Influence Of Aerobic Exercise Training And Relaxation

Influence of Aerobic Exercise Training and Relaxation Training on Physical and Psychologic Health Following Stressful Life Events DAVID L. ROTH, PHD, AND DAVID S. HOLMES, PHD An experiment was conducted to determine whether aerobic exercise training or relaxation training would be effective for reducing the deleterious effects of life stress on physical and psychologic health. Over 1000 college students were surveyed, and 55 of those who reported experiencing a high number of negative life events over the preceding year were assigned to an aerobic exercise training condition a relaxation training condition, or a no-treatment control condition. Physical and psychologic health were assessed with self-report measures before, halfway through, immediately following, and 8 weeks after the 11-week training (and control) period. Heart rate data collected during a treadmill test indicated that the aerobic exercise training was effective for improving cardiovascular fitness. Psychologic measures indicated that the exercise training condition was more effective than the other two conditions for reducing depression during the first 5 weeks of training. No differences were found among the conditions on self-report measures of physical health. These findings suggest that aerobic exercise training may be useful for reducing the severity and duration of depressive reactions following stressful life change.

The amount of stressful life change that persons report experiencing over a recent time period has been repeatedly found to be associated with the amount of physical and psychologic health problems they report over a subsequent time period (1-5). Currently, there is a great deal of interest in determining the ways in which the deleterious effects of stress can be reduced. Because it is not always practical or even possible to avoid many stressful life events, it seems that one of the most effective ways to reduce the impact of stress on health would be to enhance one's ability to tol-

Address reprint requests to: David L. Roth, Ph.D., Department of Psychology, Campbell Hall, University of Alabama at Birmingham, Birmingham, AL 35294. Received for publication August 1985; revision received October 22, 1986. From the Department of Psychology, University of Kansas, Lawrence, Kansas.

Psychosomatic Medicine 49:355-365 (1987) Copyright © 1987 by (he American Psychosomatic Society, Inc Published by Elsevier Science Publishing Co , Inc 52 Vanderbilt Avenue, New York, NY 10017

erate stress and thereby avoid developing stress-related health problems. One variable that has been proposed as a moderator of the stress-illness relationship is aerobic fitness. The results of a recent study revealed that following stressful life events, those subjects with high levels of aerobic fitness reported fewer physical health problems and tended to report less depression than did those with lower levels of fitness (4). Other comparisons of high- and low-fit persons have indicated that a high level of fitness is associated with reduced physiologic responses to laboratory stressors requiring active coping or cognitive processing (6-9). Physiologic responses to stressors requiring passive coping (e.g., viewing a stressful film, cold pressor task) have not been reliably different for high- and low-fit persons (7, 8), although high-fit persons have been found to show quicker physiologic recovery from stressors requiring either 355 0033-3174/87/J3 50

D. L. ROTH and D. S. HOLMES

active or passive coping responses (10-12). this variability may have prevented the However, because these studies did not emergence of between-group differences. utilize experimental manipulations of fit- Moreover, individual differences in fitness ness, inferences concerning causation must improvement were found to be signifibe considered tentative. cantly correlated with quicker heart rate There are a few experiments that do pro- recovery for those in the exercise group, vide some evidence concerning the role providing some support for the hypothesis that fitness may play in the response to that fitness training alters responses to stress. In one experiment, subjects who psychologic stress. participated in an exercise training class The present investigation was designed showed quicker autonomic recovery, as to experimentally test the influence of measured by electrodermal response, from aerobic exercise training on the physical laboratory stressors requiring active cog- and psychological health of a group of colnitive processing than subjects who par- lege students who reported experiencing ticipated in meditation training or music high levels of life stress. On the basis of appreciation classes (13). An experiment previous prospective studies (1, 3, 4), these conducted on cardiac patients revealed that students were considered to be at risk for those assigned to an exercise condition developing more health problems over the showed greater improvements in cardiac next few months than less stressed stufunctioning and self-concept than those dents. Subjects assigned to the aerobic exassigned to a routine care condition (14). ercise training condition were compared An experiment conducted on moderately on measures of self-reported physical and depressed college women revealed that psychologic health to subjects assigned to those who were assigned to participate in participate in a progressive relaxation an aerobic dance class showed greater de- training condition and to subjects asclines in depression than those who were signed to a no-treatment control condiassigned to a program of self-monitored tion. The exercise training condition conrelaxation activities or to a no-treatment sisted of running and brisk walking when condition (15). These investigators pointed necessary, whereas the relaxation training out, however, that because the women in condition consisted of progressive muscle the exercise condition participated in relaxation and guided imagery techniques. groups and those in the relaxation condition participated individually, social METHODS contact and support may have contributed to the effects achieved in the exercise condition. Finally, a recent experiment comSubjects paring subjects assigned to 10-week aeroThe subjects in this investigation were college stubic training, anaerobic training (weightdents enrolled in General Psychology classes at the lifting), or wait-list control conditions re- University of Kansas. A total of 1051 students indivealed no reliable group differences on cated which events on the Life Experiences Survey heart rate or self-report responses to lab- (LES) (17) they had experienced over the past year, oratory stressors (16). However, these in- and they rated each experienced event as being either vestigators reported large variability in the "good" or "bad" when it happened. Those students reported experiencing a high number of negative degree of fitness improvement in the aero- who life events and who indicated that they were not bic training group, and they suggested that currently participating in either a regular aerobic ex356

Psychosomatic Medicine 49:355-365 (1987)

EXERCISE AND HEALTH ercise training or structured relaxation training program were invited to participate in the experiment. The subjects who agreed to participate were randomly assigned to the aerobic exercise training condition, the progressive relaxation training condition, or the no-treatment control condition. Initially, 23, 21, and 21 subjects were assigned to the three conditions, respectively, but 10 subjects dropped out of the study within the first few weeks, leaving 18 subjects in the exercise training condition (9 males, 9 females), 19 subjects in the relaxation training condition (10 males, 9 females), and 18 subjects in the no-treatment control condition [8 males, 10 females). The ages of the subjects did not differ reliably across the groups. Overall, the subjects had a mean age of 18.9 years (SD = 1.3).

Procedures and Measures Preintervention Assessments. In initial meetings, the procedures to be used in the experiment were explained, and then each subject signed an informed consent statement specific to the particular condition to which he or she had been assigned. Next, each subject was given a Health Record Form (4) to take home and use to record all health problems that occurred between the preintervention and midintervention sessions. On these forms, each subject was to indicate the date(s) of each health problem, provide a brief description of the problem, indicate whether a doctor was consulted about the problem, indicate whether medications were used, and rate the severity of each problem in terms of the degree to which it interfered with normal daily activities (1 = not at all, 2 = a little, 3 = somewhat, 4 = a lot, 5 = very much). Subjects were instructed to record all physical health problems that occurred including minor diseases (e.g., head colds) and transient ailments (e.g., headaches). Next, each subject completed a Health Survey that provided a retrospective measure of physical health problems. On this measure, the subjects were provided with a list of common physical health problems, and they indicated which problems they had experienced over the past month and rated each problem experienced in terms of how much it interfered with normal daily activities on the same scale used on the Health Record Form. Next, each subject completed the LES to examine the replicability of the high-life-stress classification from the initial screening. On this administration of the LES, the subjects also rated the impact of each event experienced on a 4-point scale (0 =

Psychosomatic Medicine 49:355-365 (1987)

no effect, 1 = slight effect, 2 = moderate effect, 3 = great effect), in addition to indicating whether each event experienced was "good" or "bad" when it happened. Finally, to provide measures of psychologic health, each subject completed the Beck Depression Inventory (BDI) (18), the Spielberger Trait Anxiety Inventory (STAI) (19), and the Hopkins Symptom Checklist (SCL-90) (20). Each subject also participated in an individual session in which his or her preintervention level of aerobic fitness was assessed using a Balke treadmill test (21). After weighing the subject on an electronic scale, three electrodes were attached to the subject's chest to provide a modified Lead II monitoring of the subject's cardiac activity. The subject was then instructed to begin walking on a motorized treadmill for which the speed was set at 3.4 miles per hour. The elevation of the treadmill was increased by 1% after each minute until the subject's heart rate reached 90% of the maximum heart rate that would be predicted for a person of that age (e.g., 220 minus age). The grade at which this heart rate was reached was used to provide an estimate of aerobic capacity in milliliters per kilogram per minute according to the conversion tables reported by Londeree (22). In addition, each subject's heart rate at the 10% grade was recorded. Midintervention Assessments. After approximately 5 weeks of the intervention conditions, each subject participated in a midintervention questionnaire session. Subjects first turned in the Health Record Forms that were distributed during the preintervention session, and then they were given blank Health Record Forms to use to record any health problems that occurred between the midintervention and postintervention sessions. The subjects then completed the Health Survey, BDI, STAI, and SCL90 again. In addition, the subjects in the exercise and relaxation training conditions responded to questions regarding a) how effective they had found their training classes to have been up to that point for handling stressful situations, b) how effective they expected their training would be for assisting them in handling the effects of stress in the future, and c) how likely they were to continue to use their training activity after the class ended. All responses were made on 7-point scales (e.g., 1 = not at all effective, 7 = very effective). Postintervention Assessments. Immediately after the conclusion of the 11-week training (and control) period, all subjects participated in sessions in which they turned in the Health Record Forms and 357

D. L. ROTH and D. S. HOLMES completed the Health Survey, BDI, STAI, and SCL90 again. In individual sessions, the subjects were also administered the Balke treadmill test again, with the same procedures and predicted maximum heart rates used. Follow-up Assessments and Debriefing. Approximately 2 months after the postintervention assessments, most of the subjects were met a final time during which they completed the Health Survey, BDI, and STAI once again. (Two subjects from the exercise condition, two from the relaxation condition, and six from the no-treatment condition were not available to participate at follow-up, but comparisons on the data collected up to that point indicated that they did not differ systematically from the other subjects.) After the questionnaires were completed, each subject was given a thorough and elaborate debriefing. Subjects in both the no-treatment condition and relaxation training condition were offered a program of exercise training similar to the one provided to the subjects in the exercise training condition.

Intervention Conditions The exercise and relaxation training programs were conducted in very comparable ways. Both were structured as 1 credit-hour college courses that met for one half-hour each day, 3 days per week, over an 11-week period. Both courses were taught by the same instructor who had been trained in both clinical psychology and exercise physiology. Aerobic Exercise Training. Subjects in the aerobic exercise training condition met at a s-mile indoor track facility. Running and brisk walking were the aerobic exercise activities used in the class. Each subject also participated in warm-up and cool-down stretching exercise each day. Each subject was initially given an individualized exercise prescription that was based on the fitness level determined by the preintervention Balke treadmill test. The prescription was given in terms of a lap pace that would elicit an exercise heart rate that was approximately 75% of predicted maximum. Each subject was trained to monitor his or her own exercise heart rate. Approximately halfway through the training period, the emphasis was changed from one of monitoring exercise intensity to one of monitoring distance. By the end of the 11-week period, almost all of the subjects had at one time or another ran continously for at least 2 miles. A total of 29 formal class meetings were held. The mean number of meetings attended was 23.1 (SD = 3.0, Range = 17 to

358

28). Subjects were encouraged to exercise on their own to make up for holidays and absences in order to maintain a frequency of three exercise sessions per week. Progressive Relaxation Training. Subjects in the progressive relaxation training condition met in a classroom. During the first several meetings, the subjects were taught an abbreviated version of Jacobson's (23) progressive muscle relaxation training procedure (24). Later in the class, other techniques, such as mental imagery procedures (25), were introduced and practiced. Toward the end of the class, very few explicit instructions were provided, and individual flexibility in the utilization of the various techniques taught was encouraged. The relaxation training subjects were taught how to monitor their heart rates so that they could evaluate the effects of their training on their physiologic arousal. The subjects in the relaxation training class were informed that daily practice of these techniques is sometimes recommended by clinical practitioners, and they were encouraged to practice these techniques at home on days that the class did not meet. A total of 29 relaxation class meetings were held, and the mean attendance was 24.8 (SD = 2.4, range = 18-28). No-Treatment Controi. The subjects in the notreatment control condition were not seen except during the sessions when the dependent measures were collected. RESULTS

Life Stress Screening All of the subjects who participated in this investigation reported at least five negative life events on the LES, and that score corresponded to a percentile rank of 82 for the distribution of 1051 respondents who participated in the initial screening. The mean score of 7.2 negative life events for the subjects who participated in the investigation corresponded to a percentile rank of 92. From these results it is clear that all subjects who participated in this study were at least in the upper 20% of this stressful life events distribution. Psychosomatic Medicine 49:355-365 (1987)

EXERCISE AND HEALTH

High-life-stress scores were also observed on the LES administered during the preintervention assessments. The mean number of negative life events reported regressed to 6.6. The mean negative life change score (sum of impact ratings for experienced events designated as "bad") was 12.8. This mean is approximately 1 standard deviation above the means reported in other studies that have used the LES with college students (4,17), and these data further confirm that the subjects in the present study had recently experienced high levels of negative life change. The subjects in the exercise training, relaxation training, and no-treatment conditions did not differ reliably on any of these measures of life stress (all ps > 0.10). Effects of Intervention on Aerobic Fitness Changes in fitness were examined by comparing the heart rate data collected during the Balke treadmill test at pre- and postintervention. Improved estimates of aerobic capacity occurred if a subject was able to walk on a steeper grade at postImprovements in fitness were also evident if the subjects showed lower heart rates at

postintervention than preintervention while walking on the 10% grade. Table 1 contains the means and standard deviations for the three groups on these measures. Although the subjects assigned to the exercise condition tended to have poorer fitness scores initially, the group differences were not statistically reliable (ps > 0.10). To determine whether the aerobic exercise training led to reliable increases in aerobic fitness, analyses of covariance were conducted on the postintervention fitness scores, with the preintervention scores serving as the covariate to remove the effects of the (nonsignificant) initial differences. The analysis on the aerobic capacity estimates yielded a reliable difference among the conditions, F(2,43) = 8.65, p = 0.0007, and pairwise comparisons on the adjusted cell means using the error term from the overall analysis (26) indicated that the subjects in the exercise training condition showed greater improvements in fitness than did the subjects in the relaxation training condition, F(l,43) = 10.82, p = 0.002, and the subjects in the no-treatment control condition, F(l,43) = 14.13, than preintervention before his or her heart rate exceeded 90% of predicted maximum.

TABLE 1. Means and Standard Deviations on Measures of Fitness Post

Pre Condition Measure

Females

Males M

SD

M

SD

Males M

Females SD

M

SD

Exercise training

Aerobic capacity estimate (ml/kg/ min) Heart rate at 10% grade (bpm) Relaxation training Aerobic capacity estimate Heart rate at 10% grade No-treatment control Aerobic capacity estimate Heart rate at 10% grade

47.5

5.1

40.2

4.6

51.8

5.1

44.7

5.2

137.0

15.7

159.0

15.1

128.2

10.8

146.7

16.1

50.2 135.4

5.0 10.5

40.8 155.1

2.4 7.9

50.4 136.4

6.0 16 6

41.1 156.1

3.4 12.3

53.0 133.5

7.1 14.3

40.4 154 2

3.3 14.3

51.3 131.8

2.3 6.4

41.3 153.7

4.2 16.4

Psychosomatic Medicine 49:355-365 (1987)

359

D. L. ROTH and D. S. HOLMES

p = 0.0005 (adjusted Ms = 49.6,45.8, and 45.3 ml/kg/min for the subjects in the exercise, relaxation, and control conditions, respectively, collapsed across subject gender). The relaxation and control conditions did not differ reliably, F(l,43) = 0.22. Analyses on subject gender effects indicated that similar improvements in fitness occurred for both the males and females in the exercise training condition. (It might be noted that male and female subjects were not found to show any statistically reliable differences in the amount of change observed from preintervention levels on any of the measures examined. Males and females were therefore both included together in all reported statistical analyses.) Similar results were obtained from the analysis of covariance conducted on the 10% grade heart rate data. A reliable difference among the conditions was obtained, F(2,43) = 6.66, p = 0.003, and pairwise comparisons indicated that this heart rate decreased more for the subjects in the exercise condition than for subjects in the relaxation condition, F(1,43) = 10.59, p = 0.002, or in the no-treatment condition, F(l,43) = 6.99, p = 0.011 (adjusted Ms = 135.2, 146.0, and 144.0 bpm, respectively). Overall, these results indicate that the exercise training condition led to reliable improvements in cardiovascular fitness. The aerobic capacity estimates suggested that the subjects in the exercise training condition showed approximately a 10% improvement in fitness compared to changes of less than 1% for the other two conditions. Subjective Impressions and Expectations about Interventions One-way analyses of variance on the subjective ratings of the training classes made during the midintervention assess360

ment session revealed that the subjects in the relaxation training condition rated their training as being more beneficial up to that point, F(l,34) = 4.69, p = 0.04, as expected to provide more benefits in the future, F(l,34) = 6.00, p = 0.02, and as something they were more likely to continue using, F(l,34) = 5.00, p = 0.03, than the subjects in the exercise training condition rated their training experience (Ms [SDs] = 3.9 [1.4], 4.4 [1.4], and 4.6 [0.9] for the exercise condition and 4.7 [0.9], 5.4 [1.0], and 5.4 [1.1] for the relaxation condition, respectively). These results indicate that after participating in the training conditions for approximately 5 weeks, the subjects in the relaxation training condition had higher subjective impressions and expectations for the effectiveness of their training than did the subjects in the exercise training class. Influence of Interventions on SelfReported Physical and Psychologic Health Table 2 contains the means and standard deviations for the three groups on the self-reported measures of physical and psychologic health. To examine group differences in the changes observed across time, analyses of covariance were performed on the scores obtained at each administration, with the respective preintervention scores serving as the covariate. The Health Record Form and the Health Survey were used to provide measures of physical health. Total scores were obtained for each administration of each measure by simply adding the severity ratings of all reported problems. The analyses of covariance did not reveal reliable differences among the conditions (all ps > 0.10), and therefore it appears that neither training program had a systematic effect on general physical health. Psychosomatic Medicine 49:355-365 (1987)

EXERCISE AND HEALTH TABLE 2. Condition Measure Exercise training Health record form Health survey BDI STAI SCL-90-CSI Relaxation training Health record form Health survey BDI STAI SCI-90-GSI No-treatment control Health record form Health survey BDI STAI SCL-90-CSI

Means and Standard Deviations on Measures of Self-Reported Physical and Psychologic Health

SD

— 14.4 7.8 40.9 0.6

_

_ 14.9 6.9 39.6

M

SD

M

SD

7.5

5.7 6.9

4.6

12.4

3.1 7.7

36.8

0.4

0.5

2.9 8.3 4.5 8.9 0.5

3.9 8.7 4.7 8.7 0.3

5.6

8.3 6.4

13.6

10.3

37.1 0.5

0.6

—

4.7

4.9 9.0

5.6 13.7 6 8 40.0

0.5

0.4

0.5

_

—

14.4

6.9

8.7

Fol low-up

Post

Mid

Pre

M

9.8

13.8

7.0

69

6.9

41.8 0.6

11.6 0.4

40.5 0.5

7.0 9.1 5 8 10.5 0.4

4.1

14.2

3.1 8.9

M

SD

—

— 7.8

11.2

2.3

33.9 — —

11.2 4.8 36.2

2.2 8.4 —

_

10.1 5.5 10.3

5.4

5.6

39 0

10.4

05

0.4

6.6

5.1

11.9 5.0

7.0 4.3

4.4

4.1

37.0 0.5

9.1 0.4

37.0 —

10.6 —

7.0

3.6

With regard to self-reported psycho- p = 0.07. These results are presented logic health, the analysis of covariance on graphically in Figure 1. the BDI scores revealed a reliable differAn inspection of the amount of imence in depression among the conditions provement observed on the BDI for the at the time of the mid-intervention assess- subjects in the exercise condition indiment, F(2,51) = 3.94, p = 0.03. Pairwise cated that those who were the most decomparisons on the adjusted cell means pressed prior to the intervention showed indicated that the subjects in the exercise the greater improvement (r = 0.89, p < training condition showed greater reduc- 0.0001). This effect is almost certainly due tions in depression from preintervention in part to a floor effect since those with to midintervention than did the subjects relatively low initial levels of depression in the relaxation training condition, F(l,51) had relatively less room for improvement. = 5.95, p = 0.018, and those in the no- Nevertheless, it is interesting to note that treatment control condition, F(l,51) = 6.03, out of the six subjects in the exercise trainp = 0.018, and that the latter two groups ing condition who began the study in the were not reliably different from each other, clinically depressed range (BDI > 10 [27]), F(l,51) = 0.00. There were no reliable dif- five of them (83%) showed an improveferences among the conditions at the post- ment to a clinically normal range (BDI < intervention assessment F(2,51) = 1.03, 10) by the midintervention assessment. but there was a trend that approached staThe analyses of covariance conducted tistical reliability for the group differences on the STAI scores, the Global Symptom to reemerge at follow-up, F(2,41) = 2.91, Index score from the SCL-90, and the subPsychosomatic Medicine 49:355-365 (1987)

361

D. L. ROTH and D. S. HOLMES

RELAXATION

z o

I

6

DISCUSSION

CO CO _

LU 5 DC Q. LJJ

NO-RX EXERCISE

PRE

MID

POST

FO-UP

Fig. 1. Beck Depression Inventory scores for subjects in the aerobic exercise training, progressive relaxation training, and no-treatment control conditions at the preintervention, midintervention, postintervention, and follow-up points.

scale scores from the SCL-90 did not reveal any reliable conditions (all ps > 0.10). From these results, it can be concluded that the effects of the aerobic exercise training on the subjects' psychologic functioning were limited to the BDI measure of depression. Effects Associated with Time The subjects in all three conditions appeared to report decreasing levels of symptoms across time. To test that statistically, a series of groups by trials analyses of variance were conducted on the scores of the three conditions at the various assessment points. Those analyses revealed reliable decreases over time on the Health Record Form, F(l,46) = 5.66, p = 0.02, BDI, F(2,104) = 8.82, p = 0.0003, STAI,F(2,100) = 6.29, p = 0.003, and the SCL-90, F(2,104) = 5.11, p = 0.008. These results suggest 362

that subjects in all of the groups showed some recovery from the effects of stress. This general improvement may have limited the opportunities to observe the full health benefits of the training programs.

The subjects who participated in this experiment repeatedly reported high levels of life stress, and thus it appears that they were the high-risk type of subject for whom concerns about physical and psychologic health have been expressed. With regard to the exercise training, the treadmill tests revealed that the subjects who participated in the exercise training condition showed increases in fitness as assessed by cardiovascular indicators, whereas the subjects in the other two conditions showed little change in fitness. This indicates that the exercise training condition was conducted successfully in terms of improving levels of fitness. With regard to the effects of the interventions on psychologic health, it was found that after the first 5 weeks of training the subjects in the exercise training condition reported reliably greater reductions in depression than did subjects in either the relaxation training condition or the notreatment control condition. Those differences diminished over the next 5 weeks of training, probably owing to the general decline in depression observed over that time span, but the differences tended to reemerge 8 weeks later because subjects in the exercise training condition continued to show declines in depression. The psychologic benefits of the exercise training were limited to depression as measured by the BDI in this study. Differential effects were not found on measures of trait anxiety, for example, and the present investigation is consistent with an earlier one Psychosomatic Medicine 49:355-365 (1987)

EXERCISE AND HEALTH

(16) in suggesting that a relatively brief aerobic training program has no systematic effect on the psychologic symptomatology assessed by the SCL-90. It is noteworthy that the patterning of results concerning depression that were found in this experiment are similar to those found in an earlier study (15). In both studies, exercise training was found to be more effective than relaxation training for reducing depression. Furthermore, in both studies the full effect of this finding was apparent after only 5 weeks of participation. The methodologic improvements introduced in this experiment (e.g., an alternative training program involving comparable amounts of time and contact) and the consistency of the findings across the two experiments provide strong support for the suggestion that regular aerobic exercise is effective for reducing mild to moderate levels of depression. It is also noteworthy that in the present investigation the exercise training had the greatest effect despite the fact that the subjects in the relaxation training condition perceived and expected their training to be more effective for dealing with stressful situations. In light of these findings, it appears that the effects of exercise training on depression observed in the present study cannot be attributed to placebo or expectancy effects. Although it has been suggested by others that frequent aerobic exercise may be an effective treatment for depression in clinical settings (28-30), there appears to be no adequately controlled experimental demonstration of this effect on a patient sample (15, 31). The results of the present investigation should only be generalized to a population of college students who are experiencing high levels of life stress. It should be emphasized that these subjects were recruited to participate in the Psvchosomatic Medicine 49:355-365 (1987)

present study. They were not seeking any treatment on their own. In addition to attempting to replicate these results using clinical samples, future research might also be directed towards examining the possible mechanisms responsible for the antidepressant effects of exercise. For example, aerobic exercise training may affect mood through either biochemical (32, 33) or psychologic (e.,g., enhanced self-concept related to improvements in fitness) means or both. If the mechanism(s) responsible can be specified, then more specific exercise prescriptions aimed at producing optimal effects can be developed. Contrary to the expectations based on previous correlational research (4), the exercise training did not have a reliable influence on subjects' self-reported physical health. One possibility is that the general improvement in health reported by most subjects over the course of the experiment may have obviated the effects of the treatments. Another possibility is that aerobic fitness may serve more of a prophylactic role rather than a corrective one with regard to the influence of life stress on physical illness. The results of the previous research are consistent with the prophylactic role because in that research it was found that subjects who were already high in fitness experienced fewer physical health problems following life stress than did subjects who were low in fitness. In the present experiment, improvements in fitness were introduced after the onset of the stressors and therefore may have come too late. In summary, the results of this investigation indicated that aerobic exercise training was more effective than relaxation training or no treatment for reducing depression following exposure to high levels of life stress. This beneficial effect appeared after only 5 weeks of periodic ex363

D. L. ROTH and D. S. HOLMES

ercise. In contrast, neither exercise training The authors would like to thank Tom nor relaxation training was superior to a R- Thomas and Wayne Osness for making no-treatment condition for reducing phys- possible the aerobic fitness testing and exical illnesses following life stress. It is pos- ercise training condition. Appreciation is sible that frequent aerobic exercise pro- QJso due to B. Kent Houston and Douglas vides an assortment of prophylactic and R- Denney for their comments. This protherapeutic effects with regard to the ef- J e c t WQs supported in part by a dissertafects of life stress on physical and psy- tion fellowship awarded to the first author chologic health, and additional research by the Graduate School of the University seems necessary to further delineate the °f Kansas, nature of these effects.

REFERENCES 1. Cline DW, Chosy JJ: A prospective study of life changes and subsequent health changes. Arch Gen Psychiatry 27:51-53, 1972 2. Dohrenwend BS, Dohrenwend BP: Stressful Life Events: Their Nature and Effects. New York, Wiley, 1974 3. Rabkin JG, Struening EL: Life events, stress, and illness. Science 194:1013-1020, 1976 4. Roth DL, Holmes DS: Influence of physical fitness in determining the impact of stressful life events on physical and psychological health. Psychosom Med 47:164-173, 1985 5. Sarason IG, Sarason BR, Potter EH, Antoni MH: Life events, social support, and illness. Psychosom Med 47:156-163, 1985 6. Holmes DS, Roth DL: Association of aerobic fitness with pulse rate and subjective responses to psychological stress. Psychophysiology 22:525-529, 1985 7. Hull EM, Young SH, Ziegler MG: Aerobic fitness affects cardiovascular and catecholamine response to stressors. Psychophysiology 21:353-360, 1984 8. Light KC, Obrist PA, James SA: Self-reported exercise levels and cardiovascular responses during rest and stress (Abstract). Psychophysiology 21:586, 1984 9. Holmes DS, McGilley BM: Influence of a brief aerobic training program on heart rate and subjective response to stress. Psychosom Med, in press 10. Cantor JR, Zillman D, Day KD: Relationship between cardiorespiratory fitness and physiological responses tofilms.Percept Motor Skill 46:1123-1130, 1978 11. Cox JP, Evans JF, Jamieson JL: Aerobic power and tonic heart rate responses to psychosocial stressors. Person Soc Psychol Bull 5:160-163, 1979 12. Sinyor D, Schwartz SG, Peronnet F, Brisson S, Seraganian P: Aerobic fitness level and reactivity to psychosocial stress: Physiological, biochemical, and subjective measures. Psychosom Med 45:205-217, 1983 13. Keller S, Seraganian P: Physicalfitnesslevel and autonomic reactivity to psychosocial stress J Psychosom Res 28:279-287, 1984 14. Roviaro S, Holmes DS, Holmsten RD: Influence of a cardiac rehabilitation program on the cardiovascular, psychological, and social functioning of cardiac patients. J Behav Med 7:61-81, 1984 15. McCann IL, Holmes DS: Influence of aerobic exercise on depression. J Person Soc Psychol 46:1142-1147, 1984 16. Sinyor D, Golden M, Steinert Y, Seraganian P: Experimental manipulation of aerobic fitness and the response to psychosocial stress: Heart rate and self-report measures. Psychosom Med 48:324-337,1986 17. Sarason IG, Johnson JH, Siegel JM: Assessing the impact of life changes: Development of the Life Experiences Survey. J Consult Clin Psychol 46:932-946, 1978 18. Beck AT: Depression: Clinical, Experimental, and Theoretical Aspects. New York, Harper & Row, 1967 364

Psychosomatic Medicine 49:355-365 (1987)

EXERCISE AND HEALTH

19. 20. 21. 22.

23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33.

Spielberger CD, Gorsuch RL, Lushene RE: Manual for the State-Trait Anxiety Inventory. Palo Alto, CA, Consulting Psychologists Press, 1970 Derogatis LR: SCL-90-R Administration, Scoring, and Procedures Manual. Baltimore, Clinical Psychometric Research, 1977 Consolazio CF, Johnson RE, Pecora LJ: Physiological Measurements of Metabolic Functions in Man. New York, McGraw-Hill, 1963 Londeree BR: Circulorespiratory exercise prescription tables. In Londeree BR (ed), The Complete Exercise Prescription. Proceedings of the 1978 annual meeting of the Central States Chapter of the American College of Sports Medicine, 1979 Jacobson E: Progressive Relaxation. Chicago, University of Chicago Press, 1938 Bernstein DA, Given GA: Progressive relaxation: Abbreviated methods. In Woolfolk RL, Lehrer PM (eds), Principles and Practice of Stress Management. New York, Guilford, 1984 Lazarus A: In the Mind's Eye: The Power of Imagery for Personal Enrichment. New York, Guilford, 1977 Winer BJ: Statistical Principles in Experimental Design, 2nd ed. New York, McGraw-Hill, 1971 Bumberry W. Oliver JM, McClure JN: Validation of the Beck Depression Inventory in a university population using psychiatric estimate as the criterion. J Consult Clin Psychol 46:150-155, 1978 Blue FR: Aerobic exercise as a treatment for moderate depression. Percept Mot Skill 48:228, 1979 Brown RS, Ramirez D, Taub J: The prescription of exercise for depression. Physician Sprots Med 6:34-45, 1978 Greist J, Klein M, Eischens R, Faris J, Gurman A, Morgan W: Running as treatment for depression. Compr Psychiatry 20:41-54, 1979 Hughes JR: Psychological effects of habitual aerobic exercise: A critical review. Prev Med 13:66-78, 1984 Brown B, Payne T, Kim C, Moore G, Krebs P, Martin W: Chronic response of rat brain norepinephrine and serotonin levels to endurance training. J Appl Physiol 46:19-23, 1979 Carr DB, Bullen BA, Skrinar GS, Arnold MA, Rosenblatt M, Beitins IZ, Martin JB, McArthur JW: Physical conditioning facilitates the exercise-induced secretion of beta-endorphin and beta lipoprotein in women. N Engl J Med 305:560-563, 1981

Psychosomatic Medicine 49:355-365 (1987)

365