Review Board approved the study. Subjects were informed that the purpose of the study was to determine whether giving a speech influenced the activity choices of children. An investigator debriefed subjects and their parents after completing the experiment.
Each child was tested on 2 days, with the order of stress and control conditions randomized across subjects. Children reported to the Behavioral Medicine Research Laboratory at the same time for both visits. They were instructed not to eat for 1 hour before coming to the laboratory. At the beginning of session one, the children completed a questionnaire to determine their adequacy in and predilection for physical activity (22) and completed an interview to assess their physical activity over the past 7 days (23). At the beginning of both sessions, children rated their perceived stress and put on a heart rate monitor (Polar Vantage XL; Polar Vantage, Port Washington, NY). A baseline heart rate was recorded after the subjects sat quietly for 5 minutes. Subjects were then instructed how to use a television, a video player, and a video game machine that were placed in the testing room. After watching a video of their choice and playing a video game for 2 minutes each, children rated their liking for watching television and playing video games. Next, the children were instructed on the safe use of a cycle ergometer, and an investigator demonstrated the use of the ergometer. The children practiced riding the ergometer for 2 minutes and rated their liking for the activity. The practice exercise session was completed on both visits to the laboratory to ensure that children participated in equal amounts of activity before testing and to ensure that both conditions would last equal amounts of time.
For the next 20 minutes, subjects rested by reading children's magazines, and then the experimenter recorded the heart rate and rating of perceived stress. In the control condition, subjects were asked to read children's magazines or color pictures for an additional 20 minutes. In the stress condition, subjects were given 15 minutes to prepare and 5 minutes to deliver a videotaped speech about what would make them a good friend. Participants were led to believe that the general quality (good or bad) of their speech would be judged by a group of their peers. Heart rate and rating of perceived stress were obtained immediately after the second reading period or the speech. For the next 30 minutes, children participated in a protocol that provided them with the choice to be active or sedentary. At the end of the second visit, height, weight, and anthropometric measurements were obtained.
Physical Activity Predilection
Predilection measures the likelihood that a child will choose to be physically active rather than sedentary when given the choice (22). The Children's Self-Perceptions of Adequacy in and Predilection of Physical Activity scale has a test-retest validity of 0.78 to 0.91 among the adequacy and predilection factors for school grade groups from fourth to eighth grade. Validity was established by correlating with performance on tests of motor proficiency (r = 0.70 to 0.82).
Physical Activity Recall
At baseline, subjects were interviewed about their time spent in physical activities of various intensities with the 7-day Physical Activity Recall (23). All subjects recalled the time spent in sleep [1 metabolic equivalent (MET)1] and in moderate (4 METs, strenuous as walking), hard (6 METs, an intensity between walking and running), and very hard (10 METs, strenuous as running) activities. The greatest amount of time each day was spent in light activities (1.5 METs, activities less strenuous than walking, not including sleep) and was determined by subtraction. Before the interview, the subjects were given instructions on how to properly answer the interviewer's questions. The interview involved the subjects’ reporting activities from the previous seven days, beginning with the day before the interview and working back the next six days. To aid recall, the subjects were prompted about activities in the morning, afternoon, and evening of each day. The questionnaire has a test-retest reliability of 0.47 to 0.99 and inter-rater reliability of 0.78 to 0.86 and correlates of 0.71 with activity log and 0.61 with Vo2max (23).
Stress responses were determined with a physiological (heart-rate reactivity) and psychological (perceived stress) measure. Using a Polar Vantage XL (Polar Vantage) heart rate monitor, heart rate was recorded as described above. Using a 100-mm visual analog scale, anchored by “very relaxed”/“very stressed (worried),” perceived stress was measured by having subjects assess their current degree of psychological stress. Ratings were measured to the nearest millimeter. Subjects were divided into high and low reactors based on a median split of the change in heart rate and based on a median split of the change in perceived stress scores between the end of the 20-minute reading period and immediately after giving the speech during the stress condition.
Children rated their liking for watching videos of popular cartoons, playing video games, and riding a cycle ergometer with a 100-mm visual analog scale, anchored by “do not like at all”/“like very much.”
Children were instructed that they could choose to ride a cycle ergometer and either watch television or play video games, depending on which sedentary behavior they rated the highest, whenever they wished for the next 30 minutes, but that they would earn money at a faster rate by exercising than by being sedentary. Children exercised at 50% of heart-rate reserve. The following progressive ratio schedule was used to increase the amount of exercise that needed to be completed before a child earned a reinforcer: $3.00 for completing 1 minute of exercise; $3.00 for completing an additional 2 minutes of exercise; $3.00 for an additional 3 minutes of exercise; $3.00 for completing an additional 4 minutes of exercise; $3.00 for an additional 5 minutes of exercise; $3.00 for completing an additional 6 minutes of exercise; and $3.00 for an additional 9 minutes of exercise. Children earned $3.00 for every 10 minutes of sedentary behavior. If subjects began exercising and then stopped to watch television, the elapsed time for exercise stopped and the elapsed time for television began. We used short cartoons with a length of 7 to 10 minutes, so that children could watch the entire cartoon within a 10-minute block and then make another choice to exercise or watch cartoons. When subjects started to exercise again, the exercise time began at the point it was stopped. Children were informed that they would earn the monetary reinforcer for a stage only if they completed the entire stage. To demonstrate changes in physical activity between the control and stress days, the children had to participate in some physical activity on the control day. However, sedentary activities are very reinforcing; therefore, it is difficult for physically active alternatives to compete. We chose to increase the reinforcing value of physical activity by allowing children to earn more money by being physically active than sedentary.
To limit the effect of fitness on exercise time, exercise intensity was maintained at a constant moderate intensity, and exercise time was progressively increased, rather than maintaining the exercise time and progressively increasing the exercise intensity. The outcome measures were exercise energy expenditure and exercise time. To control for attention when the children were exercising or watching television, the investigator monitored the subjects by recording their heart rate and asking them how they were feeling during exercise and by recording heart rates and sitting with the children when they chose to watch television or play video games.
Anthropometrics and Body Composition
Body weight was measured to the nearest 0.05 kg with the subjects wearing shorts and a light shirt. Height was measured with a stadiometer (SECA, Hanover, MD). BMI was calculated according to the following formula: BMI = kilograms per meters squared. BMI percentile was calculated in relationship to 50th BMI percentile for children based on their sex and age. Subscapular, triceps, suprailiac, abdominal, thigh, and mid-calf skinfolds were measured with a skinfold caliper (Lange, Beta Technology Incorporated, Cambridge, MD) by a single trained investigator. The recommendations of Lohman et al. (18) were followed relative to landmarks and methods. Each anthropometric variable was measured three times, and the median score was used as data to avoid the effects that an outlying measure may have on the mean score. Body composition was estimated from skinfolds using equations validated against a four-compartment body composition model in children of the same age that we studied (24). The prediction equation has reliability values of R2 = 0.77 and SE of the estimate = 3.9% fat.
ANOVA was used to test between-group differences in subject characteristics based on median splits of either heart-rate reactivity (change in heart rate) or change in perceived stress. Exercise energy expenditure and exercise time were each evaluated by two-factor ANOVAs, with heart-rate reactivity (high/low) or change in perceived stress (high/low) as a between variable and condition (stress/control) as a within variable. Exercise energy expenditure and exercise time were also evaluated by three-factor ANOVAs, with stress response (high/low heart-rate reactivity or high/low change in perceived stress) and adiposity (high/low based on median split of the percentage body fat) as between variables and condition (stress/control) as a within variable. Thus, separate three-way ANOVAs were run using either heart-rate reactivity (high/low) or change in perceived stress (high/low) as a between variable. Changes in stress response over time were evaluated using separate three-factor ANOVAs, one using heart-rate reactivity (high/low) and the other using change in perceived stress (high/low) as a between variable and condition (stress/control) and time (baseline, after 20 minutes of reading, after speech preparation or reading, after speech delivery or reading) as within variables. ANOVA models followed by a priori simple effects were used to test group differences in physical characteristics, exercise energy expenditure, exercise time, and sedentary behavior time in response to a stress. Correlation analyses were used to determine the relationships between heart-rate reactivity or change in perceived stress and exercise behavior.
Physical characteristics of the subjects when grouped by a median split of heart-rate reactivity are shown in Table 1. The mean visual analog scores for liking for watching videos, playing video games, and cycle ergometry, physical activity, predilection score, and habitual physical activity hours of the groups are shown in Table 2. There were no differences (p > 0.05) in physical characteristics, visual analog scores, predilection, or self-reported physical activity when grouped by sex or by median splits of heart-rate reactivity or change in perceived stress, except for liking of playing video games, which was 53% greater (F(1, 23) = 4.97, p < 0.05) in boys than girls.
Table 1. Physical characteristics of the subjects when grouped by a median split of heart rate reactivity*
Low reactive (n = 12)
High reactive (n = 13)
There were no significant group differences when grouped by sex or by median split of the change in perceived stress.
Data are mean ± SD.
10.1 ± 1.2
10.1 ± 1.6
138.6 ± 8.5
141.0 ± 10.1
36.8 ± 13.9
35.5 ± 7.2
18.7 ± 4.8
17.7 ± 2.2
59.0 ± 28.8
57.8 ± 28.0
Percent body fat
23.8 ± 10.4
24.4 ± 7.4
Table 2. Visual analog scores of subjects’ liking for watching videos, playing video games, and cycle ergometry, physical activity predilection score, and physical activity hours of the subjects when grouped by a median split of heart rate reactivity*
Low reactive (n = 12)
High reactive (n = 13)
There were no significant group differences.
Greater value indicates greater liking.
Physical activity time = sum of moderate, hard, and very hard activity hours.
Stress responses as measured by a physiological measure, heart-rate reactivity, and a psychological measure, change in perceived stress, are shown in Figures 1 and 2. There was a significant two-way interaction of heart-rate reactivity group by time (F(3, 69) = 3.87, p < 0.05), as shown in Figure 1. For the low heart-rate reactive group, the heart rate did not change across time. However, the mean heart rate of the high heart-rate reactive group increased by 12 beats/min (11% increase) from the time of preparing the speech to immediately after giving the speech (p < 0.005). In addition, there was a main effect of time (F(3, 69) = 6.49, p < 0.001), but there were no main effects or interactions of condition (control, stress) with heart-rate reactivity.
There was a significant three-way interaction of perceived-stress group by condition by time (F(4, 92) = 5.24, p < 0.01), as shown in Figure 2. In the control condition, the high perceived-stress group remained stable while reading magazines and increased (p < 0.01) by 3.0 cm (1.4-fold increase) after the exercise/television watching choice protocol. In the stress condition, the perceived stress scores of the high perceived-stress group decreased (p < 0.001) by 1.4 cm (47%) between baseline and completing the 20-minute reading period, increased by 2.8 cm (1.8-fold increase) after preparing (p < 0.001), and increased another 3.0 cm (68%) after giving (p < 0.001) the speech. The perceived stress was reduced (p < 0.001) by 3.2 cm (43%) after the exercise/television watching choice protocol. The perceived stress of the high perceived-stress group was greater in the stress condition than in the control condition after preparing the speech (p < 0.001, 91% difference) and after giving the speech (p < 0.001, 2.5-fold difference). In the control condition, the perceived stress of the low perceived-stress group decreased 0.5 cm after reading magazines (p < 0.02) and increased (p < 0.05) by 1.8 cm after the exercise/television watching choice protocol. The perceived stress of the low perceived-stress group was 1.5 cm greater in the stress condition than in the control condition after preparing the speech (p < 0.01), but this was caused by the reduction in perceived stress in the control condition, because the perceived stress was unchanged across time (p > 0.25) in the stress condition. In addition, there were main effects for group, condition, and time. The perceived stress was greater in the high perceived stress vs. low perceived-stress group (F(1, 23) = 11.74, p < 0.01), stress vs. control condition (F(1, 23) = 16.14, p < 0.001), and after preparing and giving the speech than after reading for 20 minutes (F(4, 92) = 8.29, p < 0.001). There were two-way interactions of the following: 1) perceived-stress group by condition (F(1, 23) = 8.97, p < 0.01), in that perceived stress of the high perceived-stress group was greater than the low perceived-stress group in the stress condition; 2) perceived-stress group by time (F(4, 92) = 3.52, p < 0.01), in that the high perceived-stress group had greater perceived stress than the low perceived-stress group after giving the speech (p < 0.05), but not at other time points (p > 0.05); and 3) condition by time (F(4, 92) = 19.31, p < 0.001), in that the perceived stress was greater after preparing and after giving the speech in the stress condition than in the control condition.
Exercise Energy Expenditure and Exercise Time
As shown in Figure 3, when subjects were grouped by low vs. high heart-rate reactivity to stress, there was a two-way interaction of level of heart-rate reactivity with condition for exercise energy expenditure (F(1, 23) = 4.64, p < 0.05) and exercise time (F(1, 23) = 3.70, p < 0.06). When stressed, high heart-rate reactive children expended less energy (p < 0.001) and exercised for a shorter time (p < 0.01). There was also a main effect of condition, as the subjects had greater exercise energy expenditures (F(1, 23) = 17.27, p < 0.001) and exercise times (F(1, 23) = 8.10, p < 0.01) in the control condition.
When grouped by low vs. high change in perceived stress, as measured by a visual analog scale, there were no significant interaction effects, but there was a significant main effect of condition, as the subjects had greater exercise energy expenditures (F(1, 23) = 14.97, p < 0.001) and exercise times (F(1, 23) = 7.61, p < 0.01) on the control day than the stress day (Figure 4).
Adiposity did not interact with stress to influence the choice of being active/sedentary. When grouped by low/high heart-rate reactivity to stress or low/high change in perceived stress and low/high percentage body fat, there were no significant (p < 0.10) main effects for adiposity, two-way interactions of adiposity by stress group or adiposity by condition, or three-way interaction of adiposity by stress group by condition.
The changes (stress − control) in exercise energy expenditure (r = −0.40, p < 0.05) and exercise time (r = −0.50, p < 0.01) were inversely related to the increase in heart rate from the end of the 20-minute reading period to after giving the speech. Correlations between changes in exercise energy expenditure or exercise time and change in perceived stress were not significant (p > 0.15). The change in heart rate and change in perceived stress were directly related (r = 0.44, p < 0.05).
We have shown in this study and in a separate group of children (21) that giving a speech increases the perceived stress and heart rate of some children. In the current study, the stress protocol was used to evaluate the influence of stress on the choice of watching television or exercising. When the subjects were grouped by whether or not they had an increase in stress, the effect of stress to reduce physical activity was marked. When subjects were grouped by change in heart-rate reactivity, those subjects who had greater increases in heart rate caused by giving a speech performed 31% less exercise and spent more time watching television on the stress day than on the control day. When subjects were grouped by change in perceived stress, there was a general effect of stress to reduce physical activity. Both subjects who did and did not report increases in their perceived stress reduced their physical activity by 21% on the stress day. These data clearly demonstrate the importance of stress for decreasing physical activity.
To the best of our knowledge, these are the first controlled laboratory data to show that stress influences the choice to be active or sedentary and reduces physical activity in youths. Some studies have reported no or very modest relationships between questionnaire measures of stress or stress tolerance as a personality trait (25, 26, 27) and the physical activity of youths. Other studies (19, 20) of children have reported inverse relationships between stress, as measured by questionnaires, and physical activity, but interpretation of the results is limited by the correlational nature of the studies. Furthermore, the direction of the relationship could not be determined because stress may have reduced physical activity or greater physical activity may have reduced stress. A unique aspect of the present investigation is that we directly manipulated the stress and then used a choice of being physically active or sedentary to find a reliable reduction in children's choices to be physically active when stressed.
A limitation of this study is that it cannot address whether stress reduces only the duration of exercise or also other aspects, such as the intensity and frequency of exercise. It is unclear whether exercise duration is more sensitive to the effect of stress than other physical activity parameters. Academic examination stress has reduced the duration, but not the number, of exercise bouts in young adults (16), whereas increased perceived stress has been associated with reductions in exercise time, number of exercise bouts, exercise enjoyment, and adherence to a physical activity program (12, 14).
Stress may influence energy balance by interrupting a child's choice to participate in healthy behaviors, such as physical activity, while increasing unhealthy behaviors, such as snacking on energy-dense foods. In a previous study of youths with similar characteristics to those in this study (21), we demonstrated that stress increased snacking, particularly in those children with greater dietary restraint. Television watching and snacking are complementary behaviors (28, 29, 30), such that increased television viewing results in increased snacking (31). Children who experience minor stress on a regular basis may, therefore, experience weight gain caused by decreased physical activity, increased sedentary behaviors, and increased snacking. The increase in snacking could occur both as a coping mechanism for stress (21) and as a complementary behavior to the increase in certain sedentary behaviors.
The results show a general influence of stress on choice to be physically active when stress is measured by subjective changes in perceived stress, but a differential influence of stress as a function of individual differences in stress reactivity as defined by the physiological measure of heart-rate changes. It is well known that physiological and subjective responses to stressors may not be concordant (32), and, therefore, it is not surprising that these two methods for assessing stress reactivity are associated with different patterns of physical activity. The increase in heart rate after interpersonal stress was associated with differences in physical activity (Figure 3), and those subjects with the greatest heart-rate reactivity caused by stress had the greatest reductions in exercise time (r = −0.50, p < 0.01). However, children performed less physical activity when stressed regardless of whether they did or did not report increases in perceived stress (Figure 4), and changes in perceived stress were not related (r = −0.19, p > 0.35) to changes in exercise behavior. Investigators studying stress and food intake have shown physiological reactivity to be a better predictor of food intake than subjective changes, which is consistent with the current results (33). Thus, for children, physiological changes caused by stress may be a better marker of who will show a shift in physical activity than the subjective appraisal of the stressor.
We conclude that interpersonal stress influences the choice to perform physical activity and that heart-rate reactivity more reliably predicts reductions in physical activity than changes in perceived stress. Stress not only decreases physical activity, but also increases snacking behavior, and may lead to weight gain. Future research should determine whether stress reduces physical activity and increases snacking in the same children.
This study was sponsored by the NAASO pilot grant program. J.N.R. received a Florida's Pritikin Longevity Center Young Investigator Award in Childhood Obesity.