ÌÇÐÄvlog

Key Points

QuestionÌý Do aerobic exercise and escitalopram reduce anxiety and improve coronary heart disease biomarkers more than a placebo?

FindingsÌý In this randomized clinical trial including 128 patients with coronary heart disease and anxiety, escitalopram, but not exercise, resulted in reduced levels of anxiety and depression compared with a placebo control.

MeaningÌý For patients in this study with coronary heart disease and high anxiety, escitalopram was an effective treatment for reducing anxiety, although the extent to which this benefit may improve clinical outcomes remains uncertain.

ImportanceÌý Anxiety is common among patients with coronary heart disease (CHD) and is associated with worse health outcomes; however, effective treatment for anxiety in patients with CHD is uncertain.

ObjectiveÌý To determine whether exercise and escitalopram are better than placebo in reducing symptoms of anxiety as measured by the Hospital Anxiety and Depression-Anxiety Subscale (HADS-A) and in improving CHD risk biomarkers.

Design, Setting, and ParticipantsÌý This randomized clinical trial was conducted between January 2016 and May 2020 in a tertiary care teaching hospital in the US and included 128 outpatients with stable CHD and a diagnosed anxiety disorder or a HADS-A score of 8 or higher who were older than 40 years, sedentary, and not currently receiving mental health treatment.

InterventionsÌý Twelve weeks of aerobic exercise 3 times per week at an intensity of 70% to 85% heart rate reserve, escitalopram (up to 20 mg per day), or placebo pill equivalent.

Main Outcomes and MeasuresÌý The primary outcome was HADS-A score. CHD biomarkers included heart rate variability, baroreflex sensitivity, and flow-mediated dilation, along with 24-hour urinary catecholamines.

ResultsÌý The study included 128 participants. The mean (SD) age was 64.6 (9.6) years, and 37 participants (29%) were women. Participants randomized to the exercise group and escitalopram group reported greater reductions in HADS-A (exercise, −4.0; 95% CI, −4.7 to −3.2; escitalopram, −5.7; 95% CI, −6.4 to −5.0) compared with those randomized to placebo (−3.5; 95% CI, −4.5 to −2.4; P = .03); participants randomized to escitalopram reported less anxiety compared with those randomized to exercise (−1.67; 95% CI, −2.68 to −0.66; P = .002). Significant postintervention group differences in 24-hour urinary catecholamines were found (exercise z score = 0.05; 95% CI, −0.2 to 0.3; escitalopram z score = −0.24; 95% CI, −0.4 to 0; placebo z score = 0.36; 95% CI, 0 to 0.7), with greater reductions in the exercise group and escitalopram group compared with the placebo group (F_1,127 = 4.93; P = .01) and greater reductions in the escitalopram group compared with the exercise group (F_1,127 = 4.37; P = .04). All groups achieved comparable but small changes in CHD biomarkers, with no differences between treatment groups.

Conclusions and RelevanceÌý Treatment of anxiety with escitalopram was safe and effective for reducing anxiety in patients with CHD. However, the beneficial effects of exercise on anxiety symptoms were less consistent. Exercise and escitalopram did not improve CHD biomarkers of risk, which should prompt further investigation of these interventions on clinical outcomes in patients with anxiety and CHD.

Trial RegistrationÌý ClinicalTrials.gov Identifier:

Anxiety disorders are the most commonly diagnosed forms of mental illness in the US and are responsible for one-third of federal government expenditures for mental illness.¹ National surveys indicate that as much as 30% of the US population will experience some kind of anxiety disorder during their lifetimes,² a figure that has increased considerably over the past 2 decades and may have been further exacerbated by the COVID-19 pandemic.³ Anxiety symptoms are associated with a number of chronic diseases,⁴ especially cardiovascular disease,^5,6 and have been associated with impaired work performance, increased use of medical services, diminished well-being, and decreased physical and social functioning.^7,8

Anxiety disorders are particularly prevalent among patients with coronary heart disease (CHD)^9-11 and are associated with greater mortality in healthy individuals^12-15 and patients with CHD.^11,16-22 Despite compelling reasons for treating anxiety in patients with CHD, few studies have examined the effects of treating anxiety in patients with cardiac disease, and no studies have examined the effects of treating anxiety on CHD biomarkers of risk, which could represent potential biological pathways to improve clinical outcomes.

Although exercise has been suggested as a viable treatment for depression,²³ including among patients with CHD,²⁴ there are few studies of exercise in individuals with anxiety. Jayakody et al²⁵ suggested that exercise may be an effective adjunctive treatment for anxiety disorders, but noted that there were too few studies to provide meaningful conclusions and, to our knowledge, no studies have examined the effects of exercise in patients with CHD and anxiety. The Understanding the Benefits of Exercise and Escitalopram in Anxious Patients With Coronary Heart Disease (UNWIND) trial²⁶ was designed to examine the effects of escitalopram and aerobic exercise compared with placebo on anxiety and CHD biomarkers of risk in patients with stable CHD and elevated symptoms of anxiety.

UNWIND was a single-site, parallel-group randomized clinical trial designed to evaluate the effects of aerobic exercise, escitalopram, or placebo (in a 2:2:1 ratio) on the Hospital Anxiety and Depression-Anxiety Subscale (HADS-A),²⁷ the primary outcome, as well as CHD risk biomarkers among individuals with CHD and elevated anxiety. Enrollment began in January 2016 and ended in February 2020. Participants were recruited through a variety of sources, including institutional review board–approved print and television advertising, self-referral, and referrals from physicians and other health care professionals. The trial was approved by the institutional review board at Duke University Medical Center, and the trial protocol can be found in Supplement 1. Written informed consent was obtained from all participants. An independent data and safety monitoring board oversaw the conduct of the study and reviewed the safety and efficacy data. The study is registered on ClinicalTrials.gov (Identifier: ). This study followed the Consolidated Standards of Reporting Trials () reporting guideline.

Men and women aged 40 years or older with documented CHD and an anxiety symptom severity score of 8 or higher on the HADS-A²⁷ and/or a DSM-5 primary diagnosis of an anxiety disorder were eligible. Race and ethnicity were determined by self-report. Collection of these data are required by the National Institutes of Health and are important to determine the generalizability of the findings. The options for self-selection match the enrollment report required by the National Institutes of Health and the institutional review board of Duke University, Durham, North Carolina, and are presented in multiple choice format, with the option to select multiple answers or to write in an answer. Exclusion criteria included a primary psychiatric diagnosis other than an anxiety disorder, medications or medical comorbidities that would preclude safe participation, current mental health treatment, and participation in regular exercise (more than 1 day per week).

Participants completed modules from the Structured Clinical Interview for DSM-5 Disorders²⁸ and the 14-item Hamilton Anxiety Rating Scale.²⁷ The Spielberger State-Trait Anxiety Inventory-State (STAI-State)²⁹ was also administered weekly to assess the ongoing treatment response of state anxiety during the 12-week intervention. In addition, several supplemental trait anxiety assessments were performed, including the Anxiety Sensitivity Index,³⁰ the General Anxiety Disorder 7-item questionnaire,³¹ and the STAI-Trait.²⁹

Because symptoms of depression often coexist with anxiety, depressive symptoms were assessed by the Beck Depression Inventory-II³² and the HADS-Depression Subscale (HADS-D).²⁷

For the assessment of aerobic capacity, patients exercised on a treadmill to exhaustion or other standard endpoints.³³ Expired gases were analyzed using a Parvo Medics TrueOne 2400 measurement system (Parvo Medics).

Urinary catecholamines, an index of sympathetic nervous system activity, including urinary concentrations of epinephrine and norepinephrine, were determined by high-pressure liquid chromatography with electrochemical detection (LabCorp).

Measurements of heart rate variability,^34,35 baroreflex sensitivity,³⁶ and endothelial function assessed by flow-mediated dilation^37,38 were performed before and after the 12-week interventions (eMethods 1 in Supplement 2).

Participants exercised 3 times per week under supervision at a local state-certified cardiac rehabilitation facility. Each exercise session consisted of 10 minutes of warm-up exercises followed by 35 minutes of continuous walking, biking, or jogging at 70% to 85% heart rate reserve and 5 minutes of cool-down exercises.

Participants started taking either 5 mg of escitalopram or a placebo equivalent daily. Daily doses were titrated to 10 mg at week 2 and 15 mg or placebo equivalent at week 3 if there was no change or only minimal improvement in anxiety. At week 4, if there was no change or only minimal improvement in anxiety, and no or minimal adverse effects, a maximum daily dose of 20 mg or placebo equivalent was prescribed. Participants met with the treating psychiatrist at weeks 1, 2, 4, 8, and 12. The treating psychiatrist was blinded to pill condition and used supportive measures to help manage adverse effects associated with the medication.

Participants were randomly assigned to exercise, escitalopram, or placebo in a 2:2:1 ratio. Randomization was stratified by sex (male or female), history of myocardial infarction (yes or no), diagnosis of an anxiety disorder (yes or no), and age (aged 40 to 59 years or 60 years and older). Participants received their group assignments in sealed envelopes. Outcome assessors were unaware of patients’ treatment assignments.

All analyses were carried out using SAS 9.4 (SAS Institute Inc). Analyses of treatment group differences were carried out using general linear models, with group contrasts as factors and ethnicity, age, sex, history of myocardial infarction, presence of a diagnosed anxiety disorder, and the pretreatment measure of the outcome variable as the adjustment covariables. For the primary outcome, HADS-A, the model included posttreatment HADS-A score as the response, with the predictors including 2 preplanned contrast variables representing (1) the 2 active treatments (exercise and escitalopram) vs placebo and (2) exercise vs escitalopram. To assess weekly changes in anxiety symptoms, STAI-S was analyzed using a repeated measures mixed model (SAS Proc Mixed) with the same covariates as above. Time in weeks was also included within this model, participants were modeled as a random variable, and a first-order autoregressive error structure was prespecified. A priori interactions between treatment, time, and baseline anxiety severity within the repeated measures analyses were also examined. To control for multiple comparisons, we used a unit-weighted composite score in which all secondary anxiety subtests were combined into a z score at both pretreatment and posttreatment for secondary analyses of ancillary anxiety measures. The intent-to-treat principle was followed in all models, with missing data handled using multiple imputation through SAS (PROC MI) (eMethods 2 in Supplement 2).

Power and sample size were estimated under the following assumptions: an α of .05, a linear model with age, sex, ethnicity, history of myocardial infarction, presence of an anxiety disorder, and baseline HADS-A score as covariates; a conservative estimate of the R² of 0.20 for the full model predicting posttreatment HADS-A; a 15% attrition rate; and 2 planned contrasts, namely active treatments vs placebo and exercise vs escitalopram. The primary effect of interest was HADS-A scores for the active treatments vs placebo. A sample size of 150 (127 after attrition) was estimated to yield 0.80 power to detect at least a 0.45 SD difference between the active treatments and placebo. All P values were 2-tailed, and statistical significance was set at a P value less than .05.

The 128 participants in the 3 groups in this study had similar demographic and clinical characteristics at baseline (Table). The mean (SD) age of participants was 64.6 (9.6) years, and 37 participants (29%) were women. The mean (SD) baseline Hamilton Anxiety Rating Scale score for the sample was 15.9 (6.4), suggesting moderate to severe baseline levels of anxiety.

The CONSORT diagram for study participants is shown in Figure 1. Of a total of 278 patients who were screened, 128 met the eligibility criteria; 52 individuals were randomized to aerobic exercise, 53 to escitalopram, and 23 to placebo. During the trial, 5 participants dropped out: 2 from the exercise group, 1 from the escitalopram group, and 2 from the placebo group, leaving 123 participants with complete data. Because of restrictions for in-person visits during the COVID-19 pandemic, 7 participants were unable to complete portions of their posttreatment assessments involving in-person laboratory measures (eg, exercise testing, flow-mediated dilation, baroreflex sensitivity, and heart rate variability assessments). Using intention-to-treat analyses, all 128 participants were retained in the final analyses (eResults in Supplement 2).

Treatment adherence was excellent across all conditions. Participants in the exercise condition attended a median (interquartile range) of 35 (2) scheduled sessions. Participants in the exercise group were at or above their prescribed heart rate training range a mean 80.4% of the time.

Among participants assigned to the pill conditions, pill counts indicated excellent adherence in both the escitalopram condition (median, 100%; interquartile range, 1.2%) and placebo condition (median, 100%; interquartile range, 1.1%) conditions. The mean (SD) daily dose for participants in the escitalopram condition was 14 mg (ie, 2.9 [1.0] pills), while the mean (SD) number of pills taken per day in the placebo condition was somewhat higher (3.5 [0.9] pills).

Participants in the exercise condition achieved greater improvements in treadmill exercise duration (2.0%; 95% CI, 0.2-3.8; P < .001) and peak maximum oxygen uptake (7.5%; 95% CI, 2.7-12.0; P = .01) compared with those in the escitalopram condition. After 12 weeks, participants in the exercise group exercised more than 1 minute longer (mean, 8.5 minutes; 95% CI, 8.2-8.8) compared with escitalopram (mean, 7.4 minutes; 95% CI, 7.1-7.7) and placebo (mean, 7.6 minutes; 95% CI, 7.1-8.0). Participants in the exercise group increased their peak maximum oxygen uptake by 6.5% (95% CI, 3.2-9.9), while participants in both the escitalopram and placebo groups showed modest declines (escitalopram, −0.6%; 95% CI, −3.7 to 2.6; placebo, −2.9%; 95% CI, −8.0 to 2.1) (eTable 1 in Supplement 2).

Adverse effects associated with each treatment were examined by patient ratings on a 36-item symptom checklist (eg, nausea, muscle soreness, and insomnia) before and after treatment. Few patients reported a worsening of symptoms after treatment and adverse effects were not different by treatment group. No participants reported experiencing an adverse event during the trial.

Examination of HADS-A scores after 12 weeks revealed that all groups showed reduced levels of anxiety following treatment (−4.57; 95% CI, −5.22 to −3.92; P < .001), with mean reductions of −4.0 (95% CI, −4.7 to −3.2) in the exercise group, −5.7 (95% CI, −6.4 to −5.0) in the escitalopram group, and −3.5 (95% CI, −4.5 to −2.4) in the placebo group. Planned contrasts were lower in the active treatment groups than in the placebo group (F_1,127 = 5.06; P = .03) and HADS-A scores for the escitalopram group were lower compared with the exercise group (−1.67; 95% CI, −2.68 to −0.66; P = .002) (Figure 2). Examination of adjusted posttreatment HADS-A scores revealed the lowest mean scores were in the escitalopram group (3.4; 95% CI, 2.7-4.1), while mean scores in the exercise group were intermediate (5.2; 95% CI, 4.5-6.0) and those in the placebo group were highest (5.7; 95% CI, 4.7-6.8). A post hoc comparison of participants in the escitalopram and placebo groups revealed that escitalopram had greater reductions in HADS-A scores compared with placebo (F_1,127 = 10.82; P = .003), while the exercise and placebo groups were not different (F_1,127 = 0.25; P = .69).

Examination of weekly changes in STAI-S scores demonstrated that both active treatment groups exhibited lower anxiety scores relative to placebo controls (F_1,126 = 7.15; P = .01) (Figure 3). After 12 weeks, participants in both the exercise and escitalopram conditions had comparable levels of state anxiety (exercise, 30.3; 95% CI, 27.7-32.9; escitalopram, 30.5; 95% CI, 28.0-33.0; P = .95); the escitalopram group experienced more rapid improvements in anxiety symptoms early in the treatment course, however, while reductions in the exercise group occurred primarily in the latter 6 weeks of treatment (F_2,1374 = 5.76; P = .003 for treatment group × time interaction) (eTable 2 in Supplement 2).

Secondary trait anxiety measures were used to provide confirmatory support for the HADS-A results. All 3 groups generally improved from baseline (F_1,126 = 128; P < .001). However, the overall composite revealed that escitalopram had greater reductions in HADS-A scores compared with exercise (F_1,127 = 5.62; P = .02). Following treatment, anxiety was consistently lower in the escitalopram group compared with the exercise and placebo groups, while the results for the exercise group were not different from those for the placebo group (eTable 3 in Supplement 2).

All groups showed significant reductions in depressive symptoms following treatment (F_1,126 = 147; P < .001), with mean Beck Depression Inventory-II reductions of −7.2 (95% CI, −8.7 to −5.7) in the exercise group, −9.8 (95% CI, −11.2 to −8.3) in the escitalopram group, and −7.2 (95% CI, −9.5 to −5.0) in the placebo group. Participants in the escitalopram group achieved larger reductions compared with those in the exercise group on both the Beck Depression Inventory-II (−0.40; 95% CI, −1.22 to −0.43; P = .04) and the HADS-D (−2.42; 95% CI, −4.56 to −0.28; P = .02) (eTable 4 in Supplement 2).

Exploratory analysis of changes in aerobic fitness and changes in depressive symptoms revealed that greater aerobic fitness improvements were associated with greater reductions in depressive symptoms (r = −0.20; P = .03). However, there was no association between improved maximum oxygen uptake and improvements in anxiety (r = 0.02; P = .834).

Analyses revealed significant postintervention group differences in urinary catecholamines following treatment (composite 24-hour catecholamines [z score] for exercise, 0.05; 95% CI, −0.2 to 0.3; escitalopram, −0.24; 95% CI, −0.4 to 0; placebo, 0.36; 95% CI, 0 to 0.7), with exercise and escitalopram, demonstrating greater reductions in catecholamines compared with placebo (F_1,127 = 4.93; P = .01) and escitalopram demonstrating greater reductions compared with exercise (F_1,127 = 4.37; P = .04) (eTable 5 in Supplement 2).

There were no consistent posttreatment differences in 24-hour heart rate variability, baroreflex sensitivity, or flow-mediated dilation across treatment groups (eTable 5 in Supplement 2).

In this 12-week randomized clinical trial of the treatment of anxiety in patients with stable CHD and high anxiety, escitalopram resulted in reduced levels of anxiety compared with placebo controls. Because a 2-point improvement or 20% reduction in HADS-A scores is considered clinically meaningful,^39,40 the 6.9-point reduction in HADS-A scores representing a 44% decline in anxiety symptoms in this study is considered both statistically and clinically significant. Importantly, escitalopram was superior to exercise in reducing anxiety, lowering epinephrine levels, and diminishing symptoms of depression. Although exercise achieved comparable reductions in state anxiety relative to escitalopram after 12 weeks and greater reductions compared with placebo controls, exercise did not result in lower anxiety scores compared with placebo on the HADS-A or on any of the supplemental trait anxiety measures. There is no obvious explanation for the greater improvement in state anxiety but not trait anxiety in the exercise condition compared with placebo, but the significant improvement in HADS-A scores among participants in the placebo group likely minimized between-group differences.

To our knowledge, UNWIND is the first randomized clinical trial to evaluate the efficacy of a selective serotonin reuptake inhibitor or aerobic exercise in the treatment of anxiety in patients with CHD and high levels of anxiety. Adherence in the trial was excellent and there were no study-related adverse events and few treatment-related adverse effects. Although we were aware of the potential risk of patients developing corrected QT prolongation⁴¹ and excluded potential participants with evidence of corrected QT prolongation, no patients developed corrected QT prolongation during the trial.

Selective serotonin reuptake inhibitors have been studied in the treatment of depression in patients with cardiac disease, with mixed results. More than 2 decades ago, results from studies of paroxetine⁴² and fluoxetine⁴³ were encouraging. However, subsequent studies of sertraline were less consistent. In the Sertraline Antidepressant Heart Attack Randomized Trial (SADHART), sertraline proved superior to placebo, but only in a subgroup of more severely depressed patients.⁴⁴ Other studies provided equivocal results. Mirtazapine, an atypical antidepressant, did not improve depression in the Myocardial Infarction and Depression-Intervention Trial (MIND-IT),⁴⁵ while citalopram was superior to placebo in treating depression in the Canadian Cardiac Randomized Evaluation of Antidepressant and Psychotherapy Efficacy (CREATE) trial.⁴⁶ However, citalopram was combined with counseling sessions, so the benefits of the medication alone could not be determined.

Escitalopram has been shown to be effective in reducing anxiety and depression in the general population.⁴⁷ In the MOOD-HF trial, a study of patients with heart failure, Angermann et al³² reported that 18 months of treatment with escitalopram did not significantly reduce all-cause mortality or hospitalization compared with placebo, and there was no significant improvement in depression. In contrast, escitalopram was found to be safe and effective in preventing depression in nondepressed patients with acute coronary syndrome.⁴⁸ In a subsequent study, the Escitalopram for Depression in Acute Coronary Syndrome (EsDEPACS) trial⁴⁹ found that 24 weeks of treatment with escitalopram was effective in reducing depressive symptoms and in lowering the risk of major adverse events compared with placebo controls.⁵⁰ Moreover, in a secondary analysis⁵¹ of the EsDEPACS trial, escitalopram was shown to reduce anxiety symptoms compared with placebo, although 27% of the original sample were lost to follow-up and were not included in the analyses.

In the present study, escitalopram was found to be a safe and effective anxiolytic treatment for patients with CHD and high anxiety. Unlike previous studies that targeted patients with CHD and major depression, the UNWIND trial targeted patients with anxiety, specifically excluding patients with a primary diagnosis of depression. Almost three-fourths of the sample had a diagnosed anxiety disorder, and baseline scores on the HADS-A were in the moderate to severe range. Escitalopram produced clinically meaningful reductions in anxiety, as well as significant reductions in depression. Escitalopram also reduced urinary catecholamine levels, indicative of a reduction in sympathetic nervous system activity. Previous studies have shown that anxiety is directly related to urinary catecholamine excretion,^52,53 suggesting that escitalopram also had a favorable effect on the physiological effects of anxiety that have been implicated in the development and exacerbation of cardiovascular disease.⁵⁴ However, escitalopram did not improve such CHD biomarkers as heart rate variability, flow-mediated dilation, or baroreflex sensitivity. It is possible that 12 weeks was insufficient to produce significant improvements in these CHD biomarkers. It is also possible that these biomarkers were not in the direct pathway that was responsible for the clinical benefits of escitalopram, such as those observed in the EsDEPACS trial.⁵⁰

Surprisingly, the effects of exercise in reducing trait anxiety were no better than placebo controls and also provided no advantage compared with placebo in reducing symptoms of depression. Previous research has shown that acute bouts of exercise reduced state anxiety and that the magnitude of this effect increased with exercise training over 9 weeks.⁵⁵ Although exercise achieved similar reductions as escitalopram on the STAI-S at the end of 12 weeks of treatment, exercise was no better than placebo on the HADS-A, the primary outcome in this trial, nor on any of the ancillary anxiety measures. However, the exercise intervention did result in some evidence of lowered 24-hour urinary catecholamine excretion, indicating reduced sympathetic nervous system activity, potentially reflecting a reduction in this physiological manifestation of anxiety. Exercise did not improve depression more than placebo controls, an unexpected finding given that exercise is generally considered effective in treating depression.²³ We recently reported that the presence of comorbid anxiety may blunt the effects of exercise on depression,⁵⁶ and because our sample consisted of individuals with moderate to severe levels of anxiety, including many with a diagnosed anxiety disorder, the benefits of exercise on depression may have been diminished.

This study had limitations. This was a single-site study in which the sample was relatively small and treatment duration was relatively short. While the EsDEPACS study⁴⁹ was 24 weeks, the mean dosage of escitalopram in the EsDEPACS trial was only 7.6 mg day compared with 14 mg day in the UNWIND trial. Further, many trials of selective serotonin reuptake inhibitors are 12 weeks or less.⁵⁷ The 12-week treatment period may have been insufficient to observe significant changes in the CHD biomarkers. Interpretation of the results of this trial may be limited by the exclusion of comorbid primary psychiatric diagnoses, particularly major depressive disorder, which frequently occurs in patients with anxiety disorders.⁵⁸ Additionally, participants in the pill condition had their dosages adjusted by a psychiatrist based on their treatment response, while those in the exercise condition received a standard exercise dose typically offered by cardiac rehabilitation programs. Adjusting the exercise dose by modifying the frequency, duration, or intensity of the exercise prescription may have improved the outcome for participants in the exercise group.

In this randomized clinical trial, among a sample of 128 patients with stable CHD and high levels of anxiety, a 12-week course of escitalopram resulted in greater reductions in anxiety compared with placebo controls. However, exercise generally produced little advantage over placebo in reducing trait anxiety, and escitalopram was superior to exercise in reducing both anxiety and depressive symptoms. There were no treatment-related differences in CHD biomarkers, so the clinical significance of these findings regarding potential cardiovascular benefits is unknown.

Accepted for Publication: June 9, 2021.

Published Online: August 18, 2021. doi:10.1001/jamapsychiatry.2021.2236

Corresponding Author: James A. Blumenthal, PhD, Department of Psychiatry and Behavioral Sciences, Duke University Medical Center, 4567 Duke Hospital South, 40 Duke Medicine Cir, Durham, NC 27710 (james.blumenthal@duke.edu).

Author Contributions: Drs Blumenthal and Smith had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Blumenthal, Jiang, Hinderliter, Davidson, Sherwood.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Blumenthal, Smith, Jiang, Hoffman, Davidson.

Critical revision of the manuscript for important intellectual content: Blumenthal, Jiang, Hinderliter, Watkins, Kraus, Davidson, Liao, Sherwood.

Statistical analysis: Smith.

Obtained funding: Blumenthal, Hinderliter, Watkins, Sherwood.

Administrative, technical, or material support: Blumenthal, Jiang, Hinderliter, Hoffman, Liao, Sherwood.

Supervision: Blumenthal, Hinderliter, Sherwood.

Conflict of Interest Disclosures: Drs Blumenthal, Smith, Jiang, Hinderliter, Watkins, Hoffman, Kraus, Liao, and Sherwood report grants from the National Heart, Lung, and Blood Institute during the conduct of the study. Dr Davidson reports royalties for self-rating measures for resilience, posttraumatic stress disorder, and anxiety. No other disclosures were reported.

Funding/Support: This project was supported by grant HL125522 from the National Heart, Lung, and Blood Institute.

Role of the Funder/Sponsor: The funder had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Data Sharing Statement: See Supplement 3.

Additional Contributions: The authors thank the members of the data and safety monitoring board: Diane Catellier, PhD (Research Triangle Institute, Raleigh, North Carolina), Leo Pozuelo, MD (Cleveland Clinic, Cleveland, Ohio), and David Sheps, MD (University of Florida, Gainesville). We also thank Dwayne Callwood, MD (Kernodle Clinic, Mebane, North Carolina), Mohan Chilukuri, MD, Paula Miller, MD (University of North Carolina Hospitals, Chapel Hill), and J. Stewart Jones, DO (Duke Triangle Heart Institute, Durham, North Carolina), for their assistance in patient recruitment, and Brian Kincaid, MD (Duke University School of Medicine, Durham, North Carolina), for his assistance in the psychiatric treatment of the participants. We also thank our research staff: Bryan Feger, PhD, Hannah Malian, BA, Kate Ammerman, BA, Willis Wong, MS, Natalie Hamilton, BA, Catherine Wu, MS, Michael Ellis, RDMAS, RVT, Jeanne Schwartz, PA-C, Stephanie Mabe, MS (Duke University School of Medicine), Beth Drossman, MS (WakeMed, Raleigh, North Carolina), and Rahul Gupta, BA (University of North Carolina, Chapel Hill). Drs Catellier, Pozuelo, Sheps, Callwood, Miller, and Feger; Mss Malian, Ammerman, Hamilton, Wu, Schwartz, and Mabe; and Messrs Wong and Ellis were compensated for their contributions.