General Hospital Psychiatry 37 (2015) 1–6

Contents lists available at ScienceDirect

General Hospital Psychiatry

j ourna l homepage: ht tp : / /www.ghp journa l .com

Psychiatric–Medical ComorbidityThe Psychiatric–Medical Comorbidity sectionwill focus on the prevalence and impact of psychiatric disorders in patientswith chronicmedical illnessas well as the prevalence and impact of medical disorders in patients with chronic psychiatric illness.

Persistent andfluctuating anxiety levels in the 18months following acute

myocardial infarction: the role of personality☆,☆☆

Henneke Versteeg, Ph.D. a,b,⁎, Annelieke M. Roest, Ph.D. c, Johan Denollet, Ph.D. a

a Center of Research on Psychology in Somatic diseases (CoRPS), Department of Medical and Clinical Psychology, Tilburg University, P.O. Box 90153, 5000 LE Tilburg, The Netherlandsb Department of Cardiology, University Medical Center Utrecht, P.O. Box 85500, 3508 GA Utrecht, The Netherlandsc University of Groningen, University Medical Center Groningen, Department of Psychiatry, Interdisciplinary Center Psychopathology and Emotion regulation (ICPE),Hanzeplein 1, 9173 GZ Groningen, The Netherlands

a b s t r a c ta r t i c l e i n f o

☆ Funding source: Not applicable.☆☆ Conflicts of interest: Nothing to declare.⁎ Corresponding author. Center of Research on Ps

(CoRPS), Tilburg University, P.O. Box 90153, 5000 LTel.: +31-13-466-2115; fax: +31-13-466-2067.

E-mail addresses: h.versteeg@tilburguniversity.edu (H(A.M. Roest), j.denollet@tilburguniversity.edu (J. Denollet

http://dx.doi.org/10.1016/j.genhosppsych.2014.11.0100163-8343/© 2015 Elsevier Inc. All rights reserved.

Article history:

Received 26 September 2014Revised 7 November 2014Accepted 19 November 2014

Keywords:Myocardial infarctionAnxietyTrajectoriesPersonalityDepression

Objective: To identify the varying courses of anxiety symptoms in thefirst 18months after amyocardial infarction(MI) and to examine the importance of personality in determining elevated anxiety.Methods: Four hundred eighty-six MI patients completed the State-Trait Anxiety Inventory during hospitalizationand at 2-, 12- and 18-months post-MI. At baseline, patients also completed the DS14 Type D personality scale, Anx-iety Sensitivity Index and Beck Depression Inventory, and clinical and sociodemographic information was collected.Results:Growthmixturemodeling analysis identified four anxiety trajectories. Themajority of patients reported sta-ble anxiety scores over time, indicative of either persistent high (17%) or low(71%) anxiety. Patients in the other twosmaller groups initially reported moderate levels of anxiety that fluctuated during follow-up. Type D personality[odds ratio (OR)=5.34; 95% confidence interval (CI): 2.26–12.63], negative affectivity (OR=3.24; 95% CI:1.29–8.14) and anxiety sensitivity (OR=3.35; 95% CI: 1.69–6.62) were themost prominent determinants of persis-

tent high anxiety, independent of depression, sociodemographic and clinical factors.Conclusions: The course of anxiety in the first 18 months after MI is relatively stable for the majority of patients. Pa-tients with Type D personality, negative affectivity and anxiety sensitivity are at an increased risk for persisting el-evated anxiety and should be identified and offered appropriate treatment.

© 2015 Elsevier Inc. All rights reserved.

1. Introduction

In recent years, there has been an increased interest in the study ofanxiety as a risk factor for poor health outcomes in patients with acutemyocardial infarction (MI). Anxiety is prevalent in post-MI patients,with prevalence rates ranging from 20 to 60% and has been associatedwith impaired quality of life, poor adherence to cardiac rehabilitationand an increased risk of new cardiovascular events or mortality [1–5].The general approach to studying anxiety in the context of MI has beento examine prevalence rates or changes in mean anxiety scores overtime. However, such an approach may mask subgroups of patients withvarying anxiety trajectories.

Todate, only one studyhas explored the courseof anxiety in a sampleofMI patients [6]. Van Beek et al. assessed cardiac anxiety in 194 patients andidentified four trajectories: three patient groups had stable anxiety levels

ychology in Somatic diseasesE, Tilburg, The Netherlands.

. Versteeg), a.m.roest@umcg.nl).

during the first year after MI, indicative of high (8%), intermediate (45%)and low(30%) levels of anxiety; the fourthgroup (17%) reportedhigh levelsof anxiety that decreased over time [6]. However, this study used the Cardi-acAnxietyQuestionnaire,whichmeasures specific fears aboutheart diseaseand not general anxiety symptoms. In addition, the studywas conducted ina relatively small sample, with only 15 patients in the stable high anxiousgroup. Hence, itwas impossible to conductmultivariable analyses to exam-ine the independent determinants of high (cardiac) anxiety.

It is particularly important to get insight into the characteristics ofpatients with persistent high anxiety levels, as this might lead to lowerquality of life and poor cardiac prognosis [7]. Previous research hasproposed several associates of elevated anxiety levels after an acutecardiac event, including demographic, clinical and psychological factors[6,8–10]. Regarding psychological factors, these studies have mainlyfocused on the association between anxiety and other psychologicalstates like depression and loneliness,while the importance of personalitytraits like the “distressed” (Type D) personality and anxiety sensitivity ispoorly understood. Better knowledge in this area is essential for thetimely identification of patients at risk for long-lasting distress after MIwho may warrant additional treatment for secondary prevention.

Hence, the objectives of the current prospective cohort study were(a) to identify the trajectories of general anxiety symptoms in the first

2 H. Versteeg et al. / General Hospital Psychiatry 37 (2015) 1–6

18 months after hospitalization for acute MI and (b) to examine therelative importance of individual differences in personality (i.e., TypeD and anxiety sensitivity) in determining high anxiety levels, after ad-justment for depression and sociodemographic and clinical risk factors.

2. Methods

2.1. Study design and participants

Between May 2003 and June 2006, patients hospitalized for acute MIwere included from four hospitals in theNetherlands (CatharinaHospital,Eindhoven; St. Elisabeth Hospital, Tilburg; TweeSteden Hospital, Tilburg;St. Anna Hospital, Geldrop). MI was defined according to the followingcriteria: Troponin I levels more than twice the upper limit, with typicalischemic symptoms (e.g., chest pain), lasting more than 10 min orelectrocardiogram evidence of ST segment elevation or new pathologicalQ-waves. Patientswith significant cognitive impairments (e.g., dementia)or severe comorbidities (e.g., cancer) were excluded. Patients completeda battery of self-reported questionnaires during initial hospitalizationfor MI (baseline) and at 2-, 12- and 18 months post-MI. Demographicand clinical variables were obtained from the medical records or throughpurpose-designed questions. The research protocol was approved by themedical ethics committees of the participating hospitals, and the studywas conducted in accordance with the Helsinki Declaration. Aftercomplete description of the study to the patients, informed consentwas obtained.

2.2. Measures

2.2.1. Demographic and clinical characteristicsDemographic variables included age, gender, marital status, education

level and employment status. Clinical variables included cardiac history[MI, percutaneous coronary intervention (PCI) or coronary artery bypassgraft (CABG) prior to the index MI], left ventricular ejection fraction(LVEF), multivessel disease, anterior location of the index MI, invasive(CABG/PCI) versus conservative treatment for index MI, systolic anddiastolic blood pressure (BP) at time of admission for indexMI, hypercho-lesterolemia (total cholesterol N6.50 mmol/L), hypertension (systolicBPN140mmHg, diastolic BPN90 mmHg), somatic comorbidities [arthritis,diabetes, renal insufficiency, chronic obstructive pulmonary disease(COPD)], smoking status, body mass index (BMI) and participation incardiac rehabilitation after the index MI. In addition, the followingmedications prescribed to the patient at discharge were collected:beta-blockers, angiotensin converting enzyme (ACE) inhibitors, antico-agulants, statins, diuretics, aspirin and selective serotonin reuptakeinhibitors (SSRIs).

2.2.2. Anxiety symptomsAt four time points, symptoms of anxietywere assessed using the State

Trait Anxiety Inventory (STAI) [11]. The STAI is a self-reportmeasure consisting of two 20-item scales developed to measure the levelof general state and trait anxiety. In the current study, we used the Statescale of the STAI to assess fluctuating levels of anxiety. Items are rated ona 4-point Likert-scale from 1 (not at all) to 4 (very much so), with a higherscore indicating a higher level of anxiety (score range: 20–80). A score of≥40 is indicative of clinical levels of anxiety [12]. The STAI has been dem-onstrated to have adequate validity and reliability, with Cronbach's alpharanging from 0.87 to 0.92 [13]. Cronbach's alpha in this study rangedfrom 0.94 to 0.95 at the four time points.

2.2.3. Type D personalityType D personality is defined as the tendency to experience negative

emotions across time and situations paired with the inhibition of theseemotions andhas been consistently related to higher levels of emotionaldistress in cardiovascular patients [14]. At baseline, patients completedthe 14-item Type D scale (DS14) to assess Type D personality. The DS14

consists of two 7-item subscales: negative affectivity (NA) (e.g., “I oftenfeel unhappy”) and social inhibition (SI) (e.g., “I am a closed kind ofperson”) [15]. Items are answered on a 5-point Likert scale ranging from0 (false) to 4 (true), with total scores ranging from 0 to 28 for bothsubscales. To examine the differential predictive value of the compo-nents of Type D, we divided the patients into four groups: (a) scoreb10 on both subscales; (b) score ≥10 on the NA subscale only;(c) score ≥10 on the SI subscale only; (d) score ≥10 on both subscales(Type D) [16]. The DS14 is a valid and reliable scale with Cronbach'salpha of 0.88 and 0.86 for the NA and SI subscales, respectively [15]. Inthe current study, these alpha values were 0.87 and 0.86. The DS14 is astable measure of Type D personality over an 18-month period, andscores are not affected by cardiac disease severity [17].

2.2.4. Anxiety sensitivityAt baseline, the Anxiety Sensitivity Index (ASI) was used to assess

anxiety sensitivity, which refers to the fear of anxiety-related sensationsbased on beliefs that these sensations will have harmful consequencessuch as illness or additional anxiety [18]. It is important to distinguishanxiety sensitivity from state anxiety, with the latter referring to thefrequency of occurrence of anxiety symptoms. Anxiety sensitivity hasbeen related to high levels of distress and lower quality of life in othercardiac patient groups [19,20] but was not yet investigated in MIpatients. The ASI questionnaire comprises 16 items that tap intoconcerns of anxiety-related sensations (e.g., “When I notice that myheart is beating rapidly, I worry that I might have a heart attack.” and“It scares me when I feel faint.”). Respondents indicate their strengthof endorsement for each item on a 5-point Likert scale that rangesfrom 0 (very little) to 4 (very much). Total scores can range from 0 to64, with higher scores reflecting greater sensitivity and fear of anxietysymptoms. The ASI has proven to have a good reliability and validityas a measure of anxiety sensitivity [21]. In our study, the reliability ofthe ASI was also good, with a Cronbach's alpha of 0.91. Since there isno standardized cut-off for the ASI, we used a median split (i.e., ASIN10) to dichotomize ASI scores [20].

2.2.5. Depressive symptoms as potential confoundersTo examinewhether Type D and anxiety sensitivity traits have addi-

tive predictive value in terms of anxiety post-MI, we also controlled forthe potential confounding effect of depressive symptoms at baseline.The Beck Depression Inventory (BDI) is a 21-item self-report question-naire designed to assess the presence and severity of depressive symp-toms [22]. Each item is rated on a 0–3 scale, and a total score is obtainedby adding up all item scores, with a higher score indicatingmore severedepression. A BDI score of≥10 is indicative of at least mild to moderatesymptoms of depression and has been linked to an increased risk forpoor prognosis in cardiac patients [23]. The BDI is a reliable and well-validated measure of depressive symptoms with a Cronbach's alpha of0.81 in nonpsychiatric samples [24]. The Cronbach's alpha in the currentsample was 0.82.

2.3. Statistical analyses

Growth mixture modeling (GMM) with the STAI score as outcomewas used to identify classes of patients with similar course trajectorieson the STAI. In GMM, individual variation on anxiety within classes isallowed and is represented by random effects (Mplus User's Manual,chapter 8, www.statmodel.com). Robust Maximum likelihood estima-tion (MLR) was used to estimate model parameters. Because MLR wasused, all patients with at least the baseline measurement and oneother STAI measurement during follow-up could be included in theGMM analyses (using the Mplus mixture missing functionality). Eachanalysis was run with 250 initial random starts and 50 final stage opti-mizations to avoid local solutions. Analyses started with a one classmodel, and the number of classes was expanded until the best modelfit was found. Model fit was determined using different sources. First,

3H. Versteeg et al. / General Hospital Psychiatry 37 (2015) 1–6

the Bayesian Information Criterion (BIC) was compared across models,with the lowest value indicating the most useful model. In addition, aBootstrapped Likelihood Ratio Test (BLRT) (with 100 bootstrap draws)was performed to test whether adding an extra class led to a significantincrease in model fit compared to the previous model. Analyses wereperformed in Mplus Version 5.2.

The resulting latent classes were compared on demographic, clinicaland psychological characteristics with Pearson chi-square for categoricalvariables and analysis of variance for continuous variables. Finally, a mul-tivariable logistic regression analysis was performed to examine Type Dpersonality traits and anxiety sensitivity as the independent determinantsof a high anxiety level. Based on literature, we decided to include age,gender, low education level, cardiac history, invasive treatment, LVEF,somatic comorbidities, SSRIs and depression as statistical covariates inthe analysis. Since 112 patients had no echocardiographic data, dataimputation based on LVEF means was used to fill in missing LVEF data[25]. All tests were two-tailed and Pb .05 was used to indicate statisticalsignificance. Analyses were performed using SPSS 19.0 for Windows

Table 1Baseline characteristics for the total sample and stratified by anxiety trajectory (N=486)

Baseline characteristics Total sample(N=486)

Persistent, high(N=83)

Fluctua(N=44

PersonalityAnxiety sensitivity, mean (S.D.) 13.45 (10.4) 23.48 (13.2) 16.00 (High anxiety sensitivity (N10) 240 (50) 68 (83) 27 (NA only (NA≥10, SI≤9) 76 (16) 26 (32) 8 (SI only (SI≥10, NA≤9) 115 (24) 9 (11) 4 (Type D (NA and SI≥10) 91 (19) 37 (45) 13 (

Depressive symptomsBDI≥10 115 (24) 46 (55) 14 (

SociodemographicAge, mean (S.D.) 59.94 (11.1) 60.05 (10.3) 59.11 (Sex, female 99 (20) 24 (29) 11 (Partner 404 (83) 60 (74) 37 (Low education levela 216 (44) 50 (60) 15 (Employed 192 (40) 24 (29) 17 (

Disease severityCardiac historyb 76 (16) 18 (22) 10 (LVEF percentage, mean (S.D.) 50.03 (8.94) 50.10 (7.7) 50.51 (Multivessel disease 155 (38) 27 (37) 19 (Anterior MI location 182 (41) 30 (40) 14 (Hypercholesterolemia 49 (10) 7 (8) 7 (Hypertension 89 (22) 14 (20) 6 (

Cardiac functionSystolic BP (mmHg), mean (S.D.) 141.16 (28.4) 135.7 (27.6) 140.1 (Diastolic BP (mmHg), mean (S.D.) 82.39 (16.6) 79.1 (17.1) 82.9 (

ComorbiditiesDiabetes 64 (13) 12 (15) 8 (COPD 36 (8) 9 (11) 6 (Renal failure 21 (4) 6 (7) 1 (Arthritis 39 (8) 8 (10) 6 (

TreatmentInvasive treatmentc 306 (63) 52 (63) 28 (

MedicationB-Blockers 416 (86) 69 (83) 40 (ACE-inhibitors 177 (36) 23 (28) 16 (Anticoagulants 403 (83) 66 (80) 37 (Statins 445 (92) 77 (93) 40 (Aspirin 407 (84) 65 (78) 37 (Diuretics 79 (16) 19 (23) 6 (SSRIs 56 (12) 23 (28) 5 (

Cardiac rehabilitation 308 (63) 42 (51) 32 (Lifestyle

Smoking 175 (36) 36 (44) 22 (BMI, mean (S.D.) 27.00 (3.9) 27.06 (3.8) 26.93 (

⁎ Pb .05.⁎⁎ Pb .01.⁎⁎⁎ Pb .001.

a Secondary school or lower.b MI, PCI or CABG prior to index MI.c CABG or PCI during hospitalization for index MI.

(SPSS Inc., Chicago, IL, USA). In sensitivity analysis, the procedures wererepeated for only those patientswith a STAI score on all assessment times.

3. Results

3.1. Patient sample

Of the 540 patients that consented to participate and completed thebaseline questionnaire, 320 patients had completed the STAI at allassessment times, 166 other patients had a STAI score at baseline andat least one follow-up moment. Hence, 486 (87.8%) patients could beincluded in analyses. The demographic, clinical and psychologicalcharacteristics of these patients are shown in Table 1.

The 486 patients included in the analysis were more likely to bemale (79.6% vs. 55.6%, Pb .001) and to have a partner (84% vs. 72%,P=.03) compared to excluded patients (n=54). Regardingmedication,excluded patients were more likely to be prescribed ACE-inhibitors(52% vs. 37%, P=.03) and diuretics (35% vs. 16%, P=.001) but less likely

ting, high at 12 m)

Fluctuating, high at 18 m(N=12)

Persistent, low(N=347)

P

11.0) 14.80 (9.4) 10.70 (7.8) b .001⁎⁎⁎

62) 7 (58) 138 (40) b .001⁎⁎⁎

18) 2 (17) 40 (12) b .001⁎⁎⁎

9) 1 (8) 101 (30) b .001⁎⁎⁎

30) 3 (25) 38 (11) b .001⁎⁎⁎

32) 2 (17) 53 (15) b .001⁎⁎⁎

9.7) 52.58 (13.1) 60.27 (11.4) .1225) 1 (8) 63 (18) .0984) 9 (75) 298 (86) .0634) 4 (33) 147 (42) .01⁎⁎

39) 8 (67) 143 (41) .05

23) 0 (0) 48 (14) .0710.4) 47.81 (9.7) 50.02 (9.0) .8348) 5 (42) 111 (36) .5638) 3 (27) 135 (43) .7116) 2 (17) 33 (10) .5116) 6 (60) 63 (22) .02⁎

31.6) 153.5 (19.6) 142.1 (28.3) .1919.4) 91.0 (11.6) 82.8 (16.1) .14

18) 1 (8) 43 (12) .6814) 2 (17) 19 (6) .072) 0 (0) 13 (4) .3914) 1 (8) 24 (7) .45

64) 10 (83) 216 (63) .54

91) 12 (100) 295 (86) .3336) 4 (33) 134 (39) .3084) 11 (92) 289 (84) .6691) 12 (100) 316 (92) .7484) 8 (67) 297 (86) .1214) 1 (8) 53 (15) .3112) 2 (17) 26 (7) b .001⁎⁎⁎

73) 7 (58) 227 (65) .03⁎

50) 7 (58) 110 (32) .01⁎⁎

4.7) 29.60 (3.7) 26.91 (3.8) .17

Fig. 1. Trajectories of anxiety 18 months post-MI.

Table 2Associates of persistent high anxiety levels (adjusted analysis)

Baseline characteristics OR [95% CI] P

PersonalityHigh anxiety sensitivity (N10) 3.35 [1.69–6.62] .001⁎⁎

NA only (NA≥10, SI≤9) 3.24 [1.29–8.14] .01⁎

SI only (SI≥10, NA≤9) 1.31 [0.49–3.47] .59Type D (NA and SI≥10) 5.34 [2.26–12.63] b .001⁎⁎⁎

Depressive symptomsBDI≥10 2.25 [1.17–4.34] .02⁎

SociodemographicAge 0.99 [0.96–1.01] .30Sex, female 1.02 [0.48–2.17] .95Low education levela 2.13 [1.14–3.98] .02⁎

ClinicalCardiac historyb 1.60 [0.72–3.58] .25LVEF percentage 1.01 [0.98–1.04] .58Comorbiditiesc 1.42 [0.74–2.71] .30

TreatmentInvasive treatment d 0.85 [0.46–1.58] .61SSRIs 1.71 [0.77–3.81] .19

⁎ Pb .05.⁎⁎ Pb .01.⁎⁎⁎ Pb .001.

a Secondary school or lower.b MI, PCI or CABG prior to index MI.c Diabetes, COPD, renal failure and/or arthritis.d CABG or PCI during hospitalization for index MI.

4 H. Versteeg et al. / General Hospital Psychiatry 37 (2015) 1–6

to use statins (82% vs. 92%, P=.01) compared to included patients. Noother systematic differences were found on baseline characteristicsbetween patients that were included or excluded from analysis.

3.2. Anxiety trajectories

GMM showed that a quadratic model explained the anxiety databetter as compared to a linear model, while fixing variance andcovariance estimates for the slope and quadratic factors to zero.The solution with five classes had the lowest BIC (12412.426) whilethe BLRT was still significant (Pb .001). However, this model includedtwo small classes with relatively little distinction. Therefore, a four-class model was chosen (BIC=12418.938; BLRT: Pb .001).

Thus, we identified four classes of different anxiety trajectoriesduring the first 18 months after MI (Fig. 1). The first group (n=83;17%) had high levels of anxiety during hospitalization for MI (meanSTAI score=51.5±10.0) that persisted during the following 18 months(score range: 50.7–53.1). The second group (n=44; 9%) had moderateanxiety scores (39.2±12.0) at baseline that increased to high levels(52.4±8.4) after the first year but then decreased again to moderatelevels (36.6±7.2) at 18 months post-MI. The third group (n=12; 3%)also reported moderate anxiety scores at enrollment in the study(40.8±12.8), but the scores decreased to low levels (30.7±5.6) afterthe first year and then increased to a high level (51.8±6.5) at 18months. The largest group (n=347; 71%) consisted of patients withpersistent low levels of anxiety (score range: 29.8–35.9) across allassessment times.

Sensitivity analyses without the 166 patients with missing STAI dataidentified the same four trajectories (results not shown).

3.3. Associates of persistent high anxiety

Demographic, clinical and psychological characteristics stratified forthe four anxiety groups are shown in Table 1. Overall, significant differ-ences between the four groups were found on education level (P=.007),hypertension (P=.02), cardiac rehabilitation attendance (P=.03), useof SSRIs (Pb .001), smoking (P=.01) and depression (Pb .001). Thesefour groups were also significantly different in terms of Type D person-ality characteristics (Pb .001) and anxiety sensitivity (Pb .001).

Compared to the other patients (n=403), patients with persistenthigh anxiety levels (n=83) were significantly more likely to be female(29% vs. 19%, P=.03), single (26% vs. 16%, P=.01) and to have a loweducation level (62% vs. 42%, Pb .001) but less likely to be employed

(29% vs. 42%; P=.03). In addition, theywere less likely to attend cardiacrehabilitation (55% vs. 72%, P=.003) and more likely to use SSRIs (28%vs. 8%, Pb .001). Looking at personality profiles, patients with persistenthigh anxiety levels had a higher mean anxiety sensitivity score (23.5±13.2 vs. 11.4±11.4, Pb .001), with more patients scoring N10 points(83% vs. 43%, Pb .001). Regarding the Type D characteristics, they weremore likely to a have a high score on NA only (32% vs. 13%, Pb .001) orto have a Type D personality (45% vs. 14%, Pb .001), but less likelyto have a high score on SI only (11% vs. 27%, P=.002). Patients withpersistent high anxiety levels were also more often depressed at base-line (58% vs. 17%, Pb .001).

Multivariable logistic regression analysis showed that NA [oddsratio (OR)=3.24; 95% confidence interval (CI): 1.29–8.14; P=.01],Type D (OR=5.34; 95% CI: 2.26–12.63; Pb .001), and anxiety sensitivity(OR=3.35; 95% CI: 1.69–6.62; P=.001) were independently associated

5H. Versteeg et al. / General Hospital Psychiatry 37 (2015) 1–6

with persistent high anxiety, also after adjustment for depression (OR=2.25; 95% CI: 1.17–4.34; P=.02) and low education level (OR=2.13;95% CI: 1.14–3.98; P=.02) (Table 2). None of the clinical covariateswere predictive of persistent high anxiety.

In sensitivity analysis without the 166 patients with missing STAIdata, Type D and anxiety sensitivity remained independently associatedwith persistent high anxiety. The associations of NA, depression and loweducation level with anxiety were no longer significant, while the useof SSRIs was positively associated with persistent high anxiety (resultsnot shown).

4. Discussion

To our knowledge, this is the first study that examined the course ofgeneral anxiety symptoms in the first 18 months after acute MI. Fourdistinct anxiety trajectories were identified, with the majority ofpatients reporting stable anxiety scores over time, indicative of eitherpersistent high (17%) or persistent low (71%) anxiety levels. Patientsin the other two groups started with moderate levels of anxietythat fluctuated over time, with one group reporting high anxiety at12months and the other at 18months follow-up. Personality charac-teristics, that is, Type D, NA and anxiety sensitivity, were independentpredictors of persistent high anxiety levels, even after adjustment fordepression and low education level.

Our results emphasize the importance of going beyond prevalencerates and changes in mean anxiety scores over time, as this masks thenatural heterogeneity between patients. Previous studies on anxiety incardiac patients have also identified multiple trajectories instead ofone, although the number and course of these trajectories may differ.Our finding that anxiety levels are relatively stable over time in thema-jority of patients is in accordance with the results of a study examiningcardiac anxiety in MI patients and most [6,8,10,26], but not all [9,27],studies in other groups of patients with coronary artery disease. Overall,evidence indicates thatwhen present, elevated symptomsof anxiety arenot likely to remit spontaneously but remain for a long time after acardiac event.

Besides the patients with a persistently high anxiety score (meanscore ≥50), the two groups of patients with fluctuating anxiety levelsare of interest as they reported equally high STAI scores as the highanxiety group at some point in time, either 12 or 18 months post-MI.At baseline, patients in both these groups scored around the STAI cut-off value of 40 (39–41), which confirms that in-hospital anxiety assess-ment is valuable for risk stratification. From our results, it could berecommended that patients with a STAI-score of ≥40 during initialhospitalization for MI should be closely monitored during the first2 years after the event. The group of patients with increasing anxietylevels between 12- and 18 months post-MI was too small (N=12) toexamine in multivariable analysis, but from Table 1 it seems that thesepatients are, in general, young, employed males with an unhealthy life-style (i.e., hypertension, smoking, overweight). As this is the first studylooking at anxiety trajectories lasting more than 1 year after acute MI,larger prospective studies are needed to examine long-term fluctuationin anxiety levels.

In line with other studies on anxiety trajectories, psychosocialfactors were more strongly related to anxiety levels than the clinicalcharacteristics of patients [6,8–10,20,27,28], with Type D personalitybeing the strongest associate of persistent high anxiety. This is in accor-dance with earlier studies in cardiac patients showing that Type D per-sonality is an important risk factor for anxiety, also when adjusting fordepression [29,30]. In addition, the Type D personality profile hasbeen related to impaired mental and physical health status and poorcardiac prognosis, independent of depression and clinical factors[14,31–33]. In addition, Type D characteristics can be easily assessedwith the brief DS14 scale, suggesting that Type D screening may befeasible and efficient in clinical practice. It is of interest that regardingthe constituting components of Type D, NA was a major determinant

of anxiety in itself, while SI alone was not. However, the risk of persis-tent anxiety was most prominent in Type D patients, characterized bythe combination of NA and SI.

Independent of Type D characteristics, patients scoring high onanxiety sensitivityweremore likely to remain anxious afterMI. Patientswith high levels of anxiety sensitivity respondwith fear to bodily sensa-tions like heart rate changes, sweaty palms and shallow breathing,because they believe that these symptoms may have harmful conse-quences [18]. Anxiety sensitivity is a relatively stable personality factorand has been associated with the development and maintenance ofanxiety disorders and other mood disorders and chronic health condi-tions in the general population [34,35]. Anxiety sensitivity is mostlyneglected in cardiac studies, but it has been related to persistent highlevels of patient-reported distress and somatic symptoms in patientswith an implantable cardioverter defibrillator and lower quality of lifeand greater symptom severity in patients with atrial fibrillation[19,20]. In our sample, the mean ASI score (13.4±10.4) was lowerthan that reported for patients with various chronic health conditions(range: 14.1–34.8) and patients with atrial fibrillation (23.3±10.0)[19,34], which indicates that post-MI patients are not by definitionmore fearful of somatic sensations. Our results emphasize that individualdifferences in personality are of importance in determining the course ofdistress after MI and should be recognized in future studies and clinicalpractice. This is particularly important as (persistent) anxiety has beenassociated with lower quality of life and poor prognosis in cardiacpatients [2,3,7,36].

Cardiac rehabilitation and exercise training are proven to be benefi-cial for reducing anxiety and depression levels and improving qualitylife [36–38], but paradoxically, the current and previous studies indicatethat highly anxious or depressed patients are less likely to enter theseprograms or tend to drop out at an early stage [5,39]. In addition, TypeD personality has been associated with impaired psychological andhealth status outcomes before and after cardiac rehabilitation [40,41].These results underline the importance of identifying patients withpsychological risk factors at an early stage and offering them additionalinterventions to manage their distress. For psychological interventionsto be successful in patients at high risk due to their personality profile,the intervention should likely not be aimed at changing personality,but rather at modifying the cognitive and behavioral underpinnings,for example by cognitive behavioral therapy (CBT). Central to CBT isthe identification of maladaptive automatic thoughts and restructuringthem into more realistic and adaptive cognitions. The Enhancing Re-covery in Coronary Heart Disease patients trial and the ExhaustionIntervention Trial have shown that CBT significantly reduces emotionaldistress and improves social support in coronary heart disease patients[42,43]. CBT may also be particularly effective in depressed and Type Dpatients as it targets a broad spectrum of negative emotions and socialdifficulties [44,45]. In addition, a meta-analytic review of randomizedcontrolled trials showed that CBT is effective in reducing anxietysensitivity in individuals with (a risk for developing) mood disorders[46]. Future studies on effective treatment for (preventing) distress incardiac patients with a vulnerable personality profile are essential.

The current study has some limitations that should be discussed. First,the 54 patients that dropped out of the study in the first 2 months post-MIweremore often female, single and usedmoremedications comparedwith the patients included in the analyses, which might jeopardize thegeneralizability of our results. However, by using the mixture missingapproach, we prevented the exclusion of 166 other patients whichwould have introduced a more substantial bias to our study. Of note,sensitivity analyses excluding these 166patients didnot changeour over-all conclusions. Second, the time period between the second (2 months)and third (12 months) anxiety assessment was relatively long, and wecannot make strong statements about the course of anxiety during thatyear. However, the majority of evidence indicates that distress levelsremain relatively stable in the first year after a cardiac event. Third, aswe were interested in the predictive value of patient characteristics

6 H. Versteeg et al. / General Hospital Psychiatry 37 (2015) 1–6

during hospitalization of MI, we did not examine the effect of changesin clinical factors (e.g., recurrent MI) or psychological treatment duringthe follow-up period, but these may have affected anxiety trajectories.Yet, the fact that the anxiety levels were stable over time for themajorityof patients suggests these factors are of limited influence. Strengthsof this study include the relatively large sample of post-MI patientsonly, a follow-up period that extends beyond 1 year and the inclusionof psychological traits instead of states as potential determinants ofanxiety trajectories.

In conclusion, the course of anxiety in the first 18months after MI isrelatively stable for the majority of patients. Personality characteristicsincluding Type D personality, high NA and high anxiety sensitivitywere the strongest associates of persisting high anxiety levels. Clinicalcharacteristics, including disease severity, comorbidities and treatment,were not associated with anxiety. These results indicate that the assess-ment of anxiety levels and/or personality characteristics early afterMI isof additional value in clinical practice and may help to identify patientsat risk for long-lasting distress. Further research is warranted to find themost beneficial and cost-effective ways to treat these patients.

Acknowledgments

We like to thank Dr. Klaas Wardenaar for his help with thestatistical analyses.

References

[1] Roest AM, Martens EJ, Denollet J, de Jonge P. Prognostic association of anxiety postmyocardial infarction with mortality and new cardiac events: a meta-analysis.Psychosom Med 2010;72:563–9.

[2] Benyamini Y, Roziner I, Goldbourt U, Drory Y, Gerber Y. Israel Study Group on FirstAcuteMyocardial Infarction. Depression and anxiety followingmyocardial infarctionand their inverse associations with future health behaviors and quality of life. AnnBehav Med 2013;46:310–21.

[3] Roest AM, Heideveld A, Martens EJ, de Jonge P, Denollet J. Symptom dimensions ofanxiety following myocardial infarction: associations with depressive symptomsand prognosis. Health Psychol 2013;33:1468–76.

[4] Strik JJ, Denollet J, Lousberg R, Honig A. Comparing symptoms of depression andanxiety as predictors of cardiac events and increased health care consumptionafter myocardial infarction. JACC 2003;42:1801–7.

[5] McGrady A, McGinnis R, Badenhop D, Bentle M, Rajput M. Effects of depression andanxiety on adherence to cardiac rehabilitation. J Cardiopulm Rehabil Prev 2009;29:358–64.

[6] Van Beek MH, Mingels M, Voshaar RC, van Balkom AJ, Lappenschaar M, Pop G, et al.One-year follow up of cardiac anxiety after a myocardial infarction: a latent classanalysis. J Psychosom Res 2012;73:362–8.

[7] Moser DK, McKinley S, Riegel B, Doering LV, Meischke H, Pelter M, et al. Relationshipof persistent symptoms of anxiety to morbidity and mortality outcomes in patientswith coronary heart disease. Psychosom Med 2011;73:803–9.

[8] Murphy B, Ludeman D, Elliott P, Judd F, Humphreys J, Edington J, et al. Red flagsfor persistent or worsening anxiety and depression after an acute cardiac event: a6-month longitudinal study in regional and rural Australia. Eur J Prev Cardiol2014;21:1079–89.

[9] Murphy BM, Elliott PC, Worcester MU, Higgins RO, Le Grande MR, Roberts SB, et al.Trajectories and predictors of anxiety anddepression inwomen during the 12monthsfollowing an acute cardiac event. Br J Health Psychol 2008;13:135–53.

[10] Pedersen SS, Smith OR, De Vries J, Appels A, Denollet J. Course of anxiety symptomsover an 18-month period in exhausted patients post percutaneous coronary inter-vention. Psychosom Med 2008;70:349–55.

[11] Spielberger C, Gorsuch R, Lushene R, Vagg P, Jacobs G. Manual for the State-TraitAnxiety Inventory (Form Y). Palo Alto: Press CP; 1983.

[12] Bunevicius A, Staniute M, Brozaitiene J, Pop VJ, Neverauskas J, Bunevicius R. Screeningfor anxiety disorders in patients with coronary artery disease. Health Qual LifeOutcomes 2013;11:37.

[13] Van der Ploeg H, Defares P, Spielberger C. Handleiding bij de zelf-beoordelingsvragenlijst: een Nederlandse bewerking van de Spielberger State-Trait AnxietyInventory. Lisse: Swets & Zeitlinger; 1980.

[14] Pedersen SS, Denollet J. Is type D personality here to stay? Emerging evidence acrosscardiovascular disease patient groups. Curr Cardiol Rev 2006;2:205–13.

[15] Denollet J. DS14: standard assessment of negative affectivity, social inhibition, andtype D personality. Psychosom Med 2005;67:89–97.

[16] EmonsWH,Meijer RR, Denollet J. Negative affectivity and social inhibition in cardio-vascular disease: evaluating type D personality and its assessment using itemresponse theory. J Psychosom Res 2007;63:27–39.

[17] Martens EJ, Kupper N, Pedersen SS, Aquarius AE, Denollet J. Type-D personality is astable taxonomy in post-MI patients over an 18-month period. J Psychosom Res2007;63:545–50.

[18] Reiss S, Peterson RA, Gursky DM, McNally RJ. Anxiety sensitivity, anxiety frequencyand the prediction of fearfulness. Behav Res Ther 1986;24:1–8.

[19] Ong L, Cribbie R, Harris L, Dorian P, Newman D, Mangat I, et al. Psychological corre-lates of quality of life in atrial fibrillation. Qual Life Res 2006;15:1323–33.

[20] Van den Broek KC, Kupper N, Van der Voort PH, Alings M, Denollet J, Nyklicek I.Trajectories of perceived emotional and physical distress in patients with animplantable cardioverter defibrillator. Int J Behav Med 2014;21:149–59.

[21] Zinbarg RE,Mohlman J, HongNN. Dimensions of anxiety sensitivity. In: Taylor S, editor.Anxiety sensitivity: theory, research, and treatment of the fear of anxiety. New Jersey:Lawrence Erlbaum Associates; 1999. p. 83–114.

[22] Beck AT, Ward CH, Mendelson M, Mock J, Erbaugh J. An inventory for measuringdepression. Arch Gen Psychiatry 1961;4:561–71.

[23] Frasure-Smith N, Lesperance F. Reflections on depression as a cardiac risk factor.Psychosom Med 2005;67(Suppl. 1):S19–25.

[24] Beck AT, Steer RA, Garbin MC. Psychometric properties of the Beck DepressionInventory twenty-five years of evaluation. Clin Psychol Rev 1988;8:77–100.

[25] Martens EJ, Mols F, Burg MM, Denollet J. Type D personality predicts clinical eventsafter myocardial infarction, above and beyond disease severity and depression. J ClinPsychiatry 2010;71:778–83.

[26] Grace SL, Abbey SE, Irvine J, Shnek ZM, Stewart DE. Prospective examination ofanxiety persistence and its relationship to cardiac symptoms and recurrentcardiac events. Psychother Psychosom 2004;73:344–52.

[27] Murphy BM, Elliott PC, Higgins RO, Le Grande MR, Worcester MU, Goble AJ, et al.Anxiety and depression after coronary artery bypass graft surgery: most get better,some get worse. Eur J Cardiovasc Prev Rehabil 2008;15:434–40.

[28] Pedersen SS, Theuns DA, Jordaens L, Kupper N. Course of anxiety and device-relatedconcerns in implantable cardioverter defibrillator patients the first year post implan-tation. Europace 2010;12:1119–26.

[29] Schiffer AA, Pedersen SS, Broers H,Widdershoven JW, Denollet J. Type-D personalitybut not depression predicts severity of anxiety in heart failure patients at 1-yearfollow-up. J Affect Disord 2008;106:73–81.

[30] SpindlerH, Pedersen SS, Serruys PW, ErdmanRA, vanDomburgRT. Type-Dpersonalitypredicts chronic anxiety following percutaneous coronary intervention in the drug-eluting stent era. J Affect Disord 2007;99:173–9.

[31] Versteeg H, Spek V, Pedersen SS, Denollet J. Type D personality and health status incardiovascular disease populations: a meta-analysis of prospective studies. Eur JPrev Cardiol 2012;19:1373–80.

[32] Schiffer AA, Smith OR, Pedersen SS,Widdershoven JW, Denollet J. Type D personalityand cardiac mortality in patients with chronic heart failure. Int J Cardiol 2010;142:230–5.

[33] Denollet J, Pedersen SS. Prognostic value of type D personality compared withdepressive symptoms. Arch Int Med 2008;168:431–2.

[34] Asmundson G, Wright K, Hadjistavropoulos H. Anxiety sensitivity and disablingchronic health conditions. Scand J Behav Ther 2000;29:100–17.

[35] Naragon-Gainey K. Meta-analysis of the relations of anxiety sensitivity to thedepressive and anxiety disorders. Psychol Bull 2010;136:128–50.

[36] Hofer S, BenzerW, Alber H, Ruttmann E, Kopp M, Schussler G, et al. Determinants ofhealth-related quality of life in coronary artery disease patients: a prospective studygenerating a structural equation model. Psychosomatics 2005;46:212–23.

[37] Lavie CJ, Milani RV. Prevalence of anxiety in coronary patients with improvementfollowing cardiac rehabilitation and exercise training. Am J Cardiol 2004;93:336–9.

[38] Yohannes AM, Doherty P, Bundy C, Yalfani A. The long-term benefits of cardiacrehabilitation on depression, anxiety, physical activity and quality of life. J ClinNurs 2010;19:2806–13.

[39] Jackson JL, Emery CF. Emotional distress, alexithymia, and coping as predictors ofcardiac rehabilitation outcomes and attendance. J Cardiopulm Rehabil Prev 2013;33:26–32.

[40] Pelle AJ, Erdman RA, van Domburg RT, Spiering M, Kazemier M, Pedersen SS. Type Dpatients report poorer health status prior to and after cardiac rehabilitationcompared to non-type D patients. Ann Behav Med 2008;36:167–75.

[41] Sogaro E, Schinina F, Burgisser C, Orso F, Pallante R, Aloi T, et al. Type D personalityimpairs quality of life, coping and short-term psychological outcome in patientsattending an outpatient intensive program of cardiac rehabilitation. Monaldi ArchChest Dis 2010;74:181–91.

[42] Appels A, Bar F, van der Pol G, Erdman R, Assman M, Trijsburg W, et al. Effects oftreating exhaustion in angioplasty patients on new coronary events: results of therandomized Exhaustion Intervention Trial (EXIT). PsychosomMed 2005;67:217–23.

[43] Berkman LF, Blumenthal J, BurgM, Carney RM, Catellier D, CowanMJ, et al. Effects oftreating depression and low perceived social support on clinical events after myo-cardial infarction: the Enhancing Recovery in Coronary Heart Disease Patients(ENRICHD) Randomized Trial. JAMA 2003;289:3106–16.

[44] Pelle AJ, Van den Broek KC, Denollet J. Interventions in the context of the distressed(type D) personality. In: Dornelas E, editor. Stress proof the heart - Behavioral inter-ventions for cardiac patients. New York: Springer; 2012. p. 167–98.

[45] Dickens C, Cherrington A, Adeyemi I, Roughley K, Bower P, Garrett C, et al. Characteris-tics of psychological interventions that improve depression in people with coronaryheart disease: a systematic review and meta-regression. Psychosom Med 2013;75:211–21.

[46] Smits JA, Berry AC, Tart CD, Powers MB. The efficacy of cognitive-behavioral inter-ventions for reducing anxiety sensitivity: a meta-analytic review. Behav Res Ther2008;46:1047–54.