O

R

MSa

b

c

a

ARRAA

KCCQDHS

I

mlddiwd(c

Sf

0d

Journal of Cardiology 60 (2012) 72–77

Contents lists available at SciVerse ScienceDirect

Journal of Cardiology

jo u rn al hom epa ge: www.elsev ier .com/ loca te / j j cc

riginal article

isk factors for positive depression screens in hospitalized cardiac patients

ario A. Caro (MD)a, Gillian L. Sowden (BA)b, Carol A. Mastromauro (LICSW)a,tephanie Mahnks (RN)a, Scott R. Beach (MD)a,b, James L. Januzzi (MD)b,c, Jeff C. Huffman (MD)a,b,∗

Department of Psychiatry, Massachusetts General Hospital, Boston, MA, USAHarvard Medical School, Boston, MA, USADepartment of Medicine, Massachusetts General Hospital, Boston, MA, USA

r t i c l e i n f o

rticle history:eceived 26 September 2011eceived in revised form 27 January 2012ccepted 31 January 2012vailable online 20 March 2012

eywords:ardiovascular diseaseardiology practice/managementuality improvementepressioneart diseasescreening

a b s t r a c t

Background: Depression is common in patients with cardiac illness and is independently associated withelevated morbidity and mortality. There are screening guidelines for depression in cardiac patients, butthe feasibility and cost-effectiveness of screening all cardiac patients is controversial. This process may beimproved if a subset of cardiac patients at high risk for depression could be identified using informationreadily available to clinicians and screened.Objective: To identify risk factors for a positive depression screen at the time of admission in hospitalizedcardiac patients.Methods: A total of 561 consecutively screened cardiac inpatients underwent the Patient HealthQuestionnaire-2 (PHQ-2). A prospective chart review was performed to assess potential risk factors fordepression that would be readily available to front-line clinicians. Rates of risk factors were comparedbetween patients with positive and negative PHQ-2 depression screens, and multivariate logistic regres-sion was performed to assess whether specific risk factors were independently associated with positivescreens.Results: Of the 561 patients screened, 13.5% (n = 76) had a positive depression screen (PHQ-2 ≥ 2). In theunivariate analyses, several variables were associated with a positive depression screen. On multivari-

9

ate analysis, an elevated white blood cell (WBC) count (>10 × 10 cells per liter) and prescription of anantidepressant on admission were independently associated with a positive depression screen, whilecurrent smoking showed a trend toward significance.Conclusion: Information on these three identified risk factors (WBC count, antidepressant use, and smok-ing) is readily available to clinicians, and patients with these diagnoses may represent a cohort whowould benefit from targeted depression screening in certain settings.

2 Jap

© 201

ntroduction

Depression and heart disease are common conditions withajor public health significance. Coronary artery disease is the

eading cause of death worldwide, causing more than 7 millioneaths per year [1]. Depression, another leading cause of globalisability [1], is common in patients with cardiac disease. Approx-

mately 15–20% of patients with cardiac disease, including thoseith congestive heart failure (CHF) [2,3] or acute coronary syn-

rome (ACS) [4,5] meet criteria for major depressive disorderMDD). In these patients, depression has been independently asso-iated with decreased health-related quality of life [6], recurrent

∗ Corresponding author at: Massachusetts General Hospital, 55 Fruittreet/Warren 1220C, Boston, MA 02114, USA. Tel.: +1 617 724 2910;ax: +1 617 724 9155.

E-mail address: jhuffman@partners.org (J.C. Huffman).

914-5087/$ – see front matter © 2012 Japanese College of Cardiology. Published by Elseoi:10.1016/j.jjcc.2012.01.016

anese College of Cardiology. Published by Elsevier Ltd. All rights reserved.

cardiac events and cardiac complications [7–9], increased rates ofhospitalization [10], and increased mortality at both short- andlong-term follow-up [10,11].

Given the effects of depression in cardiac patients, the AmericanHeart Association (AHA) has recommended systematic depres-sion screening of cardiac patients using the Patient HealthQuestionnaire-2 (PHQ-2) as an initial screen [12]. Although in somecircumstances systematic screening may be feasible [13], the AHArecommendations have been controversial given that depressionscreening alone does not appear to improve medical outcomes inprimary care or cardiac patients [14] and that screening of largenumbers of patients may not be practical in some settings. In situa-tions where universal screening is impractical, it may be desirableto screen a subpopulation of cardiac patients at high risk for depres-

sion, using readily available clinical or demographic data.

Prior studies of risk factors for depression or positive depressionscreens in cardiac patients have typically focused on a single orsmall number of cardiac diseases and had relatively small sample

vier Ltd. All rights reserved.

of Card

s(t

ioescial

dac

M

O

aAsscoh

S

dsAsapinotsmdredWil

eiettpfappai

M.A. Caro et al. / Journal

izes [15]. In addition, some information identified as ‘risk factors’e.g. scores on research instruments) may not be readily availableo clinicians in busy, real-world settings.

For selective depression screening to be feasible and effective,t should utilize information that is readily available to clinicians,ccur over a broad range of applicable diagnoses to provide anconomy of scale, and be performed in patients for whom depres-ion has a substantial impact. To our knowledge, prior studies inardiac patients have not examined risk factors for depression innpatients at the time of hospital admission, assessed patients with

wide range of cardiovascular illnesses, or screened a relativelyarge number of consecutively admitted cardiac patients.

To address this gap, we examined risk factors for a positiveepression screen in a consecutively-screened cohort of patientsdmitted with a wide range of cardiac diagnoses to one of the threeardiac units at Massachusetts General Hospital.

ethods

verview

This was a prospective chart review of consecutive patientsdmitted to cardiac units at an urban academic medical center.ll patients admitted to the units received a two-item depressioncreen via the PHQ-2 as part of clinical care. Charts were sub-equently reviewed to assess for sociodemographic and medicalharacteristics that may have been associated with the presencer absence of a positive PHQ-2. This study was approved by ourealthcare system’s Institutional Review Board.

ubjects and depression screening

Subjects were patients who were admitted to one of three car-iac inpatient units (one cardiac intensive care unit and two cardiactep-down units) at an urban academic medical center betweenpril 14, 2008 and June 9, 2008 with a primary cardiac diagno-is, such as CHF, myocardial infarction (MI), unstable angina (UA),rrhythmia, congenital heart disease, stent placement, pacemakerlacement, or any other cardiac pathology that warranted hospital-

zation. Patients received depression screening from clinical cardiacurses via the PHQ-2 during the initial nursing assessment as partf clinical care. Nurses had been trained in person (and with writ-en instructions) on the administration of the PHQ-2 depressioncreening items at daytime and evening nursing staff meetings onultiple occasions. The PHQ-2 includes two items (to assess anhe-

onia and depressed mood for the prior two weeks) and its scoreange is 0–6 [16]. Patients with positive screens were evaluated forligibility for a prospective trial of depression management in car-iac patients (www.clinicaltrials.gov, trial number NCT00847132).e excluded patients for whom screening was not possible due to

ntubation, sedation, mental status changes, or due to poor Englishanguage comprehension.

The PHQ-2 was selected as a screening tool because of its brevity,asy implementation, validation and use in cardiac patients, and itsnclusion in the AHA recommendations [12,16]. For the purposes ofligibility for the clinical trial, a PHQ-2 cutoff score of ≥3 was usedo reduce the number of false positives. However, for this prospec-ive chart review, the more standard PHQ-2 score cutoff of ≥2 as aositive depression screen was used; prior work in cardiac patientsound this to be an optimal cutoff score, with a sensitivity of 82%nd specificity of 79% for major depression [17]. In addition, com-

ared to a cutoff score of ≥3, a cutoff of ≥2 identifies more potentialatients with depression, and given the relative ease of completing

follow-up assessment for depression (e.g. with the PHQ-9), theres not significant concern about being overly inclusive.

iology 60 (2012) 72–77 73

Chart review

For this prospective chart review, the medical records and PHQ-2 screening results were recorded for patients admitted to theunits between April 14, 2008 and June 9, 2008 with a primary car-diac diagnosis. For each admission, the subject’s electronic medicalrecord, including admission note, laboratory data, and dischargesummary were reviewed to assess for several demographic, psy-chosocial, medical, and medication-related characteristics.

Although there may be many characteristics associated witha positive depression screen, we limited the variables we exam-ined as potential risk factors. First, given that the goal was toidentify variables that could be used by front-line clinicians tostratify depression risk, we required that each variable be readilyavailable to clinicians from prior records or the patient as part ofroutine cardiac evaluation; any risk factor requiring inquiry aboutadditional factors (e.g. history of anxiety disorder) or additionalevaluation (e.g. data from research instruments such as social sup-port scales) was not included. Second, we carefully selected thevariables based on prior literature on risk of depression in cardiacand other medically-ill patients.

Based on these factors, we selected the following variables,given their prior association with depression and cardiac illnesses:

• Demographic characteristics: gender, age, marital status, livingalone, and employment status [18–21].

• Medical historical characteristics: diabetes mellitus (DM), hyper-tension (HTN), prior MI, prior cerebrovascular accident (CVA),history of coronary artery bypass graft (CABG), and medical hos-pitalization within the past 6 months [7,8,10,18].

• Admission characteristics: admission diagnosis (MI versusadmission for another cause), current smoking at admission,white blood cell (WBC) count on admission, and use of anxiolytics,antidepressants, or beta-blockers at admission [18,22–25].

Of these variables, we also, a priori, selected the subset of vari-ables we felt to be most likely to be associated with a positivedepression screen to be examined as part of an exploratory logis-tic regression analysis as described below. This limitation of testedvariables was done to reduce the number of variables in the sta-tistical model and therefore decrease the risk of overfitting themodel.

Statistical analysis

Statistical analysis was performed using Stata version 11.0 (Stat-aCorp LP, College Station, TX, USA). All statistical tests were twotailed, with p < 0.05 defined as statistical significance.

Univariate analysis: Each of the clinical and demographic char-acteristics above was analyzed to assess its association with apositive depression screen as indicated by a PHQ-2 score of ≥2.Between-group differences in these variables were compared usingchi-square analyses for categorical variables (e.g. DM) and indepen-dent samples t tests for continuous variables (e.g. WBC count).

Multivariate analysis: To assess the independent association ofthese variables with positive depression screens, we used logisticregression analysis to create a multivariate model for the predictionof positive depression screens indicated by a PHQ-2 score of ≥2. Toreduce overfitting in the regression model and otherwise optimizeanalysis, we avoided preliminary ‘testing’ of variables via univari-ate analysis, reduced the numbers of confounders tested, avoidedautomatic stepwise regression techniques, and did not dichotomize

continuous variables [26]. For this analysis, we selected a priorithe variables that appeared to be most clearly associated withrisk of depression in prior studies [7,8,10,18–20,22–24], namelygender, age, marital status, living alone, history of DM, admission

74 M.A. Caro et al. / Journal of Car

dpa

ssaargd

aan(u

nTv

screen were younger age (60.10 ± 11.4 years vs. 66.56 ± 14.0 years;

TB

VCt

Fig. 1. Patient Health Questionnaire (PHQ)-2 score distributions.

iagnosis (MI vs. other causes), tobacco use, hospitalization in therior 6 months, high WBC count, use of anxiolytics, beta-blockers,nd antidepressants.

Exploratory analyses. We performed several exploratory analy-es to better characterize the variables linked to positive depressioncreens. First, we performed univariate analyses (using chi-squarenalysis and independent samples t tests as appropriate) of all vari-bles using the more stringent PHQ-2 cutoff of ≥3; because of theelatively small number of patients with a PHQ-2 score of 3 orreater, we did not perform multivariate analysis using this cutoffue to the potential for overfitting [26].

Second, if any variables were independently associated with positive depression screen, we then calculated operating char-cteristics [sensitivity, specificity, positive predictive value (PPV),egative predictive value (NPV), and overall correct classificationOCC)] for predicting positive depression screens for each individ-al variable

Third, we then completed post hoc analyses by admission diag-

osis (e.g. UA) for any diagnostic group with over 100 subjects.hese, again, would be univariate analyses assessing each of theariables’ association with a positive depression screen, but would

able 1aseline characteristics of patients by depression screen status.

Negative depression screen, PHQ-2score <2 (n = 485)

Study variablesDemographic characteristicsGender, Male 313 (64.5)

Age, mean (SD) 66.39 (13.9)

Married 318 (65.6)

Lives alone 110 (22.6)

Employed 188 (38.7)

Medical variablesDM 139 (28.6)

Hyperchol. 289 (59.5)

HTN 312 (55.6)

Current smoker 51 (10.5)

Prior MI 96 (19.7)

Prior CVA 67 (13.8)

CABG 81 (16.7)

Hosp 6 months 132 (27.2)

WBC count, mean (SD) 8.48 (5.4)

Admission diagnosis of MI 17 (15.7)

Admission medicationsAnxiolytic 37 (7.6)

Beta-blocker 329 (67.8)

Antidepressant 63 (12.9)

ariables are reported as n (%) unless otherwise specified.ABG, coronary artery bypass graft; CVA, cerebrovascular accident; DM, diabetes mellituerolemia; HTN, hypertension; MI, myocardial infarction; PHQ-2, Patient Health Question

diology 60 (2012) 72–77

not include multivariate analysis via regression because of the smallsample size and thus overfitting.

Results

Overall, 561 patients were consecutively screened with thePHQ-2. The primary diagnoses for these patients were UA (n = 198;35.3%), arrhythmia (n = 133; 23.7%), MI (n = 90; 16.0%), CHF (n = 71;12.7%), closure of patient foramen ovale (n = 19; 3.4%), valvulardisease (n = 16; 2.9%), peripheral arterial disease (n = 7; 1.2%), andother diagnoses (n = 27; 4.8%). The distribution of PHQ-2 scores inthe screened patients is displayed in Fig. 1; in total, 76 (13.5%) ofpatients had a PHQ-2 score of 2 or greater and 47 (8.4%) had a PHQ-2score of 3 or greater.

On univariate analysis (Table 1), variables associated witha PHQ-2 score of 2 or greater were being unmarried (47.3%with positive depression screen were married vs. 89.8% withnegative depression screen; �2 = 9.34; p = 0.002), living alone(36.8% vs. 22.6%; �2 = 7.1; p = 0.008), current smoking (23.6% vs.10.5%; �2 = 10.56; p = 0.001), having a higher mean WBC count(10.72 ± 9.8 × 109 cells per liter vs. 8.48 ± 5.4 × 109 cells per liter;t = −2.88; p = 0.004), and prescription of antidepressant medication(30.8% vs. 12.9%; �2 = 15.1; p < 0.001). There were no significantbetween-group differences in any of the other variables.

After controlling for confounding via multivariate logisticregression analysis (Table 2), having a higher admission WBCcount [odds ratio (OR) = 2.19; 95% confidence interval (CI):1.25–3.83; p = 0.006] and prescription of antidepressant medica-tions (OR = 2.36; 95% CI: 1.30–4.28; p = 0.005) were independentlyassociated with a positive depression screen using a cutoff of PHQ-2 score of 2 or greater. Additionally, current smoking at admissionshowed a trend toward significance (OR = 1.86; 95% CI: 0.93–3.72;p = 0.079).

Regarding exploratory analyses, when a cutoff score of PHQ-2 ≥ 3 was used, variables associated with a positive depression

t = 3.08; p = 0.002), current smoking (25.5% with positive depressionscreen were smokers vs. 11.1% with negative depression screen;�2 = 8.32; p = 0.004), not having been hospitalized in the past 6

Positive depression screen, PHQ-2score ≥ 2 (n = 76)

p-Value

41 (53.9) 0.0863.61 (13.3) 0.9536 (47.3) 0.00228 (36.8) 0.00823 (30.2) 0.16

23 (30.2) 0.7751 (67.1) 0.2150 (65.7) 0.8118 (23.6) 0.00114 (18.4) 0.7815 (19.7) 0.1714 (18.4) 0.7115 (19.7) 0.1710.72 (9.8) 0.00414 (18.4) 0.54

10 (13.1) 0.1149 (64.4) 0.5623 (30.2) <0.001

s; Hosp 6 months, hospitalized in the previous 6 months; Hyperchol, hypercholes-naire-2; SD, standard deviation; WBC, white blood cell.

M.A. Caro et al. / Journal of Cardiology 60 (2012) 72–77 75

Table 2Logistic regression assessing variables associated with positive depression screens.

OR Z-score p-Value 95% Confidence interval for OR

Gender 0.76 −1.03 0.30 0.44–1.29Age 0.99 −1.26 0.21 0.97–1.01Marriage 0.77 −0.69 0.49 0.37–1.61Lives alone 1.49 1.03 0.30 0.70–3.17DM 1.13 0.41 0.69 0.64–1.98MI at admission 0.72 −0.92 0.36 0.35–1.46Current smoker 1.86 1.76 0.079 0.93–3.73Hosp 6 months 0.68 −1.17 0.24 0.36–1.29WBC count > 10 2.19 2.74 0.006 1.25–3.82Anxiolytic 1.48 0.96 0.34 0.67–3.31Beta-blocker 1.00 0.01 0.99 0.57–1.78

D I, my

maspwvd

lo(st

wlhhshwe(aa

D

usaoa

tip

TO

Np

Antidepressant 2.36 2.83

M, diabetes mellitus; Hosp 6 months, previous hospitalization within 6 months; M

onths (27.4% vs. 12.8%; �2 = 4.79; p = 0.029), prescription of annxiolytic medication (19.1% vs. 7.4%; �2 = 7.75; p = 0.005), and pre-cription of antidepressant medication (29.8% vs. 14.0%; �2 = 8.26;

= 0.004). WBC count remained numerically greater in patientsith positive depression screens (9.90 ± 5.6 × 109 cells per liter

s. 8.68 ± 6.3 × 109 cells per liter), but was no longer significantlyifferent (p = 0.21).

Operating characteristics of these predictor variables are out-ined in Table 3. Overall, 26.1% (18/69) of smokers, 22.8% (28/123)f patients with a WBC count > 10.0 × 109 cells per liter, and 30.2%26/86) of patients on antidepressants had a positive depressioncreen. Furthermore, 35.8% (19/53) subjects with two or more ofhese three factors had a positive depression screen.

Finally, for our subgroup analyses by diagnosis, for patientsith UA, we found that being unmarried (�2 = 5.66; p = 0.017),

iving alone (�2 = 5.33; p = 0.021), smoking (�2 = 6.83; p = 0.009),aving a history of depression (�2 = 23.76; p < 0.001), having aistory of anxiety (�2 = 7.82; p = 0.005), and antidepressant pre-cription (�2 = 5.54; p = 0.019) were significantly associated withaving a positive depression screen. For patients with arrhythmia,e found that smoking (�2 = 4.67; p = 0.031), history of anxi-

ty (�2 = 14.32; p < 0.001), WBC count > 10.0 × 109 cells per liter�2 = 11.33; p = 0.001), anxiolytic prescription (�2 = 7.19; p = 0.007),nd antidepressant prescription (�2 = 4.20; p = 0.040) were associ-ted with positive depression screens.

iscussion

In sum, in a sample of 561 patients admitted to inpatient cardiacnits, we found that higher WBC count at admission and pre-cription of antidepressants were independently associated with

positive PHQ-2 depression screen when using a cutoff score of 2r greater. Smoking, although not statistically significant, showed

trend toward significance.Our findings were largely consistent with previous research in

his area. Our prevalence rate of positive depression screens (13.5%)s higher than the prevalence rate of depression of 5% in the generalopulation [11] and similar to the prevalence rates of depression

able 3perating characteristics for our depression screen predictors.

WBC > 10 Antidepressant Smoking

Sensitivity 27/76 = 0.36 23/76 = 0.30 18/76 = 0.24Specificity 394/485 = 0.81 422/486 = 0.87 434/485 = 0.89PPV 27/118 = 0.23 23/86 = 0.27 18/69 = 0.26NPV 394/443 = 0.89 422/475 = 0.89 434/492 = 0.88OCC 421/561 = 0.75 445/561 = 0.79 452/561 = 0.81

PV, negative predictive value; OCC, overall correct classification; PPV, positiveredictive value; WBC, white blood cell.

0.005 1.30–4.28

ocardial infarction; OR, odds ratio; WBC, white blood cell.

found in other cardiac populations of 15–20% [2,5]; it is also similarto the rate of positive depression screens (PHQ-2 ≥ 2) of 16.8% in alarge prospective study of cardiac outpatients [17].

The relation between elevated WBC count and depression iswidely reported in the literature [23,27], including reports of suchan association in patients with cardiac disease [23]. Although thereasons for such an elevation of WBC in depressed patients isunclear, one hypothesis is that depression, as an inflammatorystate [28], may lead to elevation of WBC values in serum andto increased production of interleukins, leading to atherosclerosisprogression. However, there is conflicting evidence in the relationbetween depression in cardiac patients and inflammatory mark-ers, with some authors finding an association [22,29], while othersfindings a very small or no association [30,31]. In our sample, whichinvestigated only those parameters that were immediately avail-able to front-line clinicians, we did not assess the impact of multipleinflammatory markers, and even WBC differential counts (e.g. toassess neutrophil and lymphocyte counts) were not available inthe majority of patients.

Prescription of antidepressant medication at admission was alsoindependently associated with a positive depression screen. Thisfinding likely serves as a marker for those patients with recently orpreviously diagnosed depressive disorders (or those with illnessesfrequently comorbid with depression and treated with antidepres-sants, such as anxiety disorders). Given that the majority of patientsreceiving antidepressants do not receive treatment with adequatedoses or duration, and that several trials of treatment are oftenrequired before remission is reached [32], it is not surprising thatthose patients who have depression and are treated with antide-pressants may still have ongoing depressive symptoms; durationof antidepressant use for each patient was not known. Of note,we deliberately did not include a recorded history of depression(or anxiety) as a potential risk factor. Our clinical and researchexperience has been that a history of depression and anxiety arerather inconsistently reported in cardiac patients’ past medicalhistory, and we instead chose to use antidepressant medicationprescription as a more objective potential marker of being at riskfor depression at admission.

Finally, smoking, although not a statistically significant risk fac-tor in multivariate analysis, did show a trend toward significance.This finding is consistent with previous findings over decades[18,22,24]. The cause of this association is not yet fully understood;some authors have proposed that nicotine use may be a methodof self-medication to reduce anxiety and dysphoria in people withdepression [33]. More recent publications have suggested a step-wise relationship between tobacco use and the subsequent onset of

depressive symptoms [34,35]; it is not clear whether smoking maycause depressive symptoms or whether early smoking is a markerof psychosocial distress or other factors that are linked to depres-sion. Given the small number of current smokers in this cohort

7 of Car

(nter

ruasapfsfWa

bbctdrcineat

oipcccarriotdtspb[

sticopamffictphs

6 M.A. Caro et al. / Journal

approximately 12%), one possible explanation for this lack of sig-ificance is that our study sample did not allow for adequate powero detect a difference in depression screen rates between smok-rs and nonsmokers. It is possible that a larger sample may haveesulted in significant between-group differences.

Regarding demographic variables, only living alone and unmar-ied status were associated with positive depression screens onnivariate analysis, and no demographic variables were associ-ted with positive screens on multivariate analysis. Some priortudies in cardiac patients have found links between depressionnd younger age [20,36], female gender [23], and poor social sup-ort/living alone [36]. The findings in our sample may be due to theact that our patients had a wide variety of cardiac diseases (priortudies evaluated patients with a single cardiac illness), that someactors we analyzed were not included in the prior studies (e.g.

BC count), and that our evaluations were performed at the time ofdmission (other studies have performed screens after discharge).

We chose to assess depression at admission. Although it maye argued that depressive symptoms on hospital admission coulde a transient reaction to acute medical illness, we have found thatardiac patients identified with the PHQ-2 on admission and foundo have clinical depression on further evaluation with the PHQ-9id not have transient depression but instead often had chronic,ecurrent, and moderately severe depression [37]. In addition, iflinically-significant depression is identified in the hospital, thenpatient setting allows time for further evaluation, care coordi-ation, and initiation of treatment in a monitored setting; indeed,arly treatment initiation may be important given that depressionppears to be associated with increased rates of cardiac complica-ions even in the short term [7,9].

This study had both strengths and limitations. One strengthf this investigation involves the wide range of cardiac patientsncluded in the analyses; this is important given that screeningatients with only certain diagnoses may be more logisticallyomplicated in busy clinical settings, and depression is asso-iated with negative outcomes in nearly every cardiovascularondition [7,8,10]. We used variables that would typically bevailable to front-line cardiac nurses and physicians withoutequiring extra effort or specialized knowledge to obtain theisk factor data, potentially increasing the feasibility of perform-ng such screening. Furthermore, we evaluated the predictorsf a positive depression screen at the time of admission tohe hospital; prior studies of depression risk factors in car-iac patients have focused on correlates of depression at theime of discharge or following hospitalization [4,38]. Depressioncreening at admission flows well as part of the overall intakerocess, and we have found that such screening is well-acceptedy patients and staff and not substantially resource-intensive13].

Limitations include the fact that this study was performed in aingle academic medical center, with a largely Caucasian popula-ion. We could not analyze all possible risk factors for depressionn the logistic regression model to avoid overfitting, and even oururrent analysis must be considered somewhat exploratory despiteur relatively large sample of over 500 consecutively-screenedatients. We also did not have adequate information to analyzedditional variables of interest, including WBC differential, bodyass index, CHF class, or duration of medical symptoms. Given the

ocus of this study on ‘real-world’ depression screening and riskactor availability, depression screening and listing of risk factorsn admission notes was performed by clinicians as part of usualare without monitoring or interference from study staff. Finally,

his study only investigated positive depression screens and notatients with a formal depression diagnosis, although prior workas found high rates of correlation between PHQ-2 depressioncreens and major depression [16]. Furthermore, a recent study

[

diology 60 (2012) 72–77

found that positive PHQ-2 screens independently predicted 12month mortality in CHF patients [39].

In conclusion, in a sample of 561 consecutively screened car-diac inpatients, we found that increased WBC count at admissionand the use of antidepressants were independently associatedwith a positive PHQ-2 depression screen, while smoking showeda trend toward significance. There is currently substantial con-troversy regarding the routine screening of cardiac patients fordepression [14], and some settings may choose to focus limitedresources by screening a subgroup of the population at a higherrisk for depression. If so, a program that screens those with WBCcounts >10.0 × 109 cells per liter, those taking antidepressants,and perhaps current smokers, should yield higher rates of positivedepression screens.

Although the predictive value of these three variables is to somedegree limited, using the presence or absence of our three vari-ables as a prompt for screening would approximately double thepositive screening rate for patients while avoiding screening inthe majority of patients. Thus this information may have clini-cal utility in certain settings by substantially reducing providertime and energy. In addition, even if this information does nothave direct clinical utility, understanding more about risk fac-tors for depression/depression screens in cardiac patients may stillbe of importance, given the substantial and independent linksbetween depression and morbidity/mortality in this population.For example, this work could add to the literature on smoking anddepression in cardiac patients, both of which have been indepen-dently linked with mortality, and may prompt further examinationof interconnections between stress response/margination, depres-sion, inflammation, and cardiac outcomes, given the link betweenWBC count and positive depression screens in our analysis.

Acknowledgments

This study was funded in part by AHA Scientist DevelopmentGrant No. 0735530T to Jeff Huffman, M.D. The authors of this paperreport no financial or other conflict of interests regarding the con-tent of this manuscript.

References

[1] World Health Organization. The global burden of disease: 2004 update. Geneva,Switzerland: WHO press; 2008. Accessed 25 January 2011 at: http://www.who.int/healthinfo/global burden disease/GBD report 2004update full.pdf.

[2] Rutledge T, Reis VA, Linke SE, Greenberg BH, Mills PJ. Depression in heart failure:a meta-analytic review of prevalence, intervention effects, and associationswith clinical outcomes. J Am Coll Cardiol 2006;48:1527–37.

[3] Freedland KE, Rich MW, Skala JA, Carney RM, Dávila-Román VG, Jaffe AS.Prevalence of depression in hospitalized patients with CHF. Psychosom Med2003;65:119–28.

[4] Dickens C, McGowal L, Percival C, Douglas J, Tomenson B, Cotter L, HeaghertyA, Creed F. Association between depressive episode before first myocar-dial infarction and worse cardiac failure following infarction. Psychosomatics2005;46:523–8.

[5] Thombs BD, Bass EB, Ford DE, Stewart KJ, Tsilidis KK, Patel U, Fauerbach JA,Bush DE, Ziegelstein RC. Prevalence of depressive symptoms in survivors ofacute myocardial infarction. J Gen Intern Med 2006;21:30–8.

[6] de Jonge P, Spijkerman TA, van den Brink RH, Ormel J. Depression after myocar-dial infarction is a risk factor for declining health related quality of life andincreased disability and cardiac complaints at 12 months. Heart 2006;92:32–9.

[7] Huffman JC, Smith FA, Blais MA, Taylor AM, Januzzi JL, Fricchione GL. Pre-existing major depression predicts in-hospital complications after acutemyocardial infarction. Psychosomatics 2008;49:309–16.

[8] Frasure-Smith N, Lespérance F. Depression and anxiety as predictors of 2 yearcardiac events in patients with stable coronary artery disease. Arch Gen Psy-chiatry 2008;65:62–71.

[9] Grewal K, Stewart DE, Abbey SE, Leung YW, Irvine J, Grace SL. Timing of depres-sive symptoms onset and in-hospital complications among acute coronary

artery syndrome inpatients. Psychosomatics 2010;51:283–8.

10] Jiang W, Alexander J, Christopher E, Kuchibhatla M, Gaulden LH, Cuffe MS, Blaz-ing MA, Davenport C, Califf RM, Krishnan RR, O’Connor CM. Relationship ofdepression to increased risk of mortality and rehospitalization in patients withcongestive heart failure. Arch Intern Med 2001;161:1849–56.

of Card

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

M.A. Caro et al. / Journal

11] Glassman AH, Bigger Jr TJ, Gaffney M. Psychiatric characteristics associated withlong-term mortality among 361 patients having an acute coronary syndromeand major depression: seven-year follow-up of SADHART participants. ArchGen Psychiatry 2009;66:1022–9.

12] Lichtman JH, Bigger Jr JT, Blumenthal JA, Frasure-Smith N, Kaufmann PG,Lespérance F, Mark DB, Sheps DS, Taylor CB, Froelicher ES. AHA science advisory.A science advisory from the American Heart Association Prevention Committeeof the Council on Cardiovascular Nursing, Council on Clinical Cardiology, Coun-cil on Epidemiology and Prevention, and Interdisciplinary Council on Qualityof Care and Outcomes Research. Endorsed by the American Psychiatric Associ-ation. Circulation 2008;118:1768–75.

13] Sowden G, Mastromauro CA, Januzzi JL, Fricchione GL, Huffman JC. Detection ofdepression in cardiac patients: feasibility and results of systematic screening.Am Heart J 2010;159:780–7.

14] Thombs BD, de Jonge P, Coyne JC, Whooley MA, Frasure-Smith N, Mitchell AJ,Zuidersma M, Eze-Nliam C, Lima BB, Smith CG, Soderlund K, Ziegelstein RC.Depression screening and patient outcomes in cardiovascular care: a system-atic review. JAMA 2008;300:2161–71.

15] Safdar B, Foody JM, D’Onofrio G. Depression as a modifiable risk factor in theemergency department chest pain observation unit: a pilot. Crit Pathw Cardiol2010;9:82–7.

16] Löwe B, Kroenke K, Gräfe K. Detecting and monitoring depression with a two-item questionnaire (PHQ-2). J Psychosom Res 2005;58:163–71.

17] Thombs BD, Ziegelstein RC, Whooley MA. Optimizing detection for majordepression among patients with coronary artery disease using the patienthealth questionnaire: data from the Heart and Soul Study. J Gen Intern Med2008;12:2014–7.

18] Mimiaga MJ, Reichman WM, Safren SA, Losina E, Arbelaez C, WalenskyRP. Prevalence and correlates of clinically significant depressive symptomsin an urban hospital department. Prim Care Companion J Clin Psychiatry2010;12:e1–11.

19] Ansseau M, Fischler B, Dierick M, Albert A, Leyman S, Mignon A. Socioeconomiccorrelates of generalized anxiety disorder and major depression in primarycare: the GADIS II Study (Generalized Anxiety and Depression Impact SurveyII). Depress Anxiety 2008;25:506–13.

20] Rohyans LM, Pressler SJ. Depressive symptoms and heart failure: examiningthe sociodemographic variables. Clin Nurse Spec 2009;23:138–44.

21] Shimokawa H, Yasuda S. Myocardial ischemia: current concepts and futureperspectives. J Cardiol 2008;52:67–78.

22] Surtees P, Wainwright N, Day N, Luben R, Brayne C, Khaw KT. Association ofdepression with peripheral leukocyte counts in EPIC-Norfolk – role of sex andcigarette in smoking. J Psychosom Res 2003;54:303–6.

23] Kop WJ, Stein PK, Tracy RP, Barzilay JI, Schultz R, Gottdiener JS. Autonomicnervous system dysfunction and inflammation contribute to the increased car-diovascular risk associated with depression. Psychosom Med 2010;72:626–35.

24] Naqvi TZ, Rafique AM, Andreas V, Rahban M, Mirocha J, Naqvi SS. Pre-

dictors of depressive symptoms post-acute coronary syndrome. Gend Med2007;4:339–51.

25] Yagi H, Komukai K, Hashimoto K, Kawai M, Oqawa T, Anzawa R, MinaiK, Nagoshi T, Oqawa K, Taniguchi I, Yoshimura M. Difference in risk fac-tors between acute coronary syndrome and stable angina pectoris in the

[

iology 60 (2012) 72–77 77

Japanese: smoking as a crucial risk factor of acute coronary syndrome. J Cardiol2010;55:345–53.

26] Babyak MA. What you see may not be what you get: a brief, nontechni-cal introduction to overfitting in regression-type models. Psychosom Med2004;66:411–21.

27] Kobrosly R, van Wijngaarden E. Associations between immunologic, inflam-matory, and oxidative stress markers with severity of depressive symptoms.Neurotoxicology 2010;31:126–33.

28] Maes M, Yirmyia R, Noraberg J, Brene S, Hibbein J, Perini G, Kubera M, Bop P,Lerer B, Maj M. The inflammatory & neurodegenerative (I&D) hypothesis ofdepression: leads for future research and new drug treatments in depression.Metab Brain Dis 2009;24:27–53.

29] Deverts JD, Cohen S, DiLillo VG, Lewis CE, Kiefe C, Whooley M, Matthews KA.Depressive symptoms, race, and circulating C-reactive protein: the CoronaryArtery Risk Development in Young Adults (CARDIA) study. Psychosom Med2010;72:734–41.

30] Whooley MA, Caska CM, Hendrickson BE, Rourke MA, Ho J, Ali S. Depressionand inflammation in patients with coronary heart disease: findings from theHeart and Soul Study. Biol Psychiatry 2007;62:314–20.

31] Davidson KW, Schwartz JE, Kirkland SA, Mostofsky E, Fink D, Guernsey D,Shimbo D. Relation of inflammation to depression and incident coronary heartdisease (from the Canadian Nova Scotia Health Survey [NSHS95] ProspectivePopulation Study). Am J Cardiol 2009;103:755–61.

32] Trivedi MH, Fava M, Wisniewski SR, Thase ME, Quitkin F, Warden D, Ritz L,Nierenberg AA, Lebowitz BD, Biggs MM, Luther JF, Shores-Wilson K, Rush AJ,STAR*D Study Team. Medication augmentation after the failure of SSRIs fordepression. N Engl J Med 2006;354:1243–52.

33] Breslau N, Peterson EL, Schultz LR, Chilcoat HD, Andreski P. Major depres-sion and stages of smoking. A longitudinal investigation. Arch Gen Psychiatry1998;55:161–6.

34] Flensburg-Madsen T, von Scholten MB, Flachs EM, Mortensen EL, PrescottE, Tolstrup JS. Tobacco smoking as a risk factor for depression. A 26-yearpopulation-based follow-up study. J Psychiatr Res 2011;54:143–9.

35] Pasco JA, Williams LJ, Jacka FN, Ng F, Henry MJ, Nicholson GC, Kotowicz MA, BerkM. Tobacco smoking as a risk factor for major depressive disorder: a population-based study. Br J Psychiatry 2008;193:322–6.

36] Barefoot JC, Burg MM, Carney RM, Cornell CE, Czaikowski SM, Freedland KE,Hosking JD, Khatri P, Pitula CR, Sheps D. Aspects of social support associatedwith depression at hospitalization and follow-up assessment among cardiacpatients. J Cardiopulm Rehabil 2003;23:404–12.

37] Huffman JC, Mastromauro CA, Sowden GL, Wittmann C, Rodman R, JanuzziJL. A collaborative care depression management program for cardiac inpa-tients: depression characteristics and in-hospital outcomes. Psychosomatics2011;52:26–33.

38] Koivula M, Halme N, Asted-Kurki P. Predictors of depressive symptoms amongcoronary heart disease patients: a cross-sectional study nine years after coro-

nary artery bypass grafting. Heart Lung 2010;39:421–31.

39] Rollman BL, Belnap BH, Mazmudar S, Houck PR, He F, Alvarez RJ, Schulberg HC.Reynolds CF, McNamara DM., A Positive 2-Item Patient Health QuestionnaireDepression Screen Among Hospitalized Heart Failure Patients Is AssociatedWith Elevated 12-Month Mortality. J Card Fail 2012; in press.