Professional Documents
Culture Documents
Kendel 2011
Kendel 2011
CABG ⫽ coronary artery bypass graft; PF ⫽ physical functioning; with male patients after surgery (5–7). Yet, the role of gender
RP ⫽ role functioning; BP ⫽ bodily pain; GH ⫽ general health; over time does not seem consistent if baseline differences in
SF-36 ⫽ Medical Outcomes Study 36-Item Short Form Health Survey. HRQoL or clinical health parameters are taken into account.
Some studies (8 –10) reported similar trajectories in physical
INTRODUCTION
aspects of HRQoL, despite lower baseline values in women.
RESULTS
Unmatched Sample
Figure 1. Flowchart of participants.
In the unmatched sample, women were less educated, were more
often living alone, and had higher levels of depression. Men and
Measures
women differed on a variety of clinical parameters (Table 1; Sup-
Comorbidities, clinical risk factors, medications, intraoperative data, and
laboratory parameters were extracted from case report forms. Sociodemographic
plemental Digital Content 1, http://links.lww.com/PSYMED/A31).
and psychosocial data were collected through self-report questionnaires. Quality Compared with men, women were older, had more edema and a
of life was assessed with the Medical Outcomes Study 36-Item Short Form more frequent history of hypertension, syncope, hepatitis, and
Health Survey (SF-36) (12) (German version: Bullinger) (13). The SF-36 pro- disease of the thyroid gland. They also had lower hemoglobin
vides four aspects of physical HRQoL: 1) physical functioning (PF) (limitations levels and received calcium-channel blockers more frequently.
in performing physical activities, including bathing or dressing); 2) role function-
ing (RP) (problems with work or other daily activities as a result of physical Men suffered more often from hyperuricemia and renal failure;
health); 3) bodily pain (BP) (limitations due to pain); and 4) general health they had higher cholesterol levels, a lower left ventricular ejec-
perception (GH) (evaluation of personal health). The scores range from 0 (severe tion function, and more frequently got angiotensin-converting
impairment) to 100 (no impairment). The intercorrelations of these subscales enzyme inhibitors. Other gender differences pertained to the
ranged from r ⫽ .25 to r ⫽ .49 with no significant gender differences between the
localization and number of grafts.
correlation coefficients. Depressive symptoms were determined by means of the
nine-item module of the Patient Health Questionnaire (14) (German version:
Löwe et al.) (15), which yields a score ranging from 0 (no depression) to 27
(maximum level of depressive symptoms). Correlations between depression and Propensity Score Matching
HRQoL subscales ranged between r ⫽ .31 and r ⫽ .44. Again, there were no
Figure 2 shows the distribution of propensity scores
gender differences between correlation coefficients. Partner status was treated as
a dichotomous variable: living with a partner versus no partner (single, divorced, for women and men, based on clinical characteristics
widowed, separated). A three-level categorical variable was created to indicate (Supplemental Table 1/Supplemental Digital Content 1
the different levels of (vocational) education: 1) no vocational training; 2) ap- http://links.lww.com/PSYMED/A31).
prenticeship; and 3) academic. The large overlap of propensity scores allowed for match-
ing a high percentage of women. Of 198 women, 157 (79.3%)
Propensity Score Matching
could be matched to 157 (19.8%) of 792 men. Clinical char-
In nonrandomized studies, direct comparisons of two distinct groups may
be misleading because the groups may differ systematically. To obtain a
acteristics of the matched sample are shown in Table 2 and
comparable distribution of clinical variables among men and women, we used Supplemental Table 2 (Supplemental Digital Content 2
the propensity score matching technique. Rosenbaum and Rubin (16) showed http://links.lww.com/PSYMED/A32), respectively. After match-
that this technique allows for considering a high number of confounding ing, there were no significant gender differences in any matching
variables and that adjustment for a balancing propensity score is sufficient to variables. An excellent matching balance was also confirmed by
produce unbiased estimates of the average group effects.
The propensity score was calculated, using logistic regression. Sixty-five
a decrease in effect sizes in the matched sample as compared with
perioperative characteristics were used as predictors to estimate the probabil- the unmatched sample.
ity of each patient being female. Greedy matching techniques were applied to To determine the influence of psychosocial variables inde-
select male counterparts to the female patients by choosing the patient with pendent from the clinical health status, men and women in the
the nearest propensity score, i.e., a distance of ⱕ0.25 of the standard deviation sample were not matched for education, partner status, and
of the propensity score for each pair (16).
depression. However, the clinical variable matching removed
Statistical Analysis the gender difference in education (p ⫽ .12). The gender
In the unmatched sample, gender differences in baseline characteristics were
difference in partner status was still significant (p ⬍ .001), and
analyzed, using the 2 test for categorical variables and the Student’s t test for depressive symptoms were still more prevalent in women than
continuous variables. To test the balance of matching variables, the gender in men (p ⬍ .001) (Table 2).
Age, years (mean ⫾ SD) 69.89 ⫾ 9.47 65.98 ⫾ 8.08 ⬍.001 0.44
Education ⱕ9 years (%)a 59.9 45.8 ⬍.001 0.11
Living alone (%)a 52.5 15.7 ⬍.001 0.35
Depression (mean ⫾ SD)a 7.03 ⫾ 4.7 5.25 ⫾ 4.1 ⬍.001 0.40
Body mass index (mean ⫾ SD) 27.96 ⫾ 5.24 27.79 ⫾ 4.09 .67 0.04
LVEF (mean ⫾ SD) 58.39 ⫾ 11.79 55.31 ⫾ 13.83 .002 0.24
Dyspnea on exertion (%) 83.8 70.1 ⬍.001 0.12
Diabetes (%) 39.4 33.0 .09 0.05
Hypertension (%) 94.4 84.8 ⬍.001 0.11
History of myocardial infarction (%) 40.4 43.7 .40 0.03
Current smoking (%) 14.6 18.6 .20 0.04
Renal failure (%) 9.1 14.6 .04 0.07
Previous CABG (%) 5.6 5.4 .94 0.00
Procedure priority: emergent (%) 19.2 24.6 .11 0.05
Number of grafts (mean ⫾ SD) 2.44 ⫾ 1.07 2.89 ⫾ 0.99 ⬍.001 ⫺0.18
HRQoL
Physical functioning (mean ⫾ SD)a 42.5 ⫾ 25.57 58.11 ⫾ 25.94 ⬍.001 ⫺0.61
Role functioning (mean ⫾ SD)a 22.72 ⫾ 34.94 40.25 ⫾ 41.89 ⬍.001 ⫺0.45
Bodily pain (mean ⫾ SD)a 52.92 ⫾ 28.42 60.4 ⫾ 28.81 ⬍.001 ⫺0.26
General health status (mean ⫾ SD)a 49.64 ⫾ 17.67 53.71 ⫾ 17.21 .003 ⫺0.23
Mean differences analyzed with Student t tests (effect size: d); categorical variables with 2 test (effect size: 2).
a
Variables not included as matching variables.
SD ⫽ standard deviation; LVEF ⫽ left ventricular ejection fraction; CABG ⫽ coronary artery bypass graft; HRQoL ⫽ health-related quality of life.
DISCUSSION
Preoperative Quality of Life Our results indicate that there are substantial gender dif-
In the unmatched sample, the gender differences in all ferences in physical HRQoL both before and after CABG,
physical HRQoL subscales were pronounced (Table 1). with women being more impaired than men in several HRQoL
Matching for 65 baseline and perioperative characteristics subdomains. Interestingly, after matching for a large bundle of
diminished the gender differences (Table 2). However, pro- clinical variables, the gender differences in physical function-
pensity-matched men and women still differed markedly in PF ing and role functioning decreased but still remained signifi-
(p ⬍ .001; d ⫽ 0.43) and RP (p ⫽ .007; d ⫽ 0.32). cant. Nevertheless, both men and women do experience a
comparable improvement in their HRQoL.
Gender Differences in Postoperative HRQoL Our findings are consistent with another study (9), which
Repeated-measures analysis of variance revealed significant had also found higher HRQoL scores in men and reported
main effects of time and gender on HRQoL (Fig. 3, a–d). The similar improvements over the first year post CABG in both
TABLE 2. Differences Among Propensity-Matched Men and Women in Baseline Characteristics and Perioperative Variables (n ⴝ 314)
Age, years (mean ⫾ SD) 68.88 ⫾ 9.25 68.97 ⫾ 7.83 .80 ⫺0.01
Education ⱕ9 years (%)a 57.3 48.4 ⬍.001 0.09
Living alone (%)a 51.0 15.3 ⬍.001 0.38
Depression (mean ⫾ SD)a 7.09 ⫾ 4.79 5.42 ⫾ 3.56 ⬍.001 0.40
Body mass index (mean ⫾ SD) 28.06 ⫾ 5.12 27.66 ⫾ 3.84 .57 0.09
LVEF (mean ⫾ SD) 57.71 ⫾ 12.30 58.63 ⫾ 13.54 .44 ⫺0.07
Dyspnea on exertion (%) 81.5 78.3 .56 0.04
Diabetes (%) 36.9 38.2 .91 0.01
Hypertension (%) 93.0 92.4 1.00 0.01
History of myocardial infarction (%) 45.9 40.8 .37 0.05
Current smoking (%) 15.9 20.4 .40 0.06
Renal failure (%) 9.6 10.8 .85 0.02
Previous CABG (%) 5.7 5.1 1.00 0.01
Procedure priority: emergent (%) 20.3 23.6 .58 0.07
Number of grafts (mean ⫾ SD) 2.57 ⫾ 1.02 2.55 ⫾ 0.98 .91 0.06
HRQoL
Physical functioning (mean ⫾ SD)a 44.8 ⫾ 26.05 55.95 ⫾ 25.59 ⬍.001 ⫺0.43
Role functioning (mean ⫾ SD)a 24.1 ⫾ 35.84 35.88 ⫾ 38.83 .007 ⫺0.32
Bodily pain (mean ⫾ SD)a 54.1 ⫾ 29.24 58.56 ⫾ 28.48 .19 ⫺0.15
General health status (mean ⫾ SD)a 50.47 ⫾ 18.6 52.57 ⫾ 17.98 .30 ⫺0.11
Mean differences analyzed with paired t tests (effect size: d); categorical variables with McNemar test (effect size: 2).
a
Variables not included as matching variables.
SD ⫽ standard deviation; LVEF ⫽ left ventricular ejection fraction; CABG ⫽ coronary artery bypass graft; HRQoL ⫽ health-related quality of life.
Figure 3. a–d) Results from repeated-measures analyses of variance: changes in health-related quality of life subscale scores, baseline to 1 year in
propensity-matched men (n ⫽ 157) and women (n ⫽ 157). Range, 0 (severe impairment) to 100 (no impairment). Over 1 year, men and women show a
comparable improvement in physical aspects of health-related quality of life.
genders. The same result was found in two studies with a the present study, the nonsignificance of the gender difference
follow-up of 6 months (10) and 2 years (17), respectively. In may be due to the small number of only 22 women included
contrast, no gender differences in quality of life after 1 year in this study. In our study, the strongest improvements in
were reported by Hunt and colleagues (18). However, given HRQoL in both genders were in PF, RP, and BP. Changes in
the absolute difference in SF-36 scores that equals the one of GH were less marked. It may be assumed that positive
TABLE 3. Results From Multiple Regression Analysis: Effects of knowledge, there is only one other study (6) that used the pro-
Depression and Gender on Health-Related Quality of Life (HRQoL)a pensity matching technique in this context. However, this was
1 Year After Surgery in the Matched Sample After Adjustment for
only an additional analysis with a limited explanatory power.
Baseline HRQoL (n ⴝ 314)
Women are on average older at the time of the surgery and
Baseline Covariate  95% CI p have more comorbid illnesses and risk factors. Thus, their initial
situation is less favorable than that of men, which might account
Model 1:b Effects of depression for their reporting a worse HRQoL. As expected, propensity
and gender on the change in
physical functioning matching for a multitude of clinical variables diminished the
Baseline physical functioning 0.48 0.39–0.61 ⬍.001 gender difference in all domains of physical HRQoL. However,
Depression ⫺0.13 ⫺1.47–⫺0.17 .01 the differences in PF and RP were still significant, which indi-
Female gender ⫺0.04 ⫺7.92–3.28 .42 cates that women report worse HRQoL than men, independent
Model 2:b Effects of depression
from baseline health status and perioperative variables. It is
and gender on the change in
role functioning known that there are also gender differences in HRQoL in the
Baseline role functioning 0.26 0.18–0.43 ⬍.001 German norm sample (age group, 61–70 years) (15). This may be
Depression ⫺0.17 ⫺2.79–⫺0.59 .003 partially accounted for by the tendency of women to report a
Female gender ⫺0.11 ⫺19.62–⫺0.12 .047 worse health status, even independently of their actual health
Model 3:b Effects of depression
(27). However, although the gender difference in BP and GH in
and gender on the change in
bodily pain both the healthy norm sample (effect sizes d ⬍0.07) and our
Baseline bodily pain 0.21 0.09–0.34 .001 propensity-matched sample can be classified as small and did not
Depression ⫺0.19 ⫺2.11–⫺0.47 .002 differ considerably among the two samples, the gender difference
Female gender ⫺0.09 ⫺12.38–1.50 .12 in PF and in RP was markedly larger in our sample compared
Model 4:b Effects of depression
with the norm sample (d ⫽ 0.11 and d ⫽ 0.01 in the norm
and gender on the change in
general health perception sample). It seems possible that, particularly for women, the
Baseline general health 0.53 0.43–0.64 ⬍.001 surgery constitutes a stress factor that enlarges the gender differ-
perception ence already known from the healthy population.
Depression ⫺0.04 ⫺0.65–0.28 .43
Female gender ⫺0.06 ⫺6.09–1.64 .26 Role of Depression
a
Range, 0 –100; higher values indicating better HRQoL. To explain the persisting gender differences in HRQoL, we
b
All models adjusted for education and partner status. performed analyses on the role of depression. Depressive symp-
toms were highly predictive of all four components of physical
changes caused by the surgery, e.g., an improvement in pump- HRQoL, except for GH. Depression is closely related to HRQoL
ing capacity and dyspnea, are most likely to be reflected in PF, and overlaps with it to some extent, but the concepts should be
RP, and BP, whereas the perception of GH is more closely distinguished. HRQoL is a measure of a patient’s subjective
associated with comorbid conditions. perception of the functional effect of an illness and its therapy
Other studies (6,7,11,19 –22) do not support the finding of a (28), whereas depression is an affective disorder defined by
similar pace of recovery but rather found a poorer recovery in specific symptoms that may occur in reaction to an illness but
terms of HRQoL in women. In a study of 184 men and 96 may also exist independently from health status. The association
women (7), women did not have the same degree of improve- of presurgical depressive symptoms and HRQoL observed in the
ment after 1 year compared with men. In particular, women were present study is in accord with previous studies, which have
at greater risk for diminished work-related activities. Similarly, in shown the effect of depressive symptoms on HRQoL in patients
one study with a follow-up of 6 months (20) and in another study with coronary artery disease (29) and patients undergoing CABG
with a follow-up of 3 months (5) that matched 40 men and 40 (30,31). Because depression rates in women are double those of
women for age and body surface area, female gender was an men (32), it is probable that depression functions as a mediator
independent predictor for impaired HRQoL after CABG. Finally, between gender and HRQoL.
Peric et al. (23) identified female gender as an independent As expected, adjustment for depressive symptoms reduced the
predictor for the subscale BP. However, the explanatory power gender difference in both PF and BP, although women still
and validity of several studies (18,19,24,25) might be limited by reported significantly worse RP. In the SF-36, RP is represented
either small sample sizes and/or different time frames. For ex- by items such as “accomplishing less than wanted” as a result of
ample, the findings of Vaccarino et al. (11) and Artinian and one’s physical health, or experiencing “difficulty in performing
Duggan (19) referred to a time frame of only 6 weeks to 8 weeks daily activities.” The finding that women feel more restricted in
after surgery. Shortly after surgery, the traditional role demands their daily activities than men after CABG is supported by
may be particularly stressful for women and may thus enhance previous studies with patients recovering from myocardial infarc-
the disparities in HRQoL (26). Moreover, studies differ in their tion (33). One explanation may be that women in this age cohort
more or less systematic and exhausting inclusion of clinical are particularly challenged by resuming their traditional house-
variables. One strength of the present study is the accurate control hold activities soon after discharge from the hospital. Hildingh et
of variables which are possibly confounded with gender. To our al. (34) concluded from a meta-synthesis of seven qualitative
studies that, after surgery, women thought it was essential to 10. Sawatzky JA, Naimark BJ. The coronary artery bypass graft surgery trajectory:
protect their traditional roles, which they may feel are important gender differences revisited. Eur J Cardiovasc Nurs 2009;8:302–8.
11. Vaccarino V, Lin ZQ, Kasl SV, Mattera JA, Roumanis SA, Abramson JL,
in preserving their identity. However, time spent on household Krumholz HM. Gender differences in recovery after coronary artery
activities is not necessarily a predictor for worse quality of life in bypass surgery. J Am Coll Cardiol 2003;41:307–14.
women (35). 12. Ware JE, Snow KK, Kosinski M. SF-36 Health Survey Manual and
Interpretation Guide. Boston: New England Medical Center, The Health
Several limitations in the study should be noted. First, the Institute; 1993.
measurement of HRQoL is not disease specific. However, the 13. Bullinger M. German translation and psychometric testing of the SF-36
global instrument allows our results to be compared not only Health Survey: preliminary results from the IQOLA project. International
Quality of Life Assessment. Soc Sci Med 1995;41:1359 – 66.
with outcomes in other diseases but also with outcomes for the 14. Spitzer RL, Kroenke K, Williams JB. Validation and utility of a self-
general population. Second, the dropout rate might have given report version of PRIME-MD: the PHQ primary care study. Primary Care
biased results. Third, this is a monocentric study conducted in Evaluation of Mental Disorders. Patient Health Questionnaire. JAMA
1999;282:1737– 44.
a center with an outstanding reputation. We cannot exclude 15. Löwe B, Spitzer RL, Zipfel S, Herzog W. Patient Health Questionnaire
the possibility that patients with more severe diseases may (PHQ), German Version, Manual and Materials. 2nd ed. Karlsruhe,
have been referred to this hospital. Germany: Pfizer; 2002.
16. Rosenbaum P, Rubin D. The central role of the propensity score in
CONCLUSION observational studies for causal effects. Biometrika 1983;70:41–55.
17. Sjoland H, Wiklund I, Caidahl K, Hartford M, Karlsson T, Herlitz J.
In conclusion, women do show differences in HRQoL 1 year Improvement in quality of life differs between women and men after
after coronary bypass surgery, but with respect to PF, BP, and coronary artery bypass surgery. J Intern Med 1999;245:445–54.
GH, this seems to be mainly due to preexisting differences in 18. Hunt JO, Hendrata MV, Myls P. Quality of life 12 month after coronary
artery bypass graft surgery. Heart Lung 2000;29:401–11.
clinical health status and depression. However, gender seems to 19. Artinian NT, Duggan CH. Sex differences in patient recovery patterns
be an independent marker for RP even after adjustment for all after coronary artery bypass surgery. Heart Lung 1995;24:483–94.
covariates, which leads to the concept that rehabilitation mea- 20. Herlitz J, Wiklund I, Caidahl K, Karlson B W, Sjoland H, Hartford M,
Haglid M, Karlsson T. Determinants of an impaired quality of life five
sures designed for the specific needs of women might help to years after coronary artery bypass surgery. Heart 1999;81:342– 6.
improve their HRQoL. In particular, such interventions as stress 21. Mayer C, Ergina P, Morin JF, Gold S. Self-reported functional status as
management and attention to family role issues have been pro- a predictor of coronary artery bypass graft surgery outcome in elderly
patients. Can J Cardiol 2003;19:140 – 4.
posed for female cardiac patients (36). Further study is needed to 22. Welke KF, Stevens JP, Schults WC, Nelson EC, Beggs VL, Nugent WC.
clarify the role of depression in both men and women undergoing Patient characteristics can predict improvement in functional health after
and recovering from CABG surgery. elective coronary artery bypass grafting. Ann Thorac Surg 2003;75:
1849 –55; discussion, 1855.
We thank Elizabeth Morrison for editorial assistance. 23. Peric V, Borzanovic M, Stolic R, Jovanovic A, Sovtic S, Dimkovic S,
Marcetic Z. Predictors of worsening of patients’ quality of life six months
REFERENCES after coronary artery bypass surgery. J Card Surg 2008;23:648 –54.
24. Sorensen EA, Wang F. Social support, depression, functional status, and
1. Eagle KA, Guyton RA, Davidoff R. American College of Cardiology/
gender differences in older adults undergoing first-time coronary artery
American Heart Association ACC/AHA Guidelines for Coronary Artery
bypass graft surgery. Heart Lung 2009;38:306 –17.
Bypass Graft Surgery: a report of the American College of Cardiology/
American Heart Association Task Force on Practice Guidelines (Com- 25. Tung HH, Wei J, Chang CY. Gender differences in quality of life for post
mittee to Revise the 1991 Guidelines for Coronary Artery Bypass Graft coronary artery bypass grafting patients in Taiwan. J Nurs Res 2007;15:275–84.
Surgery). J Am Coll Cardiol 1999;34:1262–347. 26. King KB, Porter LA, Rowe MA. Functional, social, and emotional
2. Allen JK, Fitzgerald ST, Swank RT, Becker DM. Functional status after outcomes in women and men in the first year following coronary artery
coronary artery bypass grafting and percutaneous transluminal coronary bypass surgery. J Womens Health 1994;3:347–54.
angioplasty. Am J Cardiol 1990;66:921–5. 27. Verbrugge LM. Sex differentials in health. Public Health Rep 1982;97:
3. King KB, Reis HT, Porter LA, Norsen LH. Social support and long-term 417–37.
recovery from coronary artery surgery: effects on patients and spouses. 28. Schipper H, Clinch J, Olweny C. Quality of life studies: definitions and
Health Psychol 1993;12:56 – 63. conceptual issues. In: Spilker B, editor. Quality of Life and Pharmaco-
4. Mayou R, Bryant B. Quality of life after coronary artery surgery. Q J Med economics in Clinical Trials. Philadelphia: Lippincott-Raven; 1996.
1987;62:239 – 48. 29. Ruo B, Rumsfeld JS, Hlatky MA, Liu H, Browner WS, Whooley MA.
5. Keresztes PA, Merritt SL, Holm K, Penckofer S, Patel M. The coronary Depressive symptoms and health-related quality of life: the heart and soul
artery bypass experience: gender differences. Heart Lung 2003;32:308 –19. study. JAMA 2003;290:215–21.
6. Koch CG, Khandwala F, Cywinski JB, Ishwaran H, Estafanous FG, Loop 30. Goyal TM, Idler EL, Krause TJ, Contrada RJ. Quality of life following
FD, Blackstone EH. Health-related quality of life after coronary artery cardiac surgery: impact of the severity and course of depressive symp-
bypass grafting: a gender analysis using the Duke Activity Status Index. toms. Psychosom Med 2005;67:759 – 65.
J Thorac Cardiovasc Surg 2004;128:284 –95. 31. Lee GA. Determinants of quality of life five years after coronary artery
7. Phillips Bute B, Mathew J, Blumenthal JA, Welsh-Bohmer K, White bypass graft surgery. Heart Lung 2009;38:91–9.
WD, Mark D, Landolfo K, Newman MF. Female gender is associated 32. Klerman GL, Weissman MM. Increasing rates of depression. JAMA
with impaired quality of life 1 year after coronary artery bypass surgery. 1989;261:2229 –35.
Psychosom Med 2003;65:944 –51. 33. Lemos K, Suls J, Jenson M, Lounsbury P, Gordon EE. How do female
8. Ayanian JZ, Guadagnoli E, Cleary PD. Physical and psychosocial func- and male cardiac patients and their spouses share responsibilities after
tioning of women and men after coronary artery bypass surgery. JAMA discharge from the hospital? Ann Behav Med 2003;25:8 –15.
1995;274:1767–70. 34. Hildingh C, Fridlund B, Lidell E. Women’s experiences of recovery after
9. Lindquist R, Dupuis G, Terrin ML, Hoogwerf B, Czajkowski S, Herd JA, myocardial infarction: a meta-synthesis. Heart Lung 2007;36:410 –7.
Barton FB, Tracy MF, Hunninghake DB, Treat-Jacobson D, Shumaker S, 35. Kendel F, Dunkel A, Lehmkuhl E, Hetzer R, Regitz-Zagrosek V. Does time
Zyzanski S, Goldenberg I, Knatterud GL. Comparison of health-related spent on household activities or housework stress complicate recovery fol-
quality-of-life outcomes of men and women after coronary artery bypass lowing coronary artery bypass surgery? Women Health 2008;48:325–38.
surgery through 1 year: findings from the POST CABG biobehavioral 36. Linden W, Phillips MJ, Leclerc J. Psychological treatment of cardiac
study. Am Heart J 2003;146:1038 – 44. patients: a meta-analysis. Eur Heart J 2007;28:2972– 84.