Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2021 Apr 1.
Published in final edited form as: Health Psychol. 2020 Jan 9;39(4):255–264. doi: 10.1037/hea0000836

The effects of aerobic training on subclinical negative affect: A randomized controlled trial

Kathleen M McIntyre 1,*, Eli Puterman 2, Jennifer M Scodes 3, Tse-Hwei Choo 3, C Jean Choi 3, Martina Pavlicova 4, Richard P Sloan 1
PMCID: PMC7078065  NIHMSID: NIHMS1066270  PMID: 31916828

Abstract

Objective:

The anti-depressant and anxiolytic effects of aerobic exercise are well known, but less is known about its effects on subclinical levels of trait negative affect in healthy but sedentary adults. In the present study, we test the effects of a 3-month randomized controlled trial of aerobic exercise training in young to midlife adults on trait measures of depression, anxiety, hostility, and anger.

Methods:

One-hundred and nineteen men (n = 56) and women (n = 63) aged 20–45 were randomized to one of two conditions: (1) 12 weeks of aerobic exercise after which they were asked to halt exercising and decondition for 4 weeks, or (2) a 16-week waitlist control group. Assessments of depression, anxiety, hostility and anger were completed at study entry, week 12 and week 16.

Results:

At study entry, participants scored low on measures of depression, anxiety, hostility and anger. Analyses among the intent-to-treat and per protocol samples found significant treatment effects of aerobic training for hostility and depression, but not for anxiety and anger. Within-group analyses demonstrated that depression and hostility scores decreased in the exercise group over the course of the intervention, while remaining stable in the control group. These effects persisted for the exercise group at non-significant levels after four weeks of deconditioning.

Conclusions:

Aerobic exercise training has significant psychological effects even in sedentary yet euthymic adults, adding experimental data on the known benefits of exercise in this population.

Keywords: Aerobic exercise, negative affect, mental health, randomized controlled trial, deconditioning

It is now well-established that aerobic exercise exerts a protective effect against disease (Pedersen & Saltin, 2015) and both cardiovascular (Cheng et al., 2018) and all-cause mortality (Arem et al., 2015; Lee et al., 2014). The US Preventive Services Task Force recently concluded “all persons, regardless of their CVD risk status, can gain health benefits from healthy eating behaviors and appropriate physical activity” (p. 167, (Force et al., 2017). In addition to these medically-related benefits, exercise appears to have an impact on indices of wellbeing.

Research in psychiatric populations has established exercise as an effective non-pharmacological intervention, with depression and anxiety disorders receiving the most attention. In particular, a meta-analysis of 58 randomized controlled trials found exercise to be superior to control conditions as an intervention for depressive symptoms in adults (Rethorst, Wipfli, & Landers, 2009). A more recent meta–analysis revealed similar findings on the anti-depressive effect of exercise, even when adjusting for publication bias (Schuch et al., 2016). Mammen and Faulkner’s recent systematic review of 30 prospective studies demonstrated that an active lifestyle also prevents future development of depression in a dose-response relationship in community-based nonclinical samples (Mammen & Faulkner, 2013). Further, recently published results from the largest available genome-wide association study of 611,583 adults showed a significant protective effect of objectively measured physical activity for adults at risk for MDD (K. W. Choi, Chen, C., Stein, M.B., Klimentidis, Y.C., Wang, M., Koenen, K.C., Smoller, J.D. for the Major Depressive Disorder Working Group of the Psychiatric Genomics Consortium, 2019).

Similarly for anxiety disorder symptom reduction, exercise training is superior to no exercise control groups, and in some cases superior to non-exercise treatment interventions (Wipfli, Rethorst, & Landers, 2008). A more recent meta-analysis of exercise training interventions showed them to be more effective in decreasing anxiety than a wait list condition (standardized mean difference (SMD)=−0.41; 95% CI =[−0.70, −0.12], and further, that high intensity exercise was more effective than low intensity exercise (SMD=−0.38; 95% CI=[−0.68, −0.08]; (Aylett, Small, & Bower, 2018).

High trait hostility and trait anger do not constitute formal mental health disorders, although they still impair quality of life through associations with increased interpersonal conflict (Baron et al., 2007; Brondolo et al., 2003) and physical health risk (Barefoot et al., 1991; Pimple et al., 2015). Perhaps because of the strong association between hostility and anger and cardiac risk, most studies of exercise interventions on hostility and anger have been conducted in samples of patients undergoing cardiac rehabilitation (Blumenthal et al., 2005; Lavie & Milani, 1999). Compared to depression and anxiety exercise intervention studies, results from high trait hostility and anger samples are more mixed, but exercise training is generally associated with a decrease in hostility levels (Lavie, Milani, O’Keefe, & Lavie, 2011). To our knowledge, there are no studies of the impact of exercise training on trait anger, though a recent experimental study of an acute bout of exercise was associated with a decrease in angry feelings in 16 male participants scoring high on trait anger (Thom, O’Connor, Clementz, & Dishman, 2019).

For the majority of the population, however, symptoms of depression and anxiety do not reach clinical significance. Subthreshold anxiety and depression have nonetheless been associated with psychosocial impairment (Rivas-Vazquez, Saffa-Biller, Ruiz, Blais, & Rivas-Vazquez, 2004) and in some cases may constitute prodromal evidence of future disorder (Bosman et al., 2019). There is some limited research suggesting that increased exercise frequency may benefit those with subclinical negative affect. For example, a cross-sectional population study in Finland of 3403 adults aged 25 to 64 identified significant negative associations between self-reported aerobic exercise frequency and trait measures of depression, anxiety, hostility, and stress (Hassmen, Koivula, & Uutela, 2000). Additionally, a meta-analysis synthesizing results from 19 exercise training intervention studies in healthy participants reported a small but significant effect size (SMD=0.219) of exercise on reduction of subclinical anxiety symptoms (Conn, 2010) but there is a dearth of similar experimental research on other types of subclinical negative affect.

Given high levels of insufficient physical activity in the global population (e.g. (Kohl et al., 2012), further exploration of the psychological correlates of “movement as medicine” for the average person carries great public health significance. A sedentary lifestyle is associated with poorer mood (Endrighi, Steptoe, & Hamer, 2016), that is, a sedentary lifestyle may confer greater risks for negative affectivity. Negative affectivity is not a psychiatric diagnosis, but rather a trait-like tendency towards experiencing negative emotions including sadness, anxiety, anger, distrust, and low self-regard (Watson & Clark, 1984). The tendency toward negative affectivity has itself been associated with cardiac risk (Bleil, Gianaros, Jennings, Flory, & Manuck, 2008; Suls & Bunde, 2005).

Given these findings, a clear need emerges for randomized intervention studies assessing multiple domains of subclinical negative affect in physically and psychiatrically healthy but sedentary individuals, criteria that characterize the majority of the population. We conducted a randomized trial whose primary aim was to analyze the effects of a 12-week aerobic training intervention on inflammation levels in healthy, sedentary 20 to 45 year old adults. Participants also completed the Beck Depression Inventory (Beck, Ward, Mendelson, Mock, & Erbaugh, 1961), Spielberger Trait Anxiety Inventory (C.D. Spielberger, Gorsuch, Lushene, Vagg, & Jacobs, 1983), Cook Medley Hostility scale (Cook & Medley, 1954), and Spielberger State Trait and Anger Expression Inventory (C.D. Spielberger, 1988) at pre- and post- intervention and after a 4-week deconditioning period in which participants randomized to the training intervention were asked to abstain from any aerobic exercise. Because individuals reporting past or current frank mood or anxiety disorders were excluded from participation, we expected that baseline levels of trait negative affect would be low. We hypothesized that aerobic exercise training would nonetheless reproduce the anti-depressant and anxiolytic effects seen in other studies and also would result in decreased trait hostility and anger. We also hypothesized that after the 4-week deconditioning period, measures of affect would return to study entry levels, as has been shown in a recent review of exercise cessation studies (Weinstein, Koehmstedt, & Kop, 2017).

Methods

Study methods have been described in detail in the primary paper (Sloan et al., 2018) and are depicted in the CONSORT diagram (see Figure 1). The following summarizes the aspects most relevant for the current analyses.

Figure 1.

Figure 1.

Open in a new tab

CONSORT diagram

Study Participants

Participants were recruited between July 2009 and December 2014 with electronic bulletin boards and flyers posted throughout Columbia University Irving Medical Center/New York Presbyterian Hospital. The study was approved by New York State Psychiatric Institute’s Institutional Review Board and registered in ClinicalTrials.Gov: NCT01335737. Stimulated levels of tumor necrosis factor-alpha (TNF-α) were the primary outcome measures of this trial. The target sample size of 182 participants (or 128 completers) was chosen to ensure sufficient power (at least 90%) of a 2-sided test with level of significance of 5% to detect true effect sizes of 0.67 or greater between the aerobic exercise training and the wait-list groups with respect to the primary outcome. Subclinical negative affect levels were measured as secondary outcomes. The total randomized sample consisted of 119 male and female adults (age range 20–45) that were randomized to either the aerobic exercise training (n=60) or waitlist (n=59) group. Data collection concluded by May 2015.

Study Design and Procedures

Interested adults were initially screened by phone with the Baecke Physical Activity assessment (Baecke, Burema, & Frijters, 1982), and excluded if their score was equal to or greater than 10 on the scale, as this value defines regular exercisers. Participants also were excluded if they were current smokers. Eligible participants provided informed consent and were screened for contraindications for exercise, use of hormonal birth control or pregnancy, and a body mass index (BMI) less than 18 or above 33 kg/m2. Participants who remained eligible completed a maximal cardiopulmonary exercise test (CPET). Those who performed at or below average standards set by the American Heart Association (VO2max < 43 ml/kg/min for men, < 37 ml/kg/min for women) were allowed to continue their participation in the study. Following the CPET, participants completed a two-week run-in stretching period to assess their readiness to successfully complete the intervention once randomized. During this period, participants were asked to attend the Fitness Center at Columbia University Medical Center four times/week for 30 minutes of stretching, after receiving instructions from a study appointed ‘coach’ who monitored adherence by examining gym attendance records and heart rate monitor data (Polar model s610i).

Participants who completed at least seven stretching sessions during the two-week run-in period remained eligible and were scheduled for their time 1 (T1) fasting blood draw, after which they ate a light breakfast. They also completed a series of mood questionnaires at this visit, then were randomized to either a 12-week aerobic exercise-training program or a waitlist control condition with an allocation ratio of 1:1 using random block assignment stratified by sex.). The randomization schedules were generated by an independent study statistician and concealed until an eligible participant was ready for enrollment. Due to the nature of the conditions, they were not masked to participants or research assistant staff, although participants were not aware of the overall study hypotheses. All study staff who collected outcomes data or analyzed study data were unaware of study group assignments. At the end of the 12 weeks (time 2; T2), all participants returned for a blood draw, repeat mood assessment and CPET, administered by research staff blind to randomization condition. This testing session was scheduled 24 and 72 hours after their last exercise session, and negative affect questionnaires were administered prior to CPET to avoid confounding mood effects of acute exercise. Those in the exercise arm were then asked to decondition (i.e., reduce their activity levels to their previously sedentary levels) for four weeks after which and all participants returned for a final blood draw, completion of questionnaires, and CPET, (time 3, T3). Participants were compensated with $210 for the total of testing sessions, and those in the waitlist control received a free five-month gym membership. To boost adherence, participants in the exercise arm were offered an additional free two months of their free gym memberships if they achieved at least 85% adherence during the 12-week program.

Aerobic Training Arm

12-Week Training Period.

Participants attended the Fitness Center for four workouts/week on their own time. Study appointed coaches provided an individualized workout program specifying their training goals based on their maximum heart rate from their CPET. Participants gradually increased from 55–65% of their maximum heart rate (HR) in weeks 1 and 2, to 65–75% in weeks 3 and 4, to eventually train at 80% in weeks 5–12. To confirm that they exercised at target HRs, participants wore a Polar Electro model s610i heart rate monitor chest strap and watch during each training session. HR was recorded throughout the training session and uploaded by the participant to a computer stationed in the Fitness Center and monitored by the coaches. Coaches verified adherence with these data, gym attendance records, and weekly logs provided by the participants. Coaches contacted participants on a weekly basis to maintain engagement and, if participants slowed their participation at the Fitness Center, coaches called more often.

Sedentary Deconditioning.

After the 12-week training period, participants were instructed to refrain from any type of exercise for four weeks. Coaches contacted them weekly during these 4 weeks to encourage adherence to the deconditioning phase. Deconditioning was confirmed by CPET.

Waitlist Control Arm

Participants randomized to the waitlist control group were asked to remain physically inactive for the full 16 weeks of the study. Coaches contacted wait list participants weekly to verbally verify abstention from any type of exercise and sedentary status at T2 and T3 was confirmed by CPET.

Measures

Demographic and Anthropometric Measures

Age (in years), sex (male, female), race (Asian, Black/African American, White, Native Hawaiian/Pacific Islander, Other), and ethnicity (Hispanic/Not Hispanic) were recorded. Anthropometrics (weight and height) were measured three times (at blood draw, during a psychophysiology session, and CPET) at each testing session and averaged, and BMI (kg/m2) calculated.

Psychological Measures

Depression was measured using the Beck Depression Inventory Version 1 Total Score (BDI) (Beck et al., 1961), a 21-item self-report scale with a total score ranging from 0 to 63. Research using the BDI has provided support for clinically-meaningful continuity between subclinical and clinical depression constructs (Enns, Cox, & Borger, 2001; Judd, Rapaport, Paulus, & Brown, 1994). Anxiety was measured by self-report using the trait anxiety subscale of the Spielberger State-Trait Anxiety Inventory (C.D. Spielberger et al., 1983) (STAI), a 20-item scale with a range from 20 to 80, commonly used to assess sub-clinical anxiety changes in exercise studies (e.g. (Conn, 2010). Hostility was measured using a subset of the 50-item Cook-Medley Hostility Scale Total Score (CKM;(Cook & Medley, 1954). The subset included the total score of three (cynicism, hostile affect, and aggressive responding) of the seven subscale components that have been shown to best relate to health outcomes (Barefoot, Dodge, Peterson, Dahlstrom, & Williams, 1989). CKM total score subset included 27 true/false items for a total range of 0 to 27. The Spielberger State Trait and Anger Expression inventory (STAXI) (C.D. Spielberger, 1988) was also administered, with the present analyses using the trait anger, a 10item scale with a total score ranging from 10 to 40.

Statistical Approach

To examine the effect of random treatment assignment on BDI (depression), STAI (anxiety), CKM (hostility), and STAXI (anger) outcomes, longitudinal linear mixed effect models (MEMs) with generalized estimating equations (Schafer, 1997) were fit separately for each outcome. The MEMs adjusted for unequal variances between treatment groups and included a random intercept to account for the between subject variation. A log-link function was used to account for the right-skewed distribution of the BDI and STAI outcomes, and therefore estimated effects were computed as ratios and interpreted as percentage change. An identity link function was used for the approximately normally distributed CKM and STAXI outcomes, and therefore estimated effects were computed as differences and interpreted in actual unit change. For all estimated effects, the corresponding 95% confidence intervals were also computed.

Each model included the effects of treatment group (exercise vs. waitlist), time (T2 and T3), and their two-way interaction. Pre-specified contrasts were estimated for the treatment effect at T2 and T3, as well as the time effect from Time 1 (T1) to T2 within each treatment group. To estimate the within-group changes from T1 to T2 for each outcome while adjusting for T1, the observed values at all sessions were centered by subtracting the grand mean of the corresponding T1 values for all subjects. Subtracting a constant from the outcome and T1 values does not affect the relationship but provides a way to estimate the changes from T1 to T2 within group and to assess their significance (Sloan et al., 2018). In order to account for study participants’ T1 characteristics that can be influential on the outcome measures, all models adjusted for participants’ age, sex, and the corresponding outcome measure at T1.

All analyses were conducted on the intent-to-treat (ITT) sample, where all randomized subjects contribute to the grand mean centering to estimate within-group changes, and on the per-protocol sample. The per-protocol sample was defined as completing at least 50% of scheduled exercise training sessions, having a blood draw within 18 hours to 10 days after the last exercise training session, and completing the CPET within 14 days of the last exercise training session. There were 94 participants in the per-protocol sample. PROC GLIMMIX® in SAS® 9.4 was used for all analyses, and all statistical tests were 2-sided with significance level of 5%.

Two sensitivity analyses were also performed to evaluate the effect of missingness. First, logistic regression models on missing values were run to see if baseline affect values were associated with missingness and dropout. Additionally, missingness was assessed by first imputing missing values using multiple imputation via the Markov Chain Monte Carlo (MCMC) method (Schafer, 1997) with a single chain assuming multivariate normality (SAS PROC MI), and then reanalyzing the outcomes with the values included in analyses.

Results

Demographics

A total of 119 participants (56 men, 63 women) were randomized to the treatment conditions. Table 1 presents participants’ T1 demographic characteristics; exercise and waitlist groups were well-balanced. In both, participants were on average 31 years old with a normal BMI (24.9 kg/m2).

Table 1.

Demographic characteristics at Study Entry (N=119)

Mean (SD) or n (%)
Aerobic Exercise (n=60) Waitlist (n=59)
Age (years) 31.2 (5.7) 31.4 (6.2)
Sex
 Male 28 (46.7%) 28 (47.5%)
 Female 32 (53.3%) 31 (52.5%)
BMI (kg/m2)a 24.9 (3.8) 24.9 (3.8)
Race
 Asian 15 (25.0%) 18 (30.5%)
 Native Hawaiian or Pacific Islander 1 (1.7%) 0 (0.0%)
 Black or African American 10 (16.7%) 11 (18.6%)
 White 20 (33.3%) 20 (33.9%)
 Other 14 (23.3%) 10 (16.9%)
Ethnicity
 Hispanic 19 (31.7%) 12 (20.3%)
 Not Hispanic 41 (68.3%) 47 (79.7%)
Open in a new tab

SD = standard deviation

a

One subject missing BMI from the aerobic exercise group.

Aerobic Training Adherence

In the aerobic exercise group, a mean of 33.5 (SD=17.4) of the 48 training sessions (70%) were completed. Among only the 45 participants who completed T2 testing, a mean of 39.8 (SD=13.6) of the scheduled 48 training sessions (83%) were completed. Among the 40 who completed training and all 3 testing sessions, the attendance rate was 90%. There were no protocol-related adverse events. Among those in the aerobic exercise group in the per-protocol sample (n=35), a mean of 44.2 exercise training sessions were completed.

Effects on Affective Measures

Table 2 presents descriptive statistics of the observed affective measure scores at each time point. The model-estimated geometric means and geometric standard errors for the three time-points for BDI and STAI are displayed in Figures 2 and 3, respectively. The model-estimated means and standard errors for the three time-points for CKM and STAXI are displayed in Figures 4 and 5, respectively.

Table 2:

Summaries of Affective Measuresa

Aerobic Exercise Waitlist
Affective Measure Geometric Mean (SD)
Depression (BDI)
T1 2.66 (3.16) 1.92 (3.36)
T2 1.78 (3.33) 2.12 (3.33)
T3 1.73 (3.86) 1.71 (3.29)
Anxiety (STAI)
T1 33.79 (1.28) 33.35 (1.33)
T2 32.82 (1.33) 32.22 (1.32)
T3 30.89 (1.35) 32.49 (1.32)
Mean (SD)
Hostility (CKM)
T1 10.09 (5.06) 8.41 (4.27)
T2 8.62 (4.83) 8.81 (4.79)
T3 9.41 (5.68) 8.75 (5.42)
Anger (STAXI)
T1 15.59 (3.87) 14.41 (2.80)
T2 14.93 (3.75) 14.60 (3.21)
T3 15.10 (4.24) 14.32 (3.24)
Open in a new tab
a

Observed geometric means and standard deviations (SD) are presented for non-normally distributed outcomes (anxiety and depression). Means and standard deviations (SD) are presented for normally distributed outcomes (hostility and anger).

Figure 2.

Figure 2.

Open in a new tab

Observed T1 geometric mean plotted with model estimated T2 and T3 geometric means and standard errors of BDI (depression). Dashed lines show the within-group change between the observed T1 mean and adjusted model-estimated means at T2. Solid lines show the change between model-estimated means a T2 and T3

Figure 3.

Figure 3.

Open in a new tab

Observed T1 geometric mean plotted with model estimated T2 and T3 geometric means and standard errors of STAI (trait anxiety). Dashed lines show the within-group change between the observed T1 mean and adjusted model-estimated means at T2. Solid lines show the change between model-estimated means a T2 and T3

Figure 4.

Figure 4.

Open in a new tab

Observed T1 mean plotted with model estimated T2 and T3 means and standard errors of CKM (hostility). Dashed lines show the within-group change between the observed T1 mean and adjusted model-estimated means at T2. Solid lines show the change between model-estimated means a T2 and T3

Figure 5.

Figure 5.

Open in a new tab

Observed T1 mean plotted with model estimated T2 and T3 means and standard errors of STAXI (trait anger). Dashed lines show the within-group change between the observed T1 mean and adjusted model-estimated means at T2. Solid lines show the change between model-estimated means a T2 and T3

Depression (BDI).

In the ITT sample, the treatment effect at T2 was significant – the BDI scores were 39% lower in the exercise training group compared to the waitlist group (b=0.61, 95% CI=[0.42, 0.88], model estimated Cohen’s d=0.42; Figure 2). After the deconditioning phase, the treatment effect at T3 was no longer significant: the BDI scores were 27% lower in the exercise group compared to the waitlist group (b=0.73, 95% CI=[0.49, 1.07]). The change from T1 to T2 was significant in the exercise group, as the BDI scores showed a 35% decrease (b=0.65, 95% CI=[0.48, 0.87]), but not in the waitlist group (6% increase; b=1.06, 95% CI=[0.85, 1.33]). Analyses on the per-protocol sample showed similar results in terms of significance and confidence intervals.

Anxiety (STAI).

In the ITT sample, the treatment effect at T2 was not significant – the STAI scores were only 1% lower in the exercise group compared to the waitlist group (b=0.99, 95% CI=[0.93, 1.06], model estimated Cohen’s d=0.02; Figure 3). After the deconditioning phase, the treatment effect at T3 was also not significant: STAI scores were 6% lower in the exercise group compared to the waitlist group (b=0.94, 95% CI=[0.88, 1.01]). The change in STAI scores from T1 to T2 was not significant within either group: both groups decreased by 4% at T2 compared to at T1 (Exercise group: b=0.96, 95% CI=[0.92, 1.00]; Waitlist Group: b=0.96, 95% CI=[0.92, 1.01]). Analyses on the per-protocol sample showed similar results in terms of significance and confidence intervals.

Hostility (CKM).

In the ITT sample, the treatment effect at T2 was significant – the CKM scores were on average 1.69 points lower in the exercise group compared to the waitlist group (95% CI=[−3.04, −0.33], model estimated Cohen’s d=0.36; Figure 4). However, after the deconditioning phase, the treatment effect was no longer significant at T3 (1.10 points lower in the exercise group compared to the waitlist group; 95% CI=[−2.51, 0.31]). The estimated treatment effect sizes at T2 and T3 were similar, but had different levels of significance potentially due to the smaller sample size at T3. The change from T1 to T2 was significant in the exercise group, as the CKM scores decreased on average by 1.49 points (95% CI=[−2.52, −0.45]), but not in the waitlist group (b=0.20; 95% CI=[−0.66, 1.07]). Analyses on the per-protocol sample showed similar results in terms of significance and confidence intervals.

Anger (STAXI).

In the ITT sample, the treatment effect at T2 was not significant (Figure 5). The STAXI trait anger scores were only 0.17 units lower in the exercise group compared to the waitlist group (b=−0.17, 95% CI=[−1.13, 0.79], model estimated Cohen’s d=0.05; Figure 5). After the deconditioning phase, the treatment effect at T3 was also not significant, STAXI scores were 0.21 units higher in the exercise group compared to the waitlist group (b=0.21, 95% CI=[0.79, 1.21]). The change in STAXI scores from T1 to T2 was not significant within either group, the exercise group decreased by 0.30 units in T2 compared to T1 (b=−0.30, 95% CI=[−1.05, 0.44], and the waitlist group decreased by 0.13 units in T2 compared to T1 (b=−0.13, 95% CI=[0.73, 0.47]). Analyses on the per-protocol sample showed similar results in terms of significance and confidence intervals.

Sensitivity Analyses for Missing Data

For all four affective outcomes, 2 (1.7%) participants were missing T1 values and 17 (14.3%) participants were missing T2 values. For BDI, CKM, and STAXI outcome measures, 27 (22.7%) participants were missing T3 values. For the STAI outcome measure, 28 (23.5%) were missing T3 values. Missing values were largely due to study dropout, as 16 (13.4%) participants left the study between T1 and T2, and an additional 5 (4.2%; 17.6% overall) participants left the study between T2 and T3. Logistic regression of missingness for the BDI, STAI, CKM, STAXI and of dropout found no significant associations between baseline affect measures and missing values nor dropout. Additionally, to evaluate the effect of missing observations through imputation, a second sensitivity analysis was performed by imputing missing values using multiple MCMC imputations and then reanalyzing each outcome. Results based on the MCMC imputed data were consistent with the ITT analyses.

Discussion

In a sample of healthy but sedentary adults randomized to 12 weeks of aerobic exercise training, subsyndromal levels of depression and trait hostility significantly decreased from T1 to T2 for the treatment but not the wait list group. These changes were no longer significant at T3 after a four-week period of deconditioning but contrary to prediction, mean levels of each outcome at T3 remained lower than T1 rather than making a full return to baseline. Exercise training had no effect on subclinical trait anxiety or anger, contrary to the hypothesis that each of these four traits would decrease post-training and, in the case of anxiety, running contrary to findings from many other studies.

These data add experimental support to existing cross-sectional evidence that low levels of depression and hostility are inversely associated with physical activity even in non-clinical populations. Notable findings from the current study discussed below include the post-deconditioning reversal of the treatment-induced decrease in depression and hostility, the failure to replicate a significant training-induced anxiety reduction as typically reported in other studies, and the discrepancy in the effects of exercise training between the usually highly-correlated trait outcomes of anger and hostility.

One strength of this study in addition to its randomized controlled design is its four-week deconditioning period. Contrary to prediction, the training-induced reduction in BDI and CKM levels in the treatment but not the wait list group persisted at 4 weeks post-training, albeit at non-significant levels. It is impossible to say definitively why, in the present sample, the effect of exercise training on depression and hostility at T2 was still present but no longer significant at T3. However, 17.6% of all participants dropped out prior to T3 testing, which may inhibit the capacity to detect significant post-training effects (Amrhein, Greenland, & McShane, 2019), especially for participants with such low T1 trait affect levels. For those who did complete T3, there is greater variance in trait negative affect scores and smaller effect sizes. Greater heterogeneity in scores at T3 for those who remained in the study also suggests that there may be individual differences in participants’ ability to sustain mood and hostility benefits after cessation of exercise. Few studies examine the psychological effects of deconditioning in non-patient samples, but there is some research to suggest that increases in negative affect are associated with imposed exercise cessation after just one or two weeks. For example in one study of regular exercisers randomized to a withdrawal or continuation condition, BDI scores increased significantly for those who stopped exercising after just two weeks (Berlin, Kop, & Deuster, 2006). Similarly, an increase in anxiety was observed after one week of experimentally-induced sedentariness in a sample of active young adults relative to those who were randomized to continue with exercise as usual (Edwards & Loprinzi, 2016).

For subclinical depression, scores significantly decreased by 39% for the intervention group at the conclusion of 12 weeks of training and remained 27% lower than T1 after deconditioning, though this decrease was no longer was significant. There is a lack of consensus on what constitutes a minimal clinically important difference in BDI score changes among patients for whom depression is not a primary presenting concern (Masson & Tejani, 2013). However, given the heightened risk for depression associated with chronic sedentary behavior (Zhai, Zhang, & Zhang, 2015), even minimal differences in depression scores resulting from exercise could have an impact on public health as a preventive intervention to reduce depression’s high disability burden (K. W. Choi et al., 2019). In this study of non-depressed but sedentary individuals, effect sizes approach the threshold for medium, which is unsurprising given the widely-known mood benefits of exercise. In addition to benefitting physical health, exercise is also a form of behavioral activation, a highly effective treatment for depression (Veale, 2008). Though not clinically depressed at T1, exercise group participants nonetheless shifted from a sedentary lifestyle to 12 weeks of consistently scheduled, goal-directed activity in pursuit of clinically meaningful improvement in aerobic capacity. It is conceivable that after a month-long prohibition of this newly-adopted behavior, overall mood benefits would decrease, as both the acute psychological benefits of exercise and secondary emotional benefits of taking long term action to improve one’s health could have dissipated.

The biological mechanisms responsible for this training-induced reduction in subclinical levels of depression and hostility are unclear but one candidate mechanism is inflammation. Many studies report that exercise training reduces circulating levels of inflammatory cytokines (Hamer et al., 2012) and the prominent role of inflammation in depression has been well established (Kohler et al., 2017). However, we found no effect of exercise training on LPS-stimulated levels of TNFα or IL-6 (Sloan et al., 2018) or on circulating inflammatory markers (in submission).

The failure to find an effect of exercise training on anxiety is contrary to the results of two meta-analyses that have found a small but significant post-training effect of exercise on subclinical anxiety (Conn, 2010; Rebar et al., 2015). The moderate intensity level of the current exercise protocol may be a consideration here, as Conn (2010) also notes that higher intensity exercise has a greater anxiolytic effect compared to lower intensity training. Additionally, it is possible that, given the low T1 levels of anxiety in this sample (75% of which were below a score of 40 on a scale ranging 20 to 80; (C.D. Spielberger et al., 1983), post-intervention anxiety reductions may be limited by a floor effect.

Finally, as constructs, trait hostility and trait anger share a similar factor structure of affective, behavioral, and cynical cognition components (Martin, Watson, & Wan, 2000; Wilkowski & Robinson, 2010), so it is curious that in this sample, exercise training was associated with reductions only in hostility. While some support exists for the hostility reduction effect of exercise training (Lavie et al., 2011), research on anger and physical activity has tended to focus only on state anger changes as a result of bouts of acute exercise (e.g. (Thom et al., 2019). Given the high risk profile of elevated levels of trait anger (Williams, Nieto, Sanford, & Tyroler, 2001), the lack of research on aerobic training as a psychological health behavior intervention for both clinical and sub-clinical anger is surprising, but the results of the current study do not encourage further exploration.

This study has several limitations. The age range of 25–40 prevents generalization of findings to older cohorts. Further, the control condition was a passive wait list that lacked the social component of regular interaction with a supportive research assistant “coach” that could have contributed to the reduction in negative affect seen in the intervention group. Additionally, the post-intervention reduction in anxiety and other negative affect outcomes may have been limited by a floor effect because of the exclusion of participants who reported past or current frank mood or anxiety disorders. Future experimental studies of the impact of exercise on subclinical affect should allow participation of subjects with higher levels of anxiety and negative affect and should incorporate assessment of participants’ attitudes about exercise and psychological experience of engaging in the intervention. Future studies should also assess biological variables that could shed more light on the mechanisms associated with psychological benefits of exercise training in this less-studied but highly prevalent population.

Clinical Implications

The present study offers some of the first experimental evidence that a shift from sedentary behavior to aerobic conditioning over 12 weeks may be associated with moderate benefits in two measures of negative affect in a subclinical sample. In addition to building upon existing cross-sectional inverse associations between negative affect and exercise for those not currently diagnosed with depression or high hostility, the current findings add needed specificity to the type, intensity, and duration of training regimen that is likely to be effective as a public health recommendation in similar samples (Chekroud et al., 2018).

Conclusions

Psychological functioning is an important health endpoint in its own right and has been linked to the mitigation of several diseases and early mortality. The current study showed that a 3-month aerobic exercise training program lowered depression and hostility but not trait anxiety or trait anger in euthymic healthy but sedentary young adults. These results demonstrate that exercise training is psychologically beneficial even among those with low levels of negative affect and lend further support to the now well-established view that exercise has health benefits even among low risk individuals.

Acknowledgments

This research was supported in part to EP by the Canada Research Chairs program, Grants R00 HL 109247 (E.P.), R01 HL094423 (R.P.S.) from the National Heart, Lung, and Blood Institute, Grant UL1 TR001873 from the NIH Center for Advancing Translational Sciences, and the Nathaniel Wharton Fund.

Footnotes

TRIAL REGISTRATION: ClinicalTrials.gov identifier: NCT01335737

Conflict of Interest

There are no conflicts of interest to report. The results of the present study do not constitute endorsement by ACSM. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the funding agencies. Results are presented clearly, honestly, and without fabrication, falsification, or inappropriate data manipulation.

References:

  1. Amrhein V, Greenland S, & McShane B (2019). Retire statistical significance. Nature, 567(7748), 305–307. doi:DOI 10.1038/d41586-019-00857-9 [DOI] [PubMed] [Google Scholar]
  2. Arem H, Moore SC, Patel A, Hartge P, Berrington de Gonzalez A, Visvanathan K, … Matthews CE (2015). Leisure time physical activity and mortality: a detailed pooled analysis of the dose-response relationship. JAMA Intern Med, 175(6), 959–967. doi: 10.1001/jamainternmed.2015.0533 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Aylett E, Small N, & Bower P (2018). Exercise in the treatment of clinical anxiety in general practice - a systematic review and meta-analysis. BMC Health Serv Res, 18(1), 559. doi: 10.1186/s12913-018-3313-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Baecke JA, Burema J, & Frijters JE (1982). A short questionnaire for the measurement of habitual physical activity in epidemiological studies. Am J Clin Nutr, 36(5), 936–942. doi: 10.1093/ajcn/36.5.936 [DOI] [PubMed] [Google Scholar]
  5. Barefoot JC, Dodge KA, Peterson BL, Dahlstrom WG, & Williams RB Jr. (1989). The Cook-Medley hostility scale: item content and ability to predict survival. Psychosomatic medicine, 51(1), 46–57. [DOI] [PubMed] [Google Scholar]
  6. Barefoot JC, Peterson BL, Dahlstrom WG, Siegler IC, Anderson NB, & Williams RB Jr. (1991). Hostility patterns and health implications: Correlates of Cook-Medley Hostility Scale scores in a national survey. Health psychology, 10(1), 18–24. [DOI] [PubMed] [Google Scholar]
  7. Baron KG, Smith TW, Butner J, Nealey-Moore J, Hawkins MW, & Uchino BN (2007). Hostility, anger, and marital adjustment: Concurrent and prospective associations with psychosocial vulnerability. Journal of behavioral medicine, 30(1), 1–10. doi: 10.1007/s10865-006-9086-z [DOI] [PubMed] [Google Scholar]
  8. Beck AT, Ward CH, Mendelson M, Mock J, & Erbaugh J (1961). An inventory for measuring depression. Arch Gen Psychiatry, 4, 561–571. [DOI] [PubMed] [Google Scholar]
  9. Berlin AA, Kop WJ, & Deuster PA (2006). Depressive mood symptoms and fatigue after exercise withdrawal: the potential role of decreased fitness. Psychosomatic medicine, 68(2), 224–230. doi: 10.1097/01.psy.0000204628.73273.23 [DOI] [PubMed] [Google Scholar]
  10. Bleil ME, Gianaros PJ, Jennings JR, Flory JD, & Manuck SB (2008). Trait negative affect: toward an integrated model of understanding psychological risk for impairment in cardiac autonomic function. Psychosomatic medicine, 70(3), 328–337. doi: 10.1097/PSY.0b013e31816baefa [DOI] [PubMed] [Google Scholar]
  11. Blumenthal JA, Sherwood A, Babyak MA, Watkins LL, Waugh R, Georgiades A, … Hinderliter A (2005). Effects of exercise and stress management training on markers of cardiovascular risk in patients with ischemic heart disease: a randomized controlled trial. JAMA, 293(13), 1626–1634. doi: 10.1001/jama.293.13.1626 [DOI] [PubMed] [Google Scholar]
  12. Bosman RC, Ten Have M, de Graaf R, Muntingh AD, van Balkom AJ, & Batelaan NM (2019). Prevalence and course of subthreshold anxiety disorder in the general population: A three-year follow-up study. Journal of affective disorders, 247, 105–113. doi: 10.1016/j.jad.2019.01.018 [DOI] [PubMed] [Google Scholar]
  13. Brondolo E, Rieppi R, Erickson SA, Bagiella E, Shapiro PA, McKinley P, & Sloan RP (2003). Hostility, interpersonal interactions, and ambulatory blood pressure. Psychosomatic medicine, 65(6), 1003–1011. doi: 10.1097/01.PSY.0000097329.53585.A1 [DOI] [PubMed] [Google Scholar]
  14. Chekroud SR, Gueorguieva R, Zheutlin AB, Paulus M, Krumholz HM, Krystal JH, & Chekroud AM (2018). Association between physical exercise and mental health in 1.2 million individuals in the USA between 2011 and 2015: a cross-sectional study. Lancet Psychiatry, 5(9), 739–746. doi: 10.1016/S2215-0366(18)30227-X [DOI] [PubMed] [Google Scholar]
  15. Cheng W, Zhang Z, Cheng W, Yang C, Diao L, & Liu W (2018). Associations of leisure-time physical activity with cardiovascular mortality: A systematic review and meta-analysis of 44 prospective cohort studies. Eur J Prev Cardiol, 25(17), 1864–1872. doi: 10.1177/2047487318795194 [DOI] [PubMed] [Google Scholar]
  16. Choi KW, Chen CY, Stein MB, Klimentidis YC, Wang MJ, Koenen KC, … Working MDD (2019). Assessment of Bidirectional Relationships Between Physical Activity and Depression Among Adults A 2-Sample Mendelian Randomization Study. JAMA Psychiatry, 76(4), 399–408. doi: 10.1001/jamapsychiatry.2018.4175 [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Choi KW, Chen C, Stein MB, Klimentidis YC, Wang M, Koenen KC, Smoller JD for the Major Depressive Disorder Working Group of the Psychiatric Genomics Consortium. (2019). Assessment of bidirectional relationships between physical activity and depression among adults: A 2-sample Mendelian randomization study. JAMA Psychiatry, E1–E10. doi: 10.1001/jamapsychiatry.2018.4175 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Conn VS (2010). Anxiety outcomes after physical activity interventions: meta-analysis findings. Nursing research, 59(3), 224–231. doi: 10.1097/NNR.0b013e3181dbb2f8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Cook WW, & Medley DM (1954). Proposed hostility and pharisiac-virtue scales for the MMPI. Journal of applied psychology, 38, 414–418. doi: 10.1037/h0060667 [DOI] [Google Scholar]
  20. Edwards MK, & Loprinzi PD (2016). Experimentally increasing sedentary behavior results in increased anxiety in an active young adult population. Journal of affective disorders, 204, 166–173. doi: 10.1016/j.jad.2016.06.045 [DOI] [PubMed] [Google Scholar]
  21. Endrighi R, Steptoe A, & Hamer M (2016). The effect of experimentally induced sedentariness on mood and psychobiological responses to mental stress. Br J Psychiatry, 208(3), 245–251. doi: 10.1192/bjp.bp.114.150755 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Enns MW, Cox BJ, & Borger SC (2001). Correlates of analogue and clinical depression: a further test of the phenomenological continuity hypothesis. Journal of affective disorders, 66(2–3), 175–183. [DOI] [PubMed] [Google Scholar]
  23. Force USPST, Grossman DC, Bibbins-Domingo K, Curry SJ, Barry MJ, Davidson KW, … Tseng CW (2017). Behavioral Counseling to Promote a Healthful Diet and Physical Activity for Cardiovascular Disease Prevention in Adults Without Cardiovascular Risk Factors: US Preventive Services Task Force Recommendation Statement. JAMA, 318(2), 167–174. doi: 10.1001/jama.2017.7171 [DOI] [PubMed] [Google Scholar]
  24. Hamer M, Sabia S, Batty GD, Shipley MJ, Tabak AG, Singh-Manoux A, & Kivimaki M (2012). Physical activity and inflammatory markers over 10 years: follow-up in men and women from the Whitehall II cohort study. Circulation, 126(8), 928–933. doi: 10.1161/CIRCULATIONAHA.112.103879 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Hassmen P, Koivula N, & Uutela A (2000). Physical exercise and psychological well-being: a population study in Finland. Prev Med, 30(1), 17–25. doi: 10.1006/pmed.1999.0597 [DOI] [PubMed] [Google Scholar]
  26. Judd LL, Rapaport MH, Paulus MP, & Brown JL (1994). Subsyndromal symptomatic depression: a new mood disorder? J Clin Psychiatry, 55 Suppl, 18–28. [PubMed] [Google Scholar]
  27. Kohl HW 3rd, Craig CL, Lambert EV, Inoue S, Alkandari JR, Leetongin G, … Lancet Physical Activity Series Working, G. (2012). The pandemic of physical inactivity: global action for public health. Lancet, 380(9838), 294–305. doi: 10.1016/S0140-6736(12)60898-8 [DOI] [PubMed] [Google Scholar]
  28. Kohler CA, Freitas TH, Maes M, de Andrade NQ, Liu CS, Fernandes BS, … Carvalho AF (2017). Peripheral cytokine and chemokine alterations in depression: a meta-analysis of 82 studies. Acta Psychiatr Scand, 135(5), 373–387. doi: 10.1111/acps.12698 [DOI] [PubMed] [Google Scholar]
  29. Lavie CJ, & Milani RV (1999). Effects of cardiac rehabilitation and exercise training programs on coronary patients with high levels of hostility. Mayo Clin Proc, 74(10), 959–966. doi: 10.4065/74.10.959 [DOI] [PubMed] [Google Scholar]
  30. Lavie CJ, Milani RV, O’Keefe JH, & Lavie TJ (2011). Impact of exercise training on psychological risk factors. Progress in cardiovascular diseases, 53(6), 464–470. doi: 10.1016/j.pcad.2011.03.007 [DOI] [PubMed] [Google Scholar]
  31. Lee DC, Pate RR, Lavie CJ, Sui X, Church TS, & Blair SN (2014). Leisure-time running reduces all-cause and cardiovascular mortality risk. Journal of the American College of Cardiology, 64(5), 472–481. doi: 10.1016/j.jacc.2014.04.058 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Mammen G, & Faulkner G (2013). Physical activity and the prevention of depression: a systematic review of prospective studies. American journal of preventive medicine, 45(5), 649–657. doi: 10.1016/j.amepre.2013.08.001 [DOI] [PubMed] [Google Scholar]
  33. Martin R, Watson D, & Wan CK (2000). A three-factor model of trait anger: Dimensions of affect, behavior, and cognition. Journal of personality, 68(5), 869–897. [DOI] [PubMed] [Google Scholar]
  34. Masson SC, & Tejani AM (2013). Minimum clinically important differences identified for commonly used depression rating scales. Journal of Clinical Epidemiology, 66(7), 805–807. doi: 10.1016/j.jclinepi.2013.01.010 [DOI] [PubMed] [Google Scholar]
  35. Pedersen BK, & Saltin B (2015). Exercise as medicine - evidence for prescribing exercise as therapy in 26 different chronic diseases. Scandinavian Journal of Medicine & Science in Sports, 25, 1–72. doi: 10.1111/sms.12581 [DOI] [PubMed] [Google Scholar]
  36. Pimple P, Shah A, Rooks C, Bremner JD, Nye J, Ibeanu I, … Vaccarino V (2015). Association between anger and mental stress-induced myocardial ischemia. American heart journal, 169(1), 115–+. doi: 10.1016/j.ahj.2014.07.031 [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Rebar AL, Stanton R, Geard D, Short C, Duncan MJ, & Vandelanotte C (2015). A meta-meta-analysis of the effect of physical activity on depression and anxiety in non-clinical adult populations. Health Psychol Rev, 9(3), 366–378. doi: 10.1080/17437199.2015.1022901 [DOI] [PubMed] [Google Scholar]
  38. Rethorst CD, Wipfli BM, & Landers DM (2009). The antidepressive effects of exercise: a meta-analysis of randomized trials. Sports Med, 39(6), 491–511. doi: 10.2165/00007256200939060-00004 [DOI] [PubMed] [Google Scholar]
  39. Rivas-Vazquez RA, Saffa-Biller D, Ruiz I, Blais MA, & Rivas-Vazquez A (2004). Current issues in anxiety and depression: Comorbid, mixed, and subthreshold disorders. Professional Psychology-Research and Practice, 35(1), 74–83. doi: 10.1037/07357028.35.1.74 [DOI] [Google Scholar]
  40. Schafer JL (1997). Analysis of incomplete multivariate data. In Monographs on statistics and applied probability 72 Retrieved from http://www.columbia.edu/cgibin/cul/resolve?clio8681056 [Google Scholar]
  41. Schuch FB, Vancampfort D, Richards J, Rosenbaum S, Ward PB, & Stubbs B (2016). Exercise as a treatment for depression: A meta-analysis adjusting for publication bias. Journal of Psychiatric Research, 77, 42–51. doi: 10.1016/j.jpsychires.2016.02.023 [DOI] [PubMed] [Google Scholar]
  42. Sloan RP, Shapiro PA, McKinley PS, Bartels M, Shimbo D, Lauriola V, … Tracey KJ (2018). Aerobic Exercise Training and Inducible Inflammation: Results of a Randomized Controlled Trial in Healthy, Young Adults. J Am Heart Assoc, 7(17), e010201. doi: 10.1161/JAHA.118.010201 [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Spielberger CD (1988). State-trait anger expression inventory: Professional manual. Odessa, FL: Psychological Assessment Resources. [Google Scholar]
  44. Spielberger CD, Gorsuch RL, Lushene R, Vagg PR, & Jacobs GA (1983). Manual for the state-trait anxiety inventory. Palo Alto, CA: Consulting Psychologists Press. [Google Scholar]
  45. Suls J, & Bunde J (2005). Anger, anxiety, and depression as risk factors for cardiovascular disease: the problems and implications of overlapping affective dispositions. Psychological bulletin, 131(2), 260–300. doi: 10.1037/0033-2909.131.2.260 [DOI] [PubMed] [Google Scholar]
  46. Thom NJ, O’Connor PJ, Clementz B, & Dishman RK (2019). Acute Exercise Prevents Angry Mood Induction but Does Not Change Angry Emotions. Med Sci Sports Exerc doi: 10.1249/MSS.0000000000001922 [DOI] [PubMed] [Google Scholar]
  47. Veale D (2008). Behavioural activation for depression Advances in Psychiatric Treatment, 14, 29–36. doi: 10.1192/apt.bp.107.004051 [DOI] [Google Scholar]
  48. Watson D, & Clark LA (1984). Negative Affectivity - the Disposition to Experience Aversive Emotional States. Psychological bulletin, 96(3), 465–490. doi:Doi 10.1037//0033-2909.96.3.465 [DOI] [PubMed] [Google Scholar]
  49. Weinstein AA, Koehmstedt C, & Kop WJ (2017). Mental health consequences of exercise withdrawal: A systematic review. Gen Hosp Psychiatry, 49, 11–18. doi: 10.1016/j.genhosppsych.2017.06.001 [DOI] [PubMed] [Google Scholar]
  50. Wilkowski BM, & Robinson MD (2010). The anatomy of anger: an integrative cognitive model of trait anger and reactive aggression. Journal of personality, 78(1), 9–38. doi: 10.1111/j.1467-6494.2009.00607.x [DOI] [PubMed] [Google Scholar]
  51. Williams JE, Nieto FJ, Sanford CP, & Tyroler HA (2001). Effects of an angry temperament on coronary heart disease risk: The Atherosclerosis Risk in Communities Study. American journal of epidemiology, 154(3), 230–235. doi: 10.1093/aje/154.3.230 [DOI] [PubMed] [Google Scholar]
  52. Wipfli BM, Rethorst CD, & Landers DM (2008). The anxiolytic effects of exercise: a meta-analysis of randomized trials and dose-response analysis. J Sport Exerc Psychol, 30(4), 392–410. [DOI] [PubMed] [Google Scholar]
  53. Zhai L, Zhang Y, & Zhang D (2015). Sedentary behaviour and the risk of depression: a meta-analysis. Br J Sports Med, 49(11), 705–709. doi: 10.1136/bjsports-2014-093613 [DOI] [PubMed] [Google Scholar]

RESOURCES