Journal of Psychosomatic Research 75 (2013) 409–413

Contents lists available at ScienceDirect

Journal of Psychosomatic Research

Patient Health Questionnaire-9 score and adverse cardiac outcomesin patients hospitalized for acute cardiac disease☆

Scott R. Beach a,b, James L. Januzzi a,c, Carol A. Mastromauro b, Brian C. Healy d, Eleanor E. Beale b,Christopher M. Celano a,b, Jeff C. Huffman a,b,⁎a Harvard Medical School, Boston, MA, USAb Department of Psychiatry, Massachusetts General Hospital, Boston, MA, USAc Department of Medicine, Division of Cardiology, Massachusetts General Hospital, Boston, MA, USAd Department of Biostatistics, Massachusetts General Hospital, Boston, MA, USA

☆ Research conducted inDepartments of Psychiatry anGeneral Hospital.⁎ Corresponding author at: Massachusetts General H

1220D, Boston, MA 02114, USA. Tel.: +1 617 724 2910; fE-mail address: jhuffman@partners.org (J.C. Huffman)

0022-3999/$ – see front matter © 2013 Elsevier Inc. All rihttp://dx.doi.org/10.1016/j.jpsychores.2013.08.001

a b s t r a c t

a r t i c l e i n f o

Article history:

Received 6 June 2013Received in revised form 1 August 2013Accepted 2 August 2013

Keywords:Cardiovascular diseaseDepressionPatient Health Questionnaire-9

Objective: The Patient Health Questionnaire-9 (PHQ-9) is increasingly used as a depression assessment tool incardiac patients. However, in contrast to older depression instruments, there is little data linking PHQ-9 scoresto adverse cardiac outcomes. Our goal was to evaluate whether higher PHQ-9 scores were predictive of subse-quent cardiac readmissions among depressed patients hospitalized for an acute cardiac event.Methods: Patients diagnosed with depression during hospitalization for acute coronary syndrome, heart failure,or arrhythmia were enrolled in a randomized depression management trial. Participants were administeredPHQ-9 at enrollment, and data was collected regarding cardiac readmissions and mortality over the next6 months. To evaluate the independent association of PHQ-9 score with subsequent cardiac readmission, Cox re-

gression analysis that included relevant sociodemographic andmedical covariateswas used. Survival analysis ex-amining time to first event, stratified by quartile of initial PHQ-9 score, was performed using Kaplan–Meiercurves and log-rank test for trend. Analyses were then repeated using a composite (cardiac readmission or mor-tality) outcome.Results: Among 172 subjects, 62 (36.0%) had a cardiac-related rehospitalization. Higher initial PHQ-9 score pre-dicted cardiac-related rehospitalization, independent ofmultiple relevant covariates (hazard ratio 1.09 [95% con-fidence interval = 1.02–1.17]; p = 0.015). On survival analysis, log-rank test for trend revealed a significant risein event rates across increasing PHQ-9 quartiles (χ2 = 6.36; p = 0.012). Findings were similar (p b .05) for thecomposite outcome.Conclusion: In depressed cardiac patients, each additional point on the PHQ-9was independently associatedwitha 9% greater risk of cardiac readmission over the subsequent 6 months.

© 2013 Elsevier Inc. All rights reserved.

Introduction

Among patients hospitalized for acute cardiac disease, depressionhas been independently associated with subsequent cardiac events,readmissions, and mortality [1,2]. Previous studies of the associationbetween depression and cardiac outcomes have typically used diagnos-tic interviews or the 21-item Beck Depression Inventory (BDI) [3] tomeasure depression; these measures can take substantial time to admin-ister and, in the case of the BDI, do not match the current conceptualiza-tion (e.g., duration of symptoms) of major depressive disorder (MDD).A more recently developed depression measure, the Patient HealthQuestionnaire-9 (PHQ-9) [4], is used increasingly in medical settings.

d Cardiology at Massachusetts

ospital, 55 Fruit Street/Warrenax: +1 617 724 9115..

ghts reserved.

This measure contains only nine items, is simple and rapid to adminis-ter, and has been found to have very good operating characteristics forMDD among patients with cardiac illness [5–7]. Given these propertiesof the PHQ-9, the AmericanHeart Association (AHA) recommends eval-uating all cardiac patients for depression using a two-step approach thatuses the first two items of the PHQ-9 (the Patient Health Questionnaire-2 [PHQ-2 [8]]) as a screen, followed by further assessment with the fullPHQ-9 for positive-screenpatients [9,10]. These recommendations havebeen controversial, given data that systematic depression screening inisolation has not been linked to better cardiac (or patient) outcomesand is costly with respect to time and resources [11–13].

Despite its increasing use, there isminimal data linking scores on thePHQ-9 to adverse cardiac outcomes; indeed,we are aware of only a singlestudy (the Heart and Soul study of patients with stable coronary arterydisease) connecting elevated PHQ-9 scores to adverse cardiac outcomes[14]. Furthermore, to our knowledge there has been no study in patientswith recent acute cardiac illness (e.g., post-myocardial infarction) to de-termine whether there is an association of PHQ-9 scores with adverse

410 S.R. Beach et al. / Journal of Psychosomatic Research 75 (2013) 409–413

cardiac events. Given that patients with acute cardiac illness represent avery high-risk cohort for whom depression is clearly and independentlylinkedwith poor prognosis [1,15], the lack of support for a connection be-tween this recommended assessment tool for depression and adverseoutcomes (e.g., readmissions for a cardiac cause) is an important gap inthe literature.

Accordingly, we conducted a secondary analysis of a collaborativecare depressionmanagement trial in patients hospitalized for acute car-diac illness to determine whether depression severity, as measured byPHQ-9 score, was associated with subsequent cardiac rehospitalization.

Methods

This is a secondary analysis of a randomized trial comparing a low-intensity, 12-week collaborative care depressionmanagement programto treatment as usual; methods and results have been reported in detailelsewhere [16,17]. In short, patients admitted to cardiac units for acutecoronary syndrome (ACS), heart failure (HF), or arrhythmia were diag-nosedwith depression using the PHQ-9, and received either collaborativecare or enhanced usual care for 12 weeks. Subjects in the collaborativecare arm had significantly greater improvements in mental healthoutcomes at 6 and 12 weeks, but not 6 months, and there were nobetween-group differences in rates of cardiac readmissions over the6-month period [16].

For the primary trial, the PHQ-9 was administered to patients in thehospital who had responded affirmatively to initial depression screen-ing. The PHQ-9 inquires about the frequency of nine depressive symp-toms over the prior two weeks, arriving at a total score ranging from 0to 27. A cutoff score (typically 10) is used to identify depression inmost settings [4,14,18]; however, in this trial we required patients tohave depressed mood or anhedonia more than half the days, and atleast five of the nine symptoms more than half the days, to meetstudy criteria for depression. These more stringent depression criteriaallowed us to more carefully avoid diagnosing depression in patientswho had only somatic symptoms in the context of their medical event,and they more closely match formal Diagnostic and Statistical Manual,fifth edition [19] criteria for MDD.

As noted, in addition to including patients admitted with ACS andHF, we also included patients admitted for arrhythmia. Cardiac arrhyth-mias are common and important clinical conditions, with over twomil-lion Americans suffering from atrialfibrillation alone [20], and it is likelythat broad depression management programs in cardiac centers wouldinclude arrhythmia patients given the prevalence of such patients in in-patient and outpatient settings.

Furthermore, depression is common in patients with arrhythmias,with rates of clinically significant depression of 10–38% in atrial fibrilla-tion [21] and 11–28% in patients who have had implantable defibrillators(ICDs) placed for ventricular arrhythmias [22]. Depression in thispopulation has also been associated with adverse outcomes. For ex-ample, depression has been independently linked to recurrence of atrialfibrillation following cardioversion [23], linked to cardiac mortality in pa-tients with comorbid atrial fibrillation and HF [24], and associated withmortality, independent of relevant covariates, in patientswith an ICD [25].

For patients who met depression criteria and enrolled in the study,baseline sociodemographic characteristics and medical variables werecollected by interview and chart review. Participants then receivedeither a collaborative care intervention or enhanced usual care for12 weeks and had periodic outcome assessments over 6 months. Cardi-ac rehospitalizations during the 6-month study period were identifiedfrom multiple sources, including patient reports of readmission as partof follow-up assessments every 6 weeks; structured inquiry of patients'primary medical providers regarding number, location, and cause ofreadmissions; and review of the healthcare system's electronic medicalrecord spanning seven acute care hospitals. Primary reason for admis-sion was determined via record review, with as-needed adjudication(regarding cardiac vs. non-cardiac cause) by the study cardiologist

(J.J.), who was blinded to study group assignment. Readmissions thatwere planned at the time of initial discharge (e.g., planned readmissionin two weeks for pacemaker implantation) were excluded from theanalysis. A secondary, composite outcome of cardiac readmissions andall-cause mortality was also created; mortality data was obtained inthe same manner as for rehospitalizations above.

Covariates selected a priori for inclusion in multivariate analysiswere group assignment (enhanced usual care or collaborative care),age, sex, admission smoking status, diabetes mellitus, hyperlipidemia,hypertension, baseline cardiac symptom score (using a 10-symptomlist adapted from the WISE study [26]), and admission diagnosis ofHF; these covariates were chosen given prior links between these vari-ables and adverse cardiac outcomes.

Specifically, study group assignment was chosen because it was thefocus of the overarching study and because it has been linked to cardiacoutcomes in prior collaborative care studies [10,27]. Agewas chosen be-cause it has been shown to predict rehospitalizations in patients withHF, ACS, and arrhythmias [25,28–32]. Sex was chosen because studieshave shown this variable to be linked with readmissions or mortality,independent of other covariates, in patients with our diagnoses [31–34].Current smoking has predicted rehospitalization andmorbidity in severalcardiac populations [35–37]. Diabetes [31,38–40], hyperlipidemia [37,41]and hypertension [41–43] have similarly been linked to readmissions andother adverse outcomes inpatientswith cardiac disease.Wealso includedcardiac symptoms at baseline as a surrogate measure of disease severity,given the connection between baseline illness severity and readmissionin ACS and HF patients [41,44], and links between baseline symptomsand subsequent readmissions in atrial fibrillation [28].

Statistical analysis

Descriptive statistics were used to calculate and present data(e.g., means, proportions) on baseline sociodemographic, medical, andoutcome variables.Mean initial PHQ-9 scoreswere calculated for patientswith and without subsequent cardiac readmissions, and these meanswere compared between groups using an independent-samples t test.These calculations were repeated for the secondary measure (compositescore of cardiac readmissions and mortality).

Multivariate Cox proportional hazards regression analysis wasperformed to evaluate the relationship between initial PHQ-9 scoresand time to clinical events. This analysis included all covariates listedabove, and we used a null value (e.g. no hypertension) as the refer-ence value for dichotomous variables and lower values as the refer-ence value for continuous variables (e.g., age). Therefore, a positivecoefficient for the variables hypertension and age would mean thatpresence of hypertension and increasing age is associated withhigher likelihood of the outcome.

The PHQ-9 scores were then divided into quartiles, and Kaplan–Meiersurvival curves were constructed to examine the incidence of eventsamong the four quartiles. Log rank tests for trend were performed to as-sess for a statistically significant trend in terms of time to adverse out-comes across the ordered quartiles.

These multivariate and survival analyses were first performed withcardiac readmissions as the dependent variable, and then repeatedusing the secondary, composite measure (cardiac readmissions andmor-tality). Finally, we also calculated the predicted survival curves for cardiacreadmission for each of the PHQ-9 quartiles based on our Cox regressionmodel. This creates a predicted survival curve at the mean of each of thecovariates for each of the four quartile groups. Analyses were performedusing Stata (version 11.2, StataCorp, College Station, TX); all tests weretwo-tailed, and p b .05 was considered significant.

Results

Overall, 175 subjectswere enrolled in the study; three subjects diedduring admission,and post-discharge readmission and mortality data was available for the remaining 172subjects. Table 1 displays the baseline characteristics of these 172 patients. A total of 62

Table 1Baseline sociodemographic and clinical characteristicsa

Characteristic

Demographic/psychosocial characteristicsAge, mean (SD) 62.1 (12.4)Male gender 89 (51.7)Married 74 (43.0)Caucasian race 157 (91.3)Living alone 57 (33.1)Employed 42 (24.4)

Medical historyHypertension 100 (58.1)Diabetes mellitus 59 (34.3)Hyperlipidemia 102 (59.3)Current smoking 35 (20.3)Admission diagnosis ACS: 81 (47.1)

CHF: 55 (31.9)Arrhythmia: 36 (20.9)

Medications at dischargeBeta-blocker 133 (77.3)ACE inhibitor 86 (50.0)Statin 126 (73.3)Aspirin 134 (77.9)Clopidogrel 51 (29.7)Diuretic 88 (51.2)Antidepressant 120 (69.8)

Baseline symptom measuresInitial PHQ-9 score, mean (SD) 17.6 (3.5)Cardiac symptoms (#, out of 10), mean (SD) 6.1 (2.1)

Note. ACE = angiotensin-converting enzyme; ACS = acute coronary syndrome;CHF = congestive heart failure; MI = myocardial infarction; PHQ-9 = PatientHealth Questionnaire-9; SD = standard deviation; UA = unstable angina.

a Data are reported as N (%) unless otherwise specified.

0.00

0.25

0.50

0.75

1.00

0 50 100 150 200

Days

Day

s

Lowest quartile Second quartileThird quartile Top quartile

Kaplan-Meier survival estimates

Fig. 1. Rates of cardiac readmission over 6 months by quartiles of initial PHQ-9 score.

411S.R. Beach et al. / Journal of Psychosomatic Research 75 (2013) 409–413

patients (36.0%) had a readmission for a cardiac cause during the 6-month follow-up period;this rate did not differ between study group assignment (collaborative care: 32/88 = 36.3%;usual care: 30/84 = 35.7%; χ2 = 0.008; p = .93). Regarding the composite (cardiacreadmission or mortality) outcome, a total of 68 patients (39.5%) had an adverse out-come, with no difference between groups (collaborative care: 35/88 = 39.8%; usualcare: 33/84 = 39.2%; χ2 = 0.004; p = .95).

Differences inmean initial PHQ-9 scores were significantly different in thosewith andwithout subsequent cardiac readmissions (rehospitalized: 18.5 [standard deviation {SD}3.7]; no readmission: 17.0 [SD 3.3]; t = 2.71; p = 0.007). There were similar between-group differences on mean initial PHQ-9 scores for the composite (cardiac readmissionsor mortality) measure (18.4 [SD 3.7] vs. 17.0 [3.3]; t = 2.64; p = 0.009). Analysis ofPHQ-9 scores resulted in the following quartiles: 11 to 14, 15 to 16, 17 to 19, and 20 to 26.

Usingmultivariate Cox proportional hazards regression analysis (Table 2), initial PHQ-9 scorewas independently associatedwith time to cardiac readmission (hazard ratio [HR]1.09 [95% CI 1.02–1.17], p = .015). Kaplan–Meier survival curves (Fig. 1) comparing theincidence of rehospitalization among the four quartiles of initial PHQ-9 scoreswere gener-ated, with log-rank test for trend finding a significant association between initial PHQ-9quartile and time to readmission (χ2 = 6.36; df = 1; p = .012).

Secondary analyses using the composite measure (Supplementary Table 1 and Supple-mentary Fig. 1) hadhighly similar results. Initial PHQ-9 scoreswere again independently pre-dictive of the composite endpoint in Cox modeling (HR 1.09 [95% CI 1.02–1.17]; p = .013).On survival analyses, there was again significance on log-rank test for trend (χ2 = 6.06;df = 1; p = .014) using quartiles of initial PHQ-9 scores. Secondary Cox regression analysesfor both outcome variables that also included prior MI as a covariate had highly similar

Table 2Cox proportional hazards analysis examining variables associated with time to cardiacreadmission

Variable Hazard ratio SE Z p [95% CI]

Initial PHQ-9 score 1.09 0.039 2.43 0.015 1.02–1.17Treatment group 0.95 0.249 −0.18 0.86 0.57–1.59Age 1.00 0.012 0.32 0.75 0.98–1.03Male sex 1.32 0.355 1.04 0.30 0.78–2.34Cardiac symptom score 1.08 0.080 1.00 0.32 0.93–1.25Admission dx (HF vs. non-HF) 1.56 0.425 1.62 0.11 0.91–2.66Diabetes mellitus 1.05 0.292 0.17 0.87 0.61–1.81Hyperlipidemia 1.79 0.555 1.88 0.060 0.98–3.29Hypertension 0.71 0.222 −1.08 0.28 0.39–1.31Current smoking (at admission) 0.79 0.285 −0.64 0.52 0.39–1.61

ACS = acute coronary syndrome; CI = confidence interval; HF = heart failure; PHQ-9 =Patient Health Questionnaire-9; SE = standard error.

findings. Finally, predicted survival curves from the Cox regression model (SupplementaryFigs. 2 and 3) were highly similar to the unadjusted Kaplan–Meier curves, with stepwise in-creased risk of adverse outcome with each higher PHQ-9 quartile.

Discussion

In this analysis of depressed cardiac patients, higher initialPHQ-9 scores were significantly associated with subsequent cardiacrehospitalization at 6 months, independent of multiple relevant covar-iates. Initial PHQ-9 scores were similarly associated with an increasedrisk of a composite outcome of cardiac readmission or mortality overthis period. Indeed, PHQ-9 score wasmore strongly associated with ad-verse cardiac outcomes than many other variables previously found tobe predictive of these outcomes, including age, gender, diabetes, hyper-tension and smoking.

Of note, though increasing age was associated with cardiacreadmissions as in prior studies [25,31], this association was not statisti-cally significant in our cohort. One possible explanation could be thatsome of the potential effects of increased age (e.g., decreased mobility/activity) that could adversely affect cardiac prognosis may overlap withthe effects of more severe depression; another possibility is Type II errorand that in a larger sample this finding would have reached significance.

Though consistent with prior studies using other depression scales[1,2,45], to our knowledge, this is the first study to link PHQ-9 scorewith an adverse cardiac outcome in patients hospitalized for acute car-diac disease, an important finding given the increasing use of the PHQ-9in the assessment of cardiac patients. Given that depression has beenclearly and independently associated with adverse cardiac outcomes,the specific instrument used to assess depression should also be linkedto such outcomes.

These results do confirm the findings in the Heart and Soul studythat appears to be the only prior work examining the prognostic valueof PHQ-9 scores in any population of cardiac patients [14]. In that studyof outpatients with stable coronary artery disease, a PHQ-9 score of 10or greater (after a positive PHQ-2 screen) was associated with increasedrisk of subsequent cardiovascular events. Our study confirms the findingsof this prior study, and does so in a broader and higher-risk hospitalizedcardiac population.

There has been specific comment in expert reviews that additionalstudies of the prognostic value of PHQ-9 in cardiac patients are requiredgiven the paucity of data about this instrument's association with out-comes [46]. Furthermore, the prior lack of data regarding the PHQ-9and cardiac outcomes has specifically led to avoidance of its use in a re-cent major intervention trial for depression in cardiac patients [47], andour results address these gaps. Importantly, our findings do not imply

412 S.R. Beach et al. / Journal of Psychosomatic Research 75 (2013) 409–413

that all patients should be screened for depression using the PHQ-9. Thefactors that need to be considered in order to determine if screening issupported by evidence go far beyond establishing that there is a link be-tween depression (in this case measured by the PHQ-9) and healthoutcomes.

The strengths of this work are that it used prospective data from pa-tients with a wide range of cardiovascular illnesses, rather than a singlediagnosis. Furthermore, multiple overlapping methods were simulta-neously used to identify readmissions and clarify etiology, and our anal-ysis included multiple relevant covariates linked to cardiac health. Thisanalysis also had several limitations. First, as all subjectswere diagnosedwith depression, this study did not include patients in the lower rangeof PHQ-9 scores. Further, the subjects in our study had PHQ-9 scoresin the moderate to severe range, significantly limiting generalizabilityto the typical cohort of patients screened on the cardiac floors, whomay have subsyndromal or minor depressive symptoms. Additionally,we were not able to control for all possible covariates given the totalnumber of subjects and readmissions without further overfitting themodel [48], though we included multiple demographic and medicalvariables that we felt to be most likely to be linked to readmissions,including multiple measures of baseline cardiac severity (admissiondiagnosis, cardiac symptoms, and multiple cardiac risk factors). Wedid not have clinicalmeasures of HF status (e.g., rales, S3) during the ad-mission. We also did not control for antidepressant use at admission.Though not definitively shown to improve cardiac prognosis, theseagents may have the potential to influence outcomes in depressed car-diac patients [49]. A very small number of patients suffered mortality,preventing us from examining mortality alone as an outcome. Finally,the subjects were recruited from a single site at an academic medicalcenter with primarily White patients.

In summary, in a cohort of depressed cardiac patients assessed dur-ing admission and assessed for 6 months as part of an integrated caredepression trial, each point on the baseline PHQ-9 score was indepen-dently associated with a 9% greater risk of cardiac readmissions (andsimilar risk of a composite measure of readmissions and mortality).Future work should further assess these connections in both depressedand nondepressed patients, and determine whether a decrease inPHQ-9 score over time results in reduced cardiac risk.

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.jpsychores.2013.08.001.

Conflict of interest

The authors have no competing interests to report.

Acknowledgments

Funding source: American Heart Association Scientist DevelopmentGrant 0735530T (Huffman). Clinical Trial Registration: Unique Identifier:NCT00847132.

References

[1] van Melle JP, de Jonge P, Spijkerman TA, Tijssen JG, Ormel J, van Veldhuisen DJ, et al.Prognostic association of depression following myocardial infarction with mortalityand cardiovascular events: a meta-analysis. Psychosom Med 2004;66:814–22.

[2] Barth J, Schumacher M, Herrmann-Lingen C. Depression as a risk factor for mortalityin patients with coronary heart disease: a meta-analysis. Psychosom Med 2004;66:802–13.

[3] Beck AT, Ward CH, Mendelson M, Mock J, Erbaugh J. An inventory for measuring de-pression. Arch Gen Psychiatry 1961;4:561–71.

[4] Kroenke K, Spitzer RL, Williams JB. The PHQ-9: validity of a brief depression severitymeasure. J Gen Intern Med 2001;16:606–13.

[5] Stafford L, Berk M, Jackson HJ. Validity of the Hospital Anxiety and Depression Scaleand Patient Health Questionnaire-9 to screen for depression in patients with coro-nary artery disease. Gen Hosp Psychiatry 2007;29:417–24.

[6] McManus D, Pipkin SS, Whooley MA. Screening for depression in patients with cor-onary heart disease (data from the Heart and Soul Study). Am J Cardiol 2005;96:1076–81.

[7] Thombs BD, Ziegelstein RC, Whooley MA. Optimizing detection of major depressionamong patients with coronary artery disease using the patient health questionnaire:data from the Heart and Soul Study. J Gen Intern Med 2008;23:2014–7.

[8] Kroenke K, Spitzer RL, Williams JB. The Patient Health Questionnaire-2: validity of atwo-item depression screener. Med Care 2003;41:1284–92.

[9] Lichtman JH, Bigger JT, Blumenthal JA, Frasure-Smith N, Kaufmann PG, Lesperance F,et al. Depression and coronary heart disease: recommendations for screening, refer-ral, and treatment: a science advisory from the American Heart Association Preven-tion Committee of the Council on Cardiovascular Nursing, Council on ClinicalCardiology, Council on Epidemiology and Prevention, and Interdisciplinary Councilon Quality of Care and Outcomes Research: endorsed by the American PsychiatricAssociation. Circulation 2008;118:1768–75.

[10] Davidson KW, Kupfer DJ, Bigger JT, Califf RM, Carney RM, Coyne JC, et al. Assessmentand treatment of depression in patients with cardiovascular disease: National Heart,Lung, and Blood Institute Working Group Report. Psychosom Med 2006;68:645–50.

[11] Thombs BD, de Jonge P, Coyne JC, Whooley MA, Frasure-Smith N, Mitchell AJ, et al.Depression screening and patient outcomes in cardiovascular care: a systematic re-view. JAMA 2008;300:2161–71.

[12] Thombs BD, Roseman M, Coyne JC, de Jonge P, Delisle VC, Arthurs E, et al. Does ev-idence support the American Heart Association's recommendation to screen pa-tients for depression in cardiovascular care? An updated systematic review. PLoSOne 2013;8:e52654.

[13] Screening for depression in adults: U.S. preventive services task force recommenda-tion statement. Ann Intern Med 2009;151:784–92.

[14] Elderon L, Smolderen KG, Na B, Whooley MA. Accuracy and prognostic value ofAmerican Heart Association: recommended depression screening in patients withcoronary heart disease: data from the Heart and Soul Study. Circ Cardiovasc QualOutcomes 2011;4:533–40.

[15] Frasure-Smith N, Lesperance F, Talajic M. Depression and 18-month prognosis aftermyocardial infarction. Circulation 1995;91:999–1005.

[16] Huffman JC, Mastromauro CA, Sowden GL, Fricchione GL, Healy BC, Januzzi JL. Impactof a depression care management program for hospitalized cardiac patients. CircCardiovasc Qual Outcomes 2011;4:198–205.

[17] Huffman JC, Mastromauro CA, Sowden GA, Wittmann CL, Rodman R, Januzzi JL. Acollaborative care depression management program for cardiac inpatients: depres-sion characteristics and in-hospital outcomes. Psychosomatics 2011;52:26–33.

[18] KatonWJ, Lin EH, Von Korff M, Ciechanowski P, Ludman EJ, Young B, et al. Collabora-tive care for patients with depression and chronic illnesses. N Engl J Med 2010;363:2611–20.

[19] American Psychiatric Association, American Psychiatric Association. Task Force onDSM-IV. Diagnostic and statistical manual of mental disorders: DSM-IV-TR. 4thed.Washington, DC: American Psychiatric Association; 2000 [943 pp.].

[20] Roger VL, Go AS, Lloyd-Jones DM, Adams RJ, Berry JD, Brown TM, et al. Heart diseaseand stroke statistics—2011 update: a report from the American Heart Association.Circulation 2011;123:e18-209.

[21] Patel D, Mc Conkey ND, Sohaney R, Mc Neil A, Jedrzejczyk A, Armaganijan L. A sys-tematic review of depression and anxiety in patients with atrial fibrillation: themind–heart link. Cardiovasc Psychiatry Neurol 2013;2013:159850.

[22] Magyar-Russell G, Thombs BD, Cai JX, Baveja T, Kuhl EA, Singh PP, et al. Theprevalenceof anxiety and depression in adults with implantable cardioverter defibrillators: a sys-tematic review. J Psychosom Res 2011;71:223–31.

[23] Lange HW, Herrmann-Lingen C. Depressive symptoms predict recurrence of atrial fi-brillation after cardioversion. J Psychosom Res 2007;63:509–13.

[24] Frasure-Smith N, Lesperance F, Habra M, Talajic M, Khairy P, Dorian P, et al. Elevateddepression symptoms predict long-term cardiovascular mortality in patients withatrial fibrillation and heart failure. Circulation 2009;120:134–40.

[25] van den Broek KC, Tekle FB, Habibovic M, Alings M, van der Voort PH, Denollet J.Emotional distress, positive affect, and mortality in patients with an implantablecardioverter defibrillator. Int J Cardiol 2013;165:327–32.

[26] Krantz DS, Olson MB, Francis JL, Phankao C, Bairey Merz CN, Sopko G, et al. Anger,hostility, and cardiac symptoms in women with suspected coronary artery disease:the Women's Ischemia Syndrome Evaluation (WISE) Study. J Womens Health(Larchmt) 2006;15:1214–23.

[27] Katon W. Collaborative depression care models: from development to dissemina-tion. Am J Prev Med 2012;42:550–2.

[28] JohnsonBH, Smoyer-TomicKE, Siu K,WalkerDR, Sander S,HuseD, et al. Readmissionamong hospitalized patients with nonvalvular atrial fibrillation. Am J Health SystPharm 2013;70:414–22.

[29] Korte T, Jung W, Ostermann G, Wolpert C, Spehl S, Esmailzadeh B, et al. Hospitalreadmission after transvenous cardioverter/defibrillator implantation; a single cen-tre study. Eur Heart J 2000;21:1186–91.

[30] Christiansen CB, Olesen JB, Gislason G, Lock-Hansen M, Torp-Pedersen C. Cardiovas-cular and non-cardiovascular hospital admissions associated with atrial fibrillation:a Danish nationwide, retrospective cohort study. BMJ Open 2013;3.

[31] Mejhert M, Kahan T, Persson H, Edner M. Predicting readmissions and cardiovascu-lar events in heart failure patients. Int J Cardiol 2006;109:108–13.

[32] Robertson J, McElduff P, Pearson SA, Henry DA, Inder KJ, Attia JR. The health servicesburden of heart failure: an analysis using linked population health data-sets. BMCHealth Serv Res 2012;12:103.

[33] Guize L, Thomas F, Bean K, Benetos A, Pannier B. Atrial fibrillation: prevalence, riskfactors and mortality in a large French population with 15 years of follow-up. BullAcad Natl Med 2007;191:791–803.

[34] Andersson T, Magnuson A, Bryngelsson IL, Frobert O, Henriksson KM, Edvardsson N,et al. All-cause mortality in 272,186 patients hospitalized with incident atrial fibril-lation 1995–2008: a Swedish nationwide long-term case–control study. Eur Heart J2013;34:1061–7.

413S.R. Beach et al. / Journal of Psychosomatic Research 75 (2013) 409–413

[35] Castro P, Bittner V, Villarroel L. Predictors of readmission in patients with congestiveheart failure. Rev Med Chil 1998;126:1173–81.

[36] Calvillo-King L, Arnold D, Eubank KJ, LoM, Yunyongying P, Stieglitz H, et al. Impact ofsocial factors on risk of readmission or mortality in pneumonia and heart failure:systematic review. J Gen Intern Med 2013;28:269–82.

[37] Nikolic G, Miljovic S, Stojkovic A, Manojlovic S, Ciric Z, Zikic O, et al. Influence of psy-chological and coronary parameters on coronary patient rehospitalization. Srp ArhCelok Lek 2010;138:154–61.

[38] Lamblin N, FertinM, de Groote P, Bauters C. Cardiac remodeling and heart failure after afirst anterior myocardial infarction in patients with diabetes mellitus. J Cardiovasc Med(Hagerstown) 2012;13:353–9.

[39] Andres E, Cordero A,Magan P, Alegria E, LeonM, Luengo E, et al. Long-termmortalityand hospital readmission after acute myocardial infarction: an eight-year follow-upstudy. Rev Esp Cardiol (Engl Ed) 2012;65:414–20.

[40] Aranda JM, Johnson JW, Conti JB. Current trends in heart failure readmission rates:analysis of Medicare data. Clin Cardiol 2009;32:47–52.

[41] Desai MM, Stauffer BD, Feringa HH, Schreiner GC. Statistical models and patient pre-dictors of readmission for acute myocardial infarction: a systematic review. CircCardiovasc Qual Outcomes 2009;2:500–7.

[42] Cordero A, Bertomeu-Gonzalez V, Moreno-Arribas J, Agudo P, Lopez-Palop R, MasiaMD, et al. Burden of systemic hypertension in patients admitted to cardiology hos-pitalization units. Am J Cardiol 2011;108:1570–5.

[43] Tsuchihashi M, Tsutsui H, Kodama K, Kasagi F, Setoguchi S, Mohr M, et al. Medicaland socioenvironmental predictors of hospital readmission in patients with conges-tive heart failure. Am Heart J 2001;142:E7.

[44] Felker GM, Leimberger JD, Califf RM, Cuffe MS,Massie BM, Adams KF, et al. Risk strat-ification after hospitalization for decompensated heart failure. J Card Fail 2004;10:460–6.

[45] Rollman BL, Belnap BH, LeMenager MS, Mazumdar S, Houck PR, Counihan PJ, et al.Telephone-delivered collaborative care for treating post-CABG depression: a ran-domized controlled trial. JAMA 2009;302:2095–103.

[46] Rollman BL, Belnap BH, Mazumdar S, Houck PR, Zhu F, Gardner W, et al. A random-ized trial to improve the quality of treatment for panic and generalized anxiety dis-orders in primary care. Arch Gen Psychiatry 2005;62:1332–41.

[47] Whang W, Burg MM, Carney RM, Freedland KE, Bigger JT, Catellier D, et al. Designand baseline data from the vanguard of the Comparison of Depression Interventionsafter Acute Coronary Syndrome (CODIACS) randomized controlled trial. ContempClin Trials 2012;33:1003–10.

[48] Babyak MA. What you see may not be what you get: a brief, nontechnicalintroduction to overfitting in regression-type models. Psychosom Med 2004;66:411–21.

[49] Taylor CB, Youngblood ME, Catellier D, Veith RC, Carney RM, Burg MM, et al. Effectsof antidepressant medication on morbidity and mortality in depressed patients aftermyocardial infarction. Arch Gen Psychiatry 2005;62:792–8.