Effects of Acute Exercise on Mood and Well-Being in Patients with Major Depressive Disorder

John B. Bartholomew; David Morrison; Joseph T. Ciccolo


Med Sci Sports Exerc. 2005;37(12):2032-2037. 

In This Article


Participants were 15 men and 25 women, 18-55 yr of age, diagnosed with MDD according to the DSM-IV[4] criteria. Future Search Trials, Inc. (FST) referred all participants to this study, which was completed within 2 wk of their diagnosis. Thus, participation in the study coincided with the onset of their treatment. Whereas this would be expected to confound the assessment of exercise training on MDD, it did not confound the effects of a single bout of exercise in this between-subjects design, because no treatment occurred between the pre- and posttest assessments. FST operates both a clinical psychology practice providing treatment for depression and other psychological disorders, as well as a clinical trials research company. All participants were diagnosed with MDD by a board-certified adult psychiatrist employed by FST. The mean age of the participants was 38.1 yr; 68% were white, 17% were Latino, 11% were African American, and the final 4% designated themselves as other. Exclusion criteria for this study were as follows: 1) participants who reported exercising regularly, as defined by more than twice per week, during the 1 month before screening; 2) participants intolerant or resistant to exercise; 3) participants with a principal psychiatric diagnosis other than MDD, or those with a comorbid disorder (e.g., panic disorder); and 4) participants currently taking antidepressant medication. None of the potential participants were excluded using these criteria. This study was approved by the institutional review board of the University of Texas at Austin, and each participant signed an informed consent before beginning the study.

A psychiatrist referred all patients to the study once MDD was diagnosed. No participant had been diagnosed with MDD for longer than 2 wk at the time of participation. Once referred to the study, participants were then interviewed by one of the authors to determine eligibility. All individuals referred met inclusion or exclusion criteria, and each was assigned to either an aerobic exercise session or a quiet-rest control. None were told to refrain from coffee, smoking, or other mood-altering substances (e.g., supplements, energy drinks, chocolate) before testing. Pseudorandom assignment for 40 individuals was established before recruitment that placed 20 participants in each condition. Each participant was then assigned to the condition assigned to the individual's specific number (i.e., 1-40). Affective state was assessed 5 min before exercise or control, and three times after exercise or control: 5, 30, and 60 min after. Both the Profile of Mood States (POMS) and the Subjective Exercise Experiences Scale (SEES) were administered to ask the participants how they are feeling right now, at this moment. Walking speed was adjusted to maintain heart rate at target levels, with heart rate continuously monitored but not recorded by the experimenter. Rating of perceived exertion (RPE) was monitored and recorded every 5 min during exercise. All participants were paid $100 upon completion of the study.

Participants assigned to the aerobic exercise session engaged in one session of brisk walking on a treadmill for 30 min. All participants assigned to the aerobic exercise session wore a heart rate monitor (Vantage XL; Polar Inc.) and walked at an intensity equal to 60-70% of their age-predicted maximal heart rate. This exercise protocol (i.e., intensity, duration, and mode) was selected because it met the American College of Sports Medicine'sACSM) guidelines for accumulated, daily physical activity.[3] No warm-up session was used; however, patients were instructed on how to walk on the treadmill before the exercise bout and were given instruction on walking, if needed, during exercise. The participants were also instructed on how to use the RPE scale, verbatim, as it is listed in ACSM guidelines.[2] Participants assigned to the quiet rest control were given instructions to sit quietly for 30 min. These participants were provided with a comfortable, nonreclining chair, located in the same room as the treadmill. They were told not to sleep and they were not allowed to read. Neither the exercising nor the quiet rest participant experienced any other interactions during this period. At the end of the treadmill walking or quiet rest, participants were directed to a desk where they sat and completed the questionnaire at 5, 30, and 60 min. The same testing room was used for both conditions, and one experimenter was in the room throughout the entire experiment.

Because mood and well-being were the primary dependent variables, no attempt was made to use clinical measures to assess the degree of the ongoing depression. Instead, participants completed both the SEES,[26] and the short form, 30-item POMS.[28] The SEES is a 12-item questionnaire that is scored on a 7-point Likert-type scale anchored with Not at all and Very much so. The SEES provides three subscales that were used in the present study: 1) psychological well-being (great, positive, strong, terrific); 2) psychological distress (awful, crummy, discouraged, miserable); and 3) fatigue (drained, exhausted, fatigued, tired). The SEES has demonstrated adequate reliability, with Cronbach alphas ranging from 0.77 to 0.84 in the present experiment, and it has been used in studies involving exercise.[26] The short form POMS is a 30-item questionnaire that is scored on a 5-point Likert-type scale anchored with Not at all and Extremely. The POMS provides six subscales: 1) anger (e.g., grouchy, furious); 2) confusion (e.g., muddled, forgetful); 3) depression (e.g., sad, unworthy); 4) fatigue (e.g., tired, sluggish); 5) tension (e.g., nervous, anxious); and 6) vigor (e.g., lively, active). The POMS has demonstrated adequate reliability, with Cronbach alphas ranging from 0.93 to 0.95 in the present experiment, and it has been used in exercise studies with similar age and gender distributions.[24] Exercise intensity was measured with Borg's RPE scale, with numerical ratings ranging from 6 to 20 with verbal anchors at 7 (very, very light) and 19 (very, very hard).[9]

Power analysis methodology represents a mixed factorial design, with two levels of the between-subject factor of group and four levels of the within-subjects factor of time. A repeated measures ANOVA power analysis was conducted using Power Analysis and Sample Size software (PASS) 2000.[21] The effect size for this calculation used the ratio of the standard deviation of the effects for a particular factor or interaction and the standard deviation of within-subject effects.[21] In addition, each power calculation assumed a Geisser-Greenhouse corrected F test with a 5% significance level. The power analysis was conducted for a single, two-group between-subjects factor, and a single within-subjects factor assessed over four time points. For this design, 40 participants (20/condition) achieves a power of 0.80 for the between-subjects main effect at an effect size of 0.45; a power of 0.93 for the within-subjects main effect at an effect size of 0.64; and a power of 0.79 for the interaction effect at an effect size of 0.64. The variation in effect size occurs because, in a repeated measures analysis, it is often assumed that the pattern of correlations between two observations will be higher when the observations are closer in time than the correlation for two observations that occur far apart in time.[21] In sum, this design has sufficient power to detect moderate effects.

These data were analyzed via a series of 2 (group) × 4 (time) ANOVA with repeated measures on the second factor. Significant interactions were decomposed into their simple effects of time within each level of group. If all interactions were significant, this would result in 54, experiment-wise post hoc comparisons. As a result, a standard alpha correction (e.g., Bonferoni or Scheffé) would be overly restrictive for the post hoc comparisons. Instead, experiment-wise alpha inflation was protected against by setting the alpha for all post hoc comparisons equal to 0.01. Given the predicted interactions, the groups are expected to report differences in the duration of these effects. As a result, change scores were used as the basis of the effect size estimates,d), which will be presented as the mean difference in these change scores divided by their pooled standard deviation.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.