Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2017 Nov 6.
Published in final edited form as: J Clin Psychiatry. 2016 Aug;77(8):1036–1042. doi: 10.4088/JCP.15m10104

Affect Following First Exercise Session as a Predictor of Treatment Response in Depression

Anisha M Suterwala 1, Chad D Rethorst 2, Thomas J Carmody 2, Tracy L Greer 2, Bruce D Grannemann 2, Manish Jha 2, Madhukar H Trivedi 2
PMCID: PMC5673095  NIHMSID: NIHMS916684  PMID: 27561137

Abstract

Objective

Remission rates are low with antidepressant treatments either as a first step or even with a second step. Furthermore, despite extensive investments from NIH and Industry, novel treatments are not yet available in clinical care for depression. Predictors of treatment response very early in the course of treatment can avoid unnecessarily lengthy trials with ineffective treatments and reduce the trial and error process. This paper examines the expression of positive affect immediately following an acute exercise session at the end of the first exercise session as a predictor of treatment response in the NIMH-funded TrEAD (Treatment with Exercise Augmentation for Depression) study, which was conducted from April 2003–August 2007.

Method

122 subjects with DSM-IV diagnosed Major Depressive Disorder were randomized to public health dose (16 kcal/kg/week or KKW) or low dose (4 KKW) of exercise for 12 weeks. Affect following the first exercise session was assessed using the Positive and Negative Affect Scale (PANAS) and depressive symptoms were assessed weekly using the Inventory for Depressive Symptoms – Clinician Rated (IDS-C).

Results

The PANAS Composite Affective (CA) score (positive - negative total) predicted change in IDS-C score (p < .05), as well as treatment response (p < .02) and remission (p < .03) for those in the high dose group but not in the low dose group.

Conclusions

These findings suggest that the composite positive affect following the first exercise session has clinical utility to predict treatment response to exercise in depression and match the “right patient” with the “right” treatment.

Clinical Trials Registration

Clinicaltrials.gov NCT00076258

Introduction

Although numerous treatments are available for Major Depressive Disorder (MDD), selecting among these options remains the biggest challenge facing clinicians. Remission rates are low with standard antidepressant treatments with the first 1 or even second step treatments 2 in the short term and even lower remission rates in the long term 2. Non-pharmacological interventions like psychotherapy and aerobic exercise have also been shown to be efficacious for the treatment of MDD 3–5. While treatment outcomes do improve with additional treatment steps, in the absence of reliable predictors and moderators 6, the trial and error process required to identify appropriate treatment for a given individual prolongs the disease burden in these patients. Therefore, there is a clear need to identify behavioral or biological markers in order to determine who is most likely to respond to any given treatment. Identification of these markers will lead to early treatment matching, reducing the burden of multiple steps for the patient and ultimately reducing the burden of the illness.

The link between exercise and mental health is well documented. Exercise has long-term benefits for general health, and physically active people are at a lower risk of developing mood and anxiety disorders 7–12. Research also supports the use of exercise in the treatment of MDD. Exercise has proven efficacious as a monotherapy 5, 13, 14 as well as an augmentation treatment of MDD 15, 16. As an intervention, exercise has the additional advantage of targeting residual symptoms such as insomnia, psychosocial functioning, and cognitive function 17–19. Despite the benefits of exercise as a treatment, it is like other antidepressant treatments, not universally effective. Determining clinically useful behavioral and biological markers as predictors of response to exercise can improve the process of treatment matching for exercise as an antidepressant.

Post-hoc analyses of prior studies have attempted to identify predictors of response to exercise. Using this baseline model, our preliminary work has already shown that insomnia, TNF-alpha and BDNF are predictive of treatment response and suggest these as possible biological markers for exercise augmentation 17, 20, 21. However, it is clear that identification of multiple predictors is necessary in order for these predictors to have clinical utility (Establishing Moderators and Biosignatures of Antidepressant Response In Clinical Care (EMBARC): Rationale and Design, Trivedi et al Journal of Psychiatric Research resubmitted).

Exercise is not only effective in reducing depressive symptoms, but a single bout of exercise results in acute improvements in affect in patients with MDD 22, 23. However, no study has evaluated whether this acute affective response post-exercise is predictive of long-term treatment outcome. Just as long-term reductions in depressive symptoms are heterogeneous, acute affective responses to exercise also vary across individuals. This variation has important implications for exercise treatment interventions, as positive affective response to an exercise session has been associated with long-term exercise adherence 24. Furthermore, research has identified genetic predictors of affective response 25, 26, indicating this variation can be attributed to underlying biological mechanisms. It is therefore possible that these biological mechanisms underlying variations in affective response to acute exercise may be the same mechanisms underlying variations in treatment response to exercise.

In this study, we examine positive affect following the initial exercise intervention session as a predictor of final treatment outcomes. The Treatment with Exercise Augmentation for Depression (TREAD) study concluded that exercise augmentation with a public health dose is an effective means to achieve remission in patients who are nonresponsive to initial SSRI treatment, with a number needed to treat of 7.8 compared to the lower dose of exercise 15, 27. This paper examines: 1) whether affect immediately following the first exercise session predicts treatment outcomes at the end of the 12-week exercise intervention, 2) whether this predictive effect is moderated by exercise dose, and 3) if the predictive effect is mediated by exercise adherence. We hypothesize that greater post-exercise positive affect will be predictive of greater decreases in depressive symptoms at the end of the intervention. Furthermore, we hypothesize that this effect will be greater in the high dose exercise treatment group. Finally, we hypothesize that the predictive effect of post-exercise affect will be, at least partially, explained by differences in exercise adherence.

Methods

The TREAD trial was a randomized trial comparing two doses of aerobic exercise as augmentation treatment for non-remitted MDD. Full study methodology has been previously published 15, 28; included below is a description of the study procedures relevant to the current analysis.

Participants

Participants were recruited through physician referrals and advertisements. Eligible participants signed an IRB-approved informed consent before engaging in any study activities. Inclusion criteria were as follows: (1) men and women aged 18–70 years; (2) diagnosis of nonpsychotic major depressive disorder (MDD) based on the Structured Clinical Interview for DSM-IV Axis I Disorders; (3) completion of >2 and <6 months of treatment with a selective serotonin reuptake inhibitor (SSRI) with at least six weeks of adequate dose; (4) moderate residual symptomatology from SSRI monotherapy defined by a score of ≥ 14 on the Hamilton Rating Scale for Depression; (5) not engaged in regular exercise; (6) capable of exercise; and (7) able and willing to provide informed consent.

Exclusion criteria included: (1) significant medical condition; (2) depression due to a comorbid psychiatric disorder; (3) current pharmacological or psychotherapeutic treatment other than SSRI; (4) treatment resistance, defined as failure of two or more pharmacologic treatments of adequate dose and duration; (5) pregnancy or planned pregnancy.

Procedures

Participants were randomly allocated to one of two exercise groups for a 12-week exercise intervention. The dose of SSRI was kept constant during the exercise treatment. The Low Dose (LD) group was prescribed a dose of exercise at 4 kilocalories per kilogram of body weight (KKW) and the Public Health Dose (PHD) group was prescribed a higher dose of exercise at 16 KKW, reflecting public health physical activity recommendations and standards 29. The LD dose is equivalent to approximately 45 minutes of moderate-to-vigorous exercise per week, and the PHD dose is equivalent to approximately 180 minutes of moderate-to-vigorous exercise per week. Exercise intensity was self-selected and monitored with a Polar 610i heart rate monitor. Participants completed the entire first week of exercise dose under supervision by trained personnel at the Cooper Institute, and had two supervised sessions in the second week. In all subsequent weeks, participants had one supervised session per week and completed the remaining exercise dose in home-based sessions. All study procedures were approved by the local Institutional Review Board.

Measures

The Positive and Negative Affect Scale (PANAS) was administered following the first exercise session of Week 1. The PANAS is a twenty-item scale with two ten-item subscales designed to measure affect. PANAS measures affect as both positive and negative. High positive affect (PA) reflects an individual’s feelings using adjectives such as enthusiastic, active, and alert, and low PA is characterized by adjectives such as sadness and lethargy. High negative affect (NA) is a general dimension of subjective distress characterized by anger, disgust, guilt, fear, and nervousness, and low NA is characterized by calmness and serenity. The PANAS has been widely used as a measure of acute affect with good psychometric properties, 30, 31 with reliability for the immediate assessment at alpha of .89 for PA scores and .85 for NA scores and has been further validated specifically within a psychiatric population 32. In addition to the standard Positive and Negative affect scores, we also derived a composite affect score (CA) by combining the two measures in order to assess the overall general affect to the exercise session, accounting for both positive and negative aspects of affect. To evaluate the general affective state at the end of the exercise session, the positive and negative PANAS scores were combined. This was done by the subtraction of the negative PANAS total from the positive PANAS total to produce the CA score.

The Inventory for Depressive Symptoms – Clinician Rated version (IDS-C) was used to measure depressive symptoms. The IDS-C is a 30-item rating scale with excellent psychometric properties and has been widely used in clinical trials33. The IDS-C was administered weekly by blinded raters. Change in IDS-C was calculated as the Week 12 IDS-C total score minus the Baseline IDS-C total score. If data were missing for the Week12 IDS-C score, the last assessed value was carried forward. Response and remission were also determined based on the IDS-C total score values. Response was defined as a 50% reduction in IDS-C score, while remission was defined as an IDS-C score of 12 or less at study completion.

Statistical Analyses

Primary analyses

Primary analyses were designed to evaluate the composite affect (CA) score following the first acute exercise session as a predictor of change in depressive symptoms from baseline to exit. The CA score at the conclusion of the first exercise session was used as a predictor of treatment response for the primary analysis. Change in depressive symptoms was calculated using the IDS-C total. A general linear model included main effects for both Exercise Group and CA score along with Group by CA interaction as the predictors of IDS-C change (decrease from baseline to exit). The main effect of group provides a measure of differences in treatment response, the main effect of CA provides the measure of the ability of affect after an exercise session to predict treatment response, and the interaction tests if the prediction is specific to one or the other of the two treatments. This model was repeated using both the PA and NA subscale scores.

Covariate analyses

To control for the effects of the influence of possible confounding variables a series of covariate analyses were conducted. Gender and family history were included as these covariates were associated with treatment outcome in primary outcome paper15. We also included fitness (VO2max) and exercise adherence (median percent of weekly KKW dose). The covariate analysis was conducted as a two step process that would include all variables plus two way interaction as a first step. As a second step, all covariates with a p value <.10 were retained in the final covariate model34.

Prediction of response and remission analyses

We conducted logistic regression models to examine treatment response (50% reduction from baseline) and treatment remission (IDS-C total score ≤ 12). In these models, exercise group and CA score are the main effects and a group by CA interaction were predictors of response and remission. To examine the utility of the CA score to predict the likelihood of response and remission, we also conducted an ROC analysis and report the Area Under the Curve (AUC) for the ROC. Perfect prediction would have an AUC of 1.0, with an AUC above .90 still providing a clinically useful test; an AUC above .80 is considered meaningful but with modest clinical usefulness 35, 36. As an example, the AUC for the PHQ-9 self-report and a clinical diagnosis of MDD is about .88 37. Finally, to test for any impact of exercise adherence over the course of the treatment trial, a set of analyses were run that included adherence (as measured by the median number of sessions) added to the general linear models with interaction terms for both Group and CA.

Results

Baseline characteristics of the study sample are provided in Table 1. One-hundred twenty-six subjects completed the baseline assessments and were randomized to an exercise condition. Four subjects did not provide post-baseline data and were excluded from the analyses. Mean changes in IDS-C, response and remission for the total sample and by group and PANAS scores following the initial exercise session are presented in Table 2. T-tests conducted to determine if there were differences in Week 1 PANAS scores (PA, PN, & CA) yielded no significant differences.

Table 1.

Baseline demographic and clinical characteristics means (standard deviations).

Baseline Variable All (n=118) 16 kcals (n=58) 4 kcals (n=60)
Demographics
Age (yrs) 47.1 (10.0) 45.7 (10.4) 48.5 (9.4)
Female no. (%) 96 (81.4) 49(84.5) 47 (78.3)
Race
 White no. (%) 103 (86.3) 49 (81.7) 54 (93.1)
 Black no. (%) 12 (10.2) 8 (13.3) 4 (6.9)
 Hispanic no. (%) 1 (0.8) 3 (5.0) 1 (1.7)
 Other no. (%) 2 (1.7) 1 (1.6) 3 (5.2)
Marital Status
 Single, never married (%) 20 (17.0) 10(17.2) 10(16.7)
 Cohabitating (%) 5 (4.2) 3(5.2) 2(3.3)
 Married, living together (%) 62 (50.8) 31(53.4) 30(50.0)
 Married, living apart (%) 1 (0.8) 0(0.0) 1(1.7)
 Separated (%) 1 (0.8) 0(0.0) 1(1.7)
 Divorced (%) 27 (22.9) 12(20.1) 15(25.0)
 Widowed (%) 3 (2.5) 2(3.5) 1(1.7)
Weight (KG) 86.8 (20.0) 87.0 (23.6) 87.8 (17.7)
BMI 30.6 (6.09) 29.9 (6.3) 31.5 (5.5)
VO2 max (l/min) 1.7 (0.6) 1.7 (0.5) 1.8 (0.7)
Age of Onset (yrs) 27.1 (11.3) 27.4 (10.7) 26.8 (12.0)
Length of current episode (mos) 82.1 (97.1) 73.4 (95.0) 90.5 (99.0)
Baseline Symptom Severity
 HRSD17 17.8 (3.7) 17.6 (3.6) 18.0 (3.8)
 IDS-C30 33.8 (7.5) 33.2 (7.1) 34.7 (7.8)
Function
 Q-LES-Q GA 41.9 (7.6) 41.3(7.6) 42.4(7.5)
 WSAS 20.2(8.9) 19.2(7.7) 21.0(9.9)
SF-36 Physical 80.1 (20.4) 80.2 (20.1) 80.1 (20.8)
SF-36 Mental 49.4 (15.4) 49.8 (14.9) 49.0 (16.0)
Open in a new tab

Body Mass Index (BMI),17-item Hamilton Rating Scale for Depression (HRSD17), Inventory of Depressive Symptomatology–Clinician-rated (IDS-C30), Quality of Life Enjoyment and Satisfaction Questionnaire, General Activities(Q-LES-Q GA), SF-36, maximal oxygen consumption (VO2 max), and Work and Social Adjustment Scale (WSAS).

Table 2.

IDS-C decrease from baseline and PANAS scores at end of first exercise session: means (standard deviations).

Variable All (n=118) 16 Kcals (n=58) 4 Kcals (n=60)
IDSC30 Change 12.2 (11.7) 11.0 (11.6) 13.3 (9.6)
PANAS Positive 29.3 (9.1) 28.7 (8.6) 29.9 (9.6)
PANAS Negative 13.2 (4.4) 13.5 (4.6) 13.0 (4.2)
PANAS Composite 16.1 (11.0) 15.2 (11.0) 16.9 (11.0)
Open in a new tab

CA score as a predictor of change in depressive symptoms

The primary analysis for Week 1 included 118 of the 122 subjects, as 4 subjects did not provide PANAS ratings. There was a significant main effect for Group [F(1,114)=4.87 p < .03: eta2 = .041], a marginal effect for CA [F(1,115)=3.36, p< .07: eta2 = .029], and a significant Group by CA interaction [F(1,115)=4.13, p < .05: eta2 = .035]. As seen in Figure 1, higher CA scores predicted better outcomes (as measured by reduction in total IDS-C scores) for subjects in the PHD group but not in the LD group.

Figure 1.

Figure 1

Open in a new tab

Prediction of treatment outcomes using PANAS CA score at end of first exercise session (week1).

Covariate analysis

The first step of the covariate analyses included gender, family history of mental illness, fitness and exercise adherence and two-way interactions in the model with group and CA. Only the exercise adherence [F(1,106)= 5.98 p < .02: eta2= 04] and the family history of mental illness by Group interaction [F(1,106)= 3.43 p < .07; eta2= 03] produced p values < .10. The final model that include just these two covariates with Group, CA and the Group by CA interaction. The main effect for median percent KKW of dose [F(1,106)=5.9=60 p < .02]: eta2= .04], family history of mental illness [F(1,106)=4.76 p < .04]: eta2= .03], family history of mental illness by Group interaction [F(1,106)=4.66 p < .01: eta2 = .03] and the Group by CA interaction dose [F(1,106)=5.01 p < .03]: eta2= .04]. The group by CA interaction remained significant and is not moderated by the either exercise adherence or family history. Note the family history results were previously reported in the primary outcome paper (Trivedi et al 2011).

CA score as a predictor of response and remission

The logistic regression at Week 1, using CA and group to predict response, produced significant main effects for Group [X2 (3)=3.88], p < .05] and CA [X2 (3)=5.45, p < .02] and a significant Group by CA interaction [X2 (3)=5.46, p < .02: odds ratio LD =1.00, PHD=1.10]. Similarly, the logistic regression predicting remission produced a marginal main effect for Group [X2 (3)=3.07, p < .08], a significant main effect for CA [X2 (3)=8.56, p < .004] and a significant Group by CA interaction [X2 (3)=4.55, p < .03: Odds Ratios for LD =1.02, PHD=1.13]. These results indicate that higher CA scores predicted a higher likelihood of response and remission within the PHD group.

In order to better understand the extent to which acute affect following the first exercise session can be used to predict treatment response and remission, ROC analyses were also conducted. Since the models for both response and remission produced a significant interaction with group, separate ROC curves were produced for each group. For the LD group, CA does not predict response [X2 (1) =.00, p < .99: ROC curve Area Under the Curve .51], but CA does predict response for the PHD group [X2 (1)=8.4, p < .004: ROC curve Area Under the Curve .76]. The same pattern appears for remission, with CA not predicting remission for the LD group [X2 (1) =.51, p < .48: ROC curve Area Under the Curve .57], but predicting remission for the PHD group [X2 (1)=9.2, p < .003: ROC curve Area Under the Curve .81]. ROC curves are presented in Figure 2.

Figure 2.

Figure 2

Open in a new tab

ROC analysis of the PANAS CA scores for prediction of Response and Remisson.

To better understand these results, analyses using the PA and NA subscales of the PANAS were conducted. The model included a significant main effect for Group [F(1,114)=4.09 p < .05: eta2 = .02], a significant main effect for PA [F(1,114)=6.14, p< .02: eta2 = .05], and a marginal Group by PA interaction [F(1,114)=3.34., p < .08: eta2 = .03]. The results for the NA analysis indicated there is no main for Group [F(1,115)=1.19 p < .36], no main effect for NA [F(1,115)=.10, p< .76: eta2 = .00008], and a marginal Group by RN interaction [F(1,115)=3.37, p < .07: eta2 = .03]. These finding are only worth noting because they provide some insight as to why CA finding are more robust than the positive finding alone. For the PHD group higher NA scores predict less symptom reduction, but for the LD groups the contrary is true with higher NA scores predicting greater symptom reduction.

Discussion

The primary goal of this study was to examine affect following the first exercise session as a predictor of treatment response for patients with non-remitted MDD. Our results specifically indicate that greater positive affect following the first exercise session was predictive of greater improvements in depressive symptoms, response and remission at the end of the 12-week exercise intervention. Most interestingly, this effect was moderated by treatment group, such that the predictive value of positive affect was only significant in the high dose exercise group but not in the low dose group.

Previous research has shown variation in affective responses to acute exercise sessions 32, 33, 25, 26. Our results suggest that variations in post-exercise affect might be predictive of antidepressant response. In light of these results, it is of interest to determine the underlying mechanism of this effect. Potential explanations for the association of post-exercise affect with the longer-term acute phase antidepressant effect of exercise could be a result of enhanced self-efficacy, improved adherence to the exercise intervention, the patient’s perception of early antidepressant effect or the capacity of the underlying neural circuitry to respond. Previous work has linked positive acute affect following a single exercise session with improved exercise adherence over an extended period of time 24. However, including adherence in the regression model did not diminish the observed effect of post-exercise affect.

Variations in affective response to an acute exercise session can also be due to the characteristics of exercise itself. For example, high exercise intensity, specifically intensity above the ventilator threshold, results in more negative post-exercise affect 38, 39. A unique feature of our exercise intervention in the current study was for participants to self-select their exercise intensity, which has been associated with positive post-exercise affect. The fact that the affective response to exercise is modulated by characteristics of the exercise might also explain the moderating effect of treatment group observed in our analysis. Due to the higher exercise dose, exercise sessions among individuals in the PHD group were of a greater duration (mean duration of 45 minutes vs. 25 minutes) and perhaps the longer sessions may elicit an affective response that is more predictive of treatment outcome. Given that characteristics of the exercise session can alter post-exercise affect, it is possible that tailoring an exercise intervention to elicit positive affect could result in a more effective exercise intervention.

Finally, the predictive relationship between post-exercise affect and long-term antidepressant outcomes may be indicative of common underlying biology. Research has identified genetic associations with variations in post-exercise affect 25, 26 as well as to antidepressant response to an exercise intervention 40. It is possible that common genetic polymorphisms may predict both acute affective response and treatment outcomes in response to exercise.

Exercise has been shown to be efficacious in the treatment of MDD but many patients do not improve following an exercise intervention. Given the heterogeneity in treatment response, it is necessary to identify predictors of treatment response in order to optimally treat MDD. Our results suggest that post-exercise affect may be a potential predictor of long-term antidepressant outcomes. This analysis continues a line of research examining predictors of treatment response to exercise in patients with MDD 17, 20, 41, 42. Despite these encouraging findings as a predictor of treatment response following the first exercise session, the current analysis does have limitations. This analysis represents a secondary data analysis, as the TREAD trial was not specifically designed to identify predictors of treatment response. Instead, our results should be viewed as “hypothesis generating” and future trials should be designed to specifically identify predictors of treatment response. Second, the PANAS was only collected post-exercise and therefore we were unable to examine changes in affect pre- and post-exercise. Finally, it is clear that a single predictor will not suffice to guide treatment selection and that a metric comprised of several markers is very likely to be needed to enhance the accuracy of prediction 43(Establishing Moderators and Biosignatures of Antidepressant Response In Clinical Care (EMBARC): Rationale and Design, Trivedi et al Journal of Psychiatric Research resubmitted). As work in the field continues to identify predictors of treatment response, these individual predictors must be integrated into comprehensive algorithms to improve predictive abilities. Ultimately, this work will lead to matching patients with treatments most likely to elicit a treatment response for them.

Clinical Points.

  • Optimal treatment of MDD requires identification of treatment outcome predictors

  • Post-exercise affect appears to be a potential predictor of depression treatment outcome

  • However, predictive capability of this, or any other individual predictor, is limited. Future work should aim to synthesize multiple predictors.

Acknowledgments

Funding/support: This work was supported by the National Institute for Mental Health (1-R01-MH067692-01; PI: MH Trivedi) and in part by a National Alliance for Research on Schizophrenia and Depression (NARSAD) Independent Investigator Award (MHT), and Young Investigator Award (TLG). Chad D. Rethorst is supported by the National Institute of Mental Health of the National Institutes of Health under Award Number K01MH097847.

The authors would like to thank all those who assisted with this project. The authors also recognize, with great appreciation, all the study participants who contributed to this project.

Footnotes

Role of the Sponsors: Neither NIMH nor NARSAD were involved in the design or completion of the study.

Previous presentation: None

Potential conflicts of interest (Past 12 months):

Dr. Trivedi has been an advisor/consultant and received fees from - Alkermes, AstraZeneca, Cerecor, Eli Lilly & Company, Lundbeck, Naurex, Neuronetics, Otsuka Pharmaceuticals, Pamlab, Pfizer Inc., SHIRE Development and Takeda.

In addition, Dr. Trivedi has received grants/research support from: National Institute of Mental Health (NIMH) and National Institute on Drug Abuse (NIDA).

Dr. Greer has received honoraria, speakers or advisory boards and/or consultant fees from H. Lundbeck A/S.

Ms. Suterwala, Dr. Rethorst, Dr. Carmody, Dr. Manish Jha and Mr. Grannemann have no conflict of interests to disclose.

References

  • 1.Trivedi MH, Rush AJ, Wisniewski SR, et al. Evaluation of outcomes with citalopram for depression using measurement-based care in STAR*D: implications for clinical practice. The American journal of psychiatry. 2006 Jan;163(1):28–40. doi: 10.1176/appi.ajp.163.1.28. [DOI] [PubMed] [Google Scholar]
  • 2.Rush AJ, Trivedi MH, Wisniewski SR, et al. Acute and longer-term outcomes in depressed outpatients requiring one or several treatment steps: a STAR*D report. The American journal of psychiatry. 2006 Nov;163(11):1905–1917. doi: 10.1176/ajp.2006.163.11.1905. [DOI] [PubMed] [Google Scholar]
  • 3.Dobson KS. A meta-analysis of the efficacy of cognitive therapy for depression. Journal of consulting and clinical psychology. 1989 Jun;57(3):414–419. doi: 10.1037//0022-006x.57.3.414. [DOI] [PubMed] [Google Scholar]
  • 4.Gloaguen V, Cottraux J, Cucherat M, Blackburn IM. A meta-analysis of the effects of cognitive therapy in depressed patients. Journal of affective disorders. 1998 Apr;49(1):59–72. doi: 10.1016/s0165-0327(97)00199-7. [DOI] [PubMed] [Google Scholar]
  • 5.Rethorst CD, Wipfli BM, Landers DM. The antidepressive effects of exercise: a meta-analysis of randomized trials. Sports medicine (Auckland, NZ) 2009;39(6):491–511. doi: 10.2165/00007256-200939060-00004. [DOI] [PubMed] [Google Scholar]
  • 6.Trivedi MH. Modeling predictors, moderators and mediators of treatment outcome and resistance in depression. Biological psychiatry. 2013 Jul 1;74(1):2–4. doi: 10.1016/j.biopsych.2013.05.009. [DOI] [PubMed] [Google Scholar]
  • 7.Deslandes AC. Exercise and Mental Health: What did We Learn in the Last 20 Years? Frontiers in psychiatry. 2014;5:66. doi: 10.3389/fpsyt.2014.00066. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Galper DI, Trivedi MH, Barlow CE, Dunn AL, Kampert JB. Inverse association between physical inactivity and mental health in men and women. Medicine and science in sports and exercise. 2006 Jan;38(1):173–178. doi: 10.1249/01.mss.0000180883.32116.28. [DOI] [PubMed] [Google Scholar]
  • 9.Camacho TC, Roberts RE, Lazarus NB, Kaplan GA, Cohen RD. Physical activity and depression: evidence from the Alameda County Study. American journal of epidemiology. 1991 Jul 15;134(2):220–231. doi: 10.1093/oxfordjournals.aje.a116074. [DOI] [PubMed] [Google Scholar]
  • 10.Farmer ME, Locke BZ, Moscicki EK, Dannenberg AL, Larson DB, Radloff LS. Physical activity and depressive symptoms: the NHANES I Epidemiologic Follow-up Study. American journal of epidemiology. 1988 Dec;128(6):1340–1351. doi: 10.1093/oxfordjournals.aje.a115087. [DOI] [PubMed] [Google Scholar]
  • 11.Dishman RK, Buckworth J. Increasing physical activity: a quantitative synthesis. Medicine and science in sports and exercise. 1996 Jun;28(6):706–719. doi: 10.1097/00005768-199606000-00010. [DOI] [PubMed] [Google Scholar]
  • 12.Dishman RK, Berthoud HR, Booth FW, et al. Neurobiology of exercise. Obesity (Silver Spring, Md) 2006 Mar;14(3):345–356. doi: 10.1038/oby.2006.46. [DOI] [PubMed] [Google Scholar]
  • 13.Blumenthal JA, Babyak MA, Doraiswamy PM, et al. Exercise and pharmacotherapy in the treatment of major depressive disorder. Psychosomatic medicine. 2007 Sep-Oct;69(7):587–596. doi: 10.1097/PSY.0b013e318148c19a. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Dunn AL, Trivedi MH, Kampert JB, Clark CG, Chambliss HO. Exercise treatment for depression: efficacy and dose response. American journal of preventive medicine. 2005 Jan;28(1):1–8. doi: 10.1016/j.amepre.2004.09.003. [DOI] [PubMed] [Google Scholar]
  • 15.Trivedi MH, Greer TL, Church TS, et al. Exercise as an augmentation treatment for nonremitted major depressive disorder: a randomized, parallel dose comparison. The Journal of clinical psychiatry. 2011 May;72(5):677–684. doi: 10.4088/JCP.10m06743. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Mather AS, Rodriguez C, Guthrie MF, McHarg AM, Reid IC, McMurdo ME. Effects of exercise on depressive symptoms in older adults with poorly responsive depressive disorder: randomised controlled trial. The British journal of psychiatry: the journal of mental science. 2002 May;180:411–415. doi: 10.1192/bjp.180.5.411. [DOI] [PubMed] [Google Scholar]
  • 17.Rethorst CD, Sunderajan P, Greer TL, et al. Does exercise improve self-reported sleep quality in non-remitted major depressive disorder? Psychological medicine. 2013 Apr;43(4):699–709. doi: 10.1017/S0033291712001675. [DOI] [PubMed] [Google Scholar]
  • 18.Mota-Pereira J, Carvalho S, Silverio J, et al. Moderate physical exercise and quality of life in patients with treatment-resistant major depressive disorder. Journal of psychiatric research. 2011 Dec;45(12):1657–1659. doi: 10.1016/j.jpsychires.2011.08.008. [DOI] [PubMed] [Google Scholar]
  • 19.Greer TL, Grannemann BD, Chansard M, Karim AI, Trivedi MH. Dose-dependent changes in cognitive function with exercise augmentation for major depression: Results from the TREAD study. European neuropsychopharmacology: the journal of the European College of Neuropsychopharmacology. 2015 Feb;25(2):248–256. doi: 10.1016/j.euroneuro.2014.10.001. [DOI] [PubMed] [Google Scholar]
  • 20.Toups MS, Greer TL, Kurian BT, et al. Effects of serum Brain Derived Neurotrophic Factor on exercise augmentation treatment of depression. Journal of psychiatric research. 2011 Oct;45(10):1301–1306. doi: 10.1016/j.jpsychires.2011.05.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Rethorst CD, Toups MS, Greer TL, et al. Pro-inflammatory cytokines as predictors of antidepressant effects of exercise in major depressive disorder. Molecular psychiatry. 2013 Oct;18(10):1119–1124. doi: 10.1038/mp.2012.125. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Bartholomew JB, Morrison D, Ciccolo JT. Effects of acute exercise on mood and well-being in patients with major depressive disorder. Medicine and science in sports and exercise. 2005 Dec;37(12):2032–2037. doi: 10.1249/01.mss.0000178101.78322.dd. [DOI] [PubMed] [Google Scholar]
  • 23.Weinstein AA, Deuster PA, Francis JL, Beadling C, Kop WJ. The role of depression in short-term mood and fatigue responses to acute exercise. International journal of behavioral medicine. 2010 Mar;17(1):51–57. doi: 10.1007/s12529-009-9046-4. [DOI] [PubMed] [Google Scholar]
  • 24.Williams DM. Exercise, affect, and adherence: an integrated model and a case for self-paced exercise. Journal of sport & exercise psychology. 2008 Oct;30(5):471–496. doi: 10.1123/jsep.30.5.471. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.de Geus EJ, Bartels M, Kaprio J, Lightfoot JT, Thomis M. Genetics of regular exercise and sedentary behaviors. Twin research and human genetics: the official journal of the International Society for Twin Studies. 2014 Aug;17(4):262–271. doi: 10.1017/thg.2014.42. [DOI] [PubMed] [Google Scholar]
  • 26.Karoly HC, Stevens CJ, Magnan RE, Harlaar N, Hutchison KE, Bryan AD. Genetic Influences on Physiological and Subjective Responses to an Aerobic Exercise Session among Sedentary Adults. Journal of cancer epidemiology. 2012;2012:540563. doi: 10.1155/2012/540563. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Trivedi MH, Greer TL, Grannemann BD, Chambliss HO, Jordan AN. Exercise as an augmentation strategy for treatment of major depression. Journal of psychiatric practice. 2006 Jul;12(4):205–213. doi: 10.1097/00131746-200607000-00002. [DOI] [PubMed] [Google Scholar]
  • 28.Trivedi MH, Greer TL, Grannemann BD, et al. TREAD: TReatment with Exercise Augmentation for Depression: study rationale and design. Clinical trials (London, England) 2006;3(3):291–305. doi: 10.1191/1740774506cn151oa. [DOI] [PubMed] [Google Scholar]
  • 29.Haskell WL, Lee IM, Pate RR, et al. Physical activity and public health: updated recommendation for adults from the American College of Sports Medicine and the American Heart Association. Circulation. 2007 Aug 28;116(9):1081–1093. doi: 10.1161/CIRCULATIONAHA.107.185649. [DOI] [PubMed] [Google Scholar]
  • 30.Watson D, Clark LA, Tellegen A. Development and validation of brief measures of positive and negative affect: the PANAS scales. Journal of personality and social psychology. 1988 Jun;54(6):1063–1070. doi: 10.1037//0022-3514.54.6.1063. [DOI] [PubMed] [Google Scholar]
  • 31.Crawford JR, Henry JD. The positive and negative affect schedule (PANAS): construct validity, measurement properties and normative data in a large non-clinical sample. The British journal of clinical psychology/the British Psychological Society. 2004 Sep;43(Pt 3):245–265. doi: 10.1348/0144665031752934. [DOI] [PubMed] [Google Scholar]
  • 32.Dyck MJ, Jolly JB, Kramer T. An evaluation of positive affectivity, negative affectivity, and hyperarousal as markers for assessing between syndrome relationships. Personality and Individual Differences. 1994 Nov;17(5):637–646. [Google Scholar]
  • 33.Trivedi MH, Rush AJ, Ibrahim HM, et al. The Inventory of Depressive Symptomatology, Clinician Rating (IDS-C) and Self-Report (IDS-SR), and the Quick Inventory of Depressive Symptomatology, Clinician Rating (QIDS-C) and Self-Report (QIDS-SR) in public sector patients with mood disorders: a psychometric evaluation. Psychological medicine. 2004 Jan;34(1):73–82. doi: 10.1017/s0033291703001107. [DOI] [PubMed] [Google Scholar]
  • 34.Bursac Z, Gauss CH, Williams DK, Hosmer DW. Purposeful selection of variables in logistic regression. Source code for biology and medicine. 2008;3:17. doi: 10.1186/1751-0473-3-17. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Fan J, Upadhye S, Worster A. Understanding receiver operating characteristic (ROC) curves. Cjem. 2006 Jan;8(1):19–20. doi: 10.1017/s1481803500013336. [DOI] [PubMed] [Google Scholar]
  • 36.Fan J, Worster A, Fernandes CM. Predictive validity of the triage risk screening tool for elderly patients in a Canadian emergency department. The American journal of emergency medicine. 2006 Sep;24(5):540–544. doi: 10.1016/j.ajem.2006.01.015. [DOI] [PubMed] [Google Scholar]
  • 37.Kiely KM, Butterworth P. Validation of four measures of mental health against depression and generalized anxiety in a community based sample. Psychiatry research. 2015 Feb 28;225(3):291–298. doi: 10.1016/j.psychres.2014.12.023. [DOI] [PubMed] [Google Scholar]
  • 38.Ekkekakis P, Lind E, Vazou S. Affective responses to increasing levels of exercise intensity in normal-weight, overweight, and obese middle-aged women. Obesity (Silver Spring, Md) 2010 Jan;18(1):79–85. doi: 10.1038/oby.2009.204. [DOI] [PubMed] [Google Scholar]
  • 39.Ekkekakis P, Parfitt G, Petruzzello SJ. The pleasure and displeasure people feel when they exercise at different intensities: decennial update and progress towards a tripartite rationale for exercise intensity prescription. Sports medicine (Auckland, NZ) 2011 Aug 1;41(8):641–671. doi: 10.2165/11590680-000000000-00000. [DOI] [PubMed] [Google Scholar]
  • 40.Rethorst CD, Landers DM, Nagoshi CT, Ross JT. Efficacy of exercise in reducing depressive symptoms across 5-HTTLPR genotypes. Medicine and science in sports and exercise. 2010 Nov;42(11):2141–2147. doi: 10.1249/MSS.0b013e3181de7d51. [DOI] [PubMed] [Google Scholar]
  • 41.Rethorst CD, Moynihan J, Lyness JM, Heffner KL, Chapman BP. Moderating effects of moderate-intensity physical activity in the relationship between depressive symptoms and interleukin-6 in primary care patients. Psychosomatic medicine. 2011 Apr;73(3):265–269. doi: 10.1097/PSY.0b013e3182108412. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Rethorst CD, Trivedi MH. Evidence-based recommendations for the prescription of exercise for major depressive disorder. Journal of psychiatric practice. 2013 May;19(3):204–212. doi: 10.1097/01.pra.0000430504.16952.3e. [DOI] [PubMed] [Google Scholar]
  • 43.Wallace ML, Frank E, Kraemer HC. A novel approach for developing and interpreting treatment moderator profiles in randomized clinical trials. JAMA psychiatry. 2013 Nov;70(11):1241–1247. doi: 10.1001/jamapsychiatry.2013.1960. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES