Depression predicts perioperative outcomes following coronary artery bypass graft surgery

Abstract

Objectives. To assess preoperative depression in middle-aged men undergoing coronary artery bypass graft surgery (CABG) and to determine if depression is related to perioperative outcomes. Design. One hundred and nine middle-aged male patients were randomly selected and assessed for depression one day before CABG using the Symptom Checklist-90 Revised (SCL-90R). Perioperative outcomes were: (1) postoperative length of hospital stay, (2) the presence of any early complications (at intensive care unit), and (3) the presence of any late complications (at cardiac surgery unit). Results. Twenty-five (23%) patients had a high level of depression. Preoperative depression scores significantly predicted postoperative length of hospital stay (p<0.001) and the incidence of late perioperative complications (p<0.05) independently from biomedical and sociodemographic factors. Each increase in depression T score increased the odds of occurrence of late complications by 10% (p=0.018, CI 95% 1.02–1.19). Conclusions. Depression is common in middle-aged men undergoing CABG and is an independent predictor of postoperative length of hospital stay and late perioperative complications.

Key words::

• CABG
• depression
• length of hospital stay
• perioperative outcomes

There is substantial evidence that depression is associated with morbidity and mortality in cardiac patients (1). It is also known that younger patients undergoing coronary artery bypass graft surgery (CABG) are at higher risk of psychological distress (2). CABG is the most widely used surgical procedure for relieving symptoms of coronary artery disease. Although some adverse outcomes from surgery can be attributed to physical factors, the considerable variation of outcomes remains unexplained. It is reported that the prevalence of depression before CABG is in the range of 20–47% (3,4). Some studies have determined that depression is an independent prognostic factor for mortality, readmission, cardiac events (CE) and lack of functional benefits (5), greater atherosclerotic progression (6), recurrence of angina (7), and poorer quality of life (8). Scheier and colleagues (9) found that patients with significant depressive symptoms had an odds ratio of experiencing CE in the first six months after surgery that was almost twice that of patients without depression. Moreover, it has been estimated that depression is associated significantly with cardiac morbidity five to 12 years following CABG (3,7).

There have been few investigations of the association between depression and perioperative outcomes in surgical patients. Some studies have found depression to be independently related to postoperative pain and morphine consumption by patients after gynecologic surgery (10) and to delayed recovery after CABG (11). Another study demonstrated that the length of postoperative hospital stay after surgical treatment could be shortened to 2.7 days by special preoperative psychological preparation (12). However, there are no studies exploring the relation of depression to perioperative outcomes in middle-aged men undergoing CABG.

The aims of the current study were: (1) to investigate the prevalence of preoperative depression in a sample of middle-aged men undergoing CABG; (2) to assess the impact of depression on the length of postoperative hospital stay; and (3) to determine whether depression is related to early (at intensive care unit) and late (at cardiac surgery unit) occurrence of any complications.

Materials and methods

Patients

The sample was comprised of 132 randomly chosen middle-aged male patients (≤65 years of age) undergoing isolated CABG in the Cardiac, Thoracic and Vascular Clinic of Kaunas Medical University Hospital in 2006. Eight patients completed the psychological questionnaires incorrectly and 15 patients did not return the questionnaires; these patients were excluded from the study. The only significant differences between the 23 excluded patients and the remaining 109 participants in the study were a longer postoperative hospital stay for the excluded patients (p<0.02), and higher percentages of excluded patients were disabled by their cardiac illness (p<0.05), developed heart failure as an early perioperative complication (p=0.01), or developed arrhythmias as a late perioperative complication (p=0.01). One of the excluded patients met criteria for New York Heart Association (NYHA) Class IV. Another excluded patient expressed suicidal ideation during the postoperative period and was diagnosed with major depression and treated with antidepressant medication. None of the excluded patients or participants in the study died during their time in hospital.

Psychological assessment

Depression was measured with the depression scale of the widely used 90-item Symptom Checklist Revised (SCL-90R), a self-rating inventory with nine clinical scales for assessing psychological symptoms and three global indices (13). The SCL-90R is a reliable and valid measure of psychopathology (13–15) and has been used in previous studies of patients with coronary artery disease (16). In the current study Cronbach's alpha coefficient for the depression subscale was 0.824. A T score of ≥71 on this subscale was considered an indicator of high depression. Psychometric testing was performed one day before surgery.

Baseline biomedical and sociodemographic factors

Preoperative biomedical and sociodemographic factors assessed in the study were age, smoking, education, disability from illness vs. ability to be employed, arterial hypertension (AH), diabetes, ejection fraction (EF), New York Heart Association (NYHA) Class, the number of grafts, history of myocardial infarction (MI), and history of percutaneous transluminal coronary angioplasty (PTCA). The data were collected from the medical records or by interview. Age and education (in years) were analyzed as continuous variables. The other variables were dichotomized as follows: smoking – into yes (if regular smoking was at least one year before surgery) or no; disability, hypertension, diabetes, and history of MI and/or PTCA – into present or absent; EF – into ≥35% or <35% (the cut-off for the determination of cardiac severity of illness); NYHA.

Class – into Class II or Class III (there were no patients with Class I and IV among the participants in the study); surgery type – into ≥3 or <3 grafts (the cut-off for the determination of cardiac severity of illness).

Perioperative outcomes

The perioperative outcomes assessed were the postoperative hospital length of stay (LOS), and any incident of early (at intensive care unit) and late (at cardiac surgery unit) complications reported in the medical records. The possible early complications were arrhythmias, bleeding (with or without hemorrhagic shock), emergency resternotomy, positive results of culture or treatment with antibiotics, prolonged ventilation time, fever, delirium, heart failure, angina pectoris (with or without cardiogenic shock), acute bleeding gastric ulcer, acute hypotension, pneumothorax, renal failure, and MI. The possible late complications were angina pectoris (with or without cardiogenic shock), leg wound infection requiring excision of tissue, pleural effusion, mediastinitis requiring drainage, sternal dehiscence, cerebrovascular insufficiency, fever, arrhythmias (requiring or not requiring a pacemaker), pneumonia, pericardial effusion, urination disorder, delirium, and hypotension. The scoring was dichotomized into yes/no for both early and late complications.

Statistical analysis

Data were analyzed with SPSS for Windows version 13.0. Results are expressed as mean ± standard deviation (SD), percentages, odds ratios and 95% confidence intervals (95%CI). LOS expressed in days was selected as an outcome of surgery. The other perioperative outcomes were presence of early and late complications.

The Kolmogorov-Smirnov test was used to test the assumption of normality for continuous variables. Because the LOS outcome variable was positively skewed, a natural log transformation was performed to obtain a distribution more appropriate for multiple regression. Univariate associations between LOS, adverse perioperative outcomes, baseline sociodemographic and biomedical factors of the patients and depression scores were examined using t-tests or Mann-Whitney tests (when normality assumption did not hold), χ² test or exact Fisher‘s test, and bivariate correlation analyses. The differences and relations were considered significant when p<0.05. Regression analysis was used to examine multi-variate associations between outcomes and variables. A hierarchical regression analysis was used to evaluate the LOS outcome. To determine whether depression as a risk factor for LOS was independent of other patient characteristics, sociodemographic and biomedical covariates identified on the basis of prior research were entered in the first two steps of the regression model and depression was entered in the third step. Logistic regression analysis was used to evaluate the impact of depression on the odds of incidents of early and late complications.

The Kaunas Regional Committee for Biomedical Research Ethics approved the study on March 8, 2006 (issue registration number – BE-2-6).

Results

Baseline biomedical and sociodemographic data of the participants, and their LOS and SCL- 90R depression subscale T scores, are shown in Table I. Thirty-four patients (31.2%) developed early perioperative complications and 16 patients (14.7%) developed late perioperative complications; the number of incidents of specific early and late complications are presented in Table II. Twenty- five patients (23%) obtained scores ≥ the cut-off of 71 for high depression on the depression subscale of the SCL-90R before surgery. Depression scale T scores were normally distributed for the total sample and for subgroups of patients with/without early complications, patients with/without late complications, and disabled vs. employable patients; they were not normally distributed for subgroups based on the number of vessels grafted.

Table I. Sociodemographic and biomedical variables, depression scores, and length of hospital stay (N=109).

Variables	n	%	Mean	SD	Range
Age (years)	109		57.8	6.7	37–65
Education (years)	109		10.5	3.0	4–16
Smoking (yes)	76	69.7
Length of hospital stay (days)	109		11.37	7.08	6–54
Adjusted length of stay (using the natural log)	109		2.33	0.39	2–4
Depression subscale T score	109		63.13	8.22	35–80
Ejection fraction < 35%	30	27.5
NYHA Class: I class	0	0
II class	66	60.6
III class	43	39.4
IV class	0	0
Disability	25	22.9
Arterial hypertension	84	77.1
Diabetes	10	9.2
History of MI	83	76.1
PTCA history	11	10.1
≥ 3 vessels surgery group	99	90.8

NYHA, New York Heart Association; MI, myocardial infarction; PTCA, percutaneous transluminal coronary angioplasty; SD, standard deviation.

Table II. Incidents of early and late perioperative complications (N=109).

Early Complications	n	%	Late Complications	N	%
Total	34	31.2	Total	16	14.7
Arrhythmias	12	11	Leg wound infection	3	2.8
Bleeding	6	5.5	Pleural effusion	2	1.8
Emergency resternotomy	2	1.8	Mediastinitis	1	0.9
Positive culture or treatment withantibiotics	1	0.9	Sternal dehiscence	1	0.9
Prolonged ventilation time	5	4.6	Cerebrovascular insufficiency	1	0.9
Fever	4	3.7	Fever	3	2.8
Delirirum	6	5.5	Arrhythmias	2	1.8
Heart failure	2	1.8	Pneumonia	1	0.9
Angina pectoris	1	0.9	Pericardial effusion	1	0.9
Acute hypotension	1	0.9	Urination disorder	1	0.9
Pneumothorax	2	1.8	Delirium	1	0.9
Renal failure	2	1.8	Hypotension	1	0.9
Myocardial infarction	2	1.8	Renal failure	0	0.0

When relations between depression and the biomedical and sociodemographic factors were analyzed, statistically significant differences in preoperative depression scores were found between the disabled patients (mean=66.2, SD=7.2) and the employable patients (mean=62.2, SD=8.3) (t(107)=–2.232, p<0.05) and between patients who had ≥3 vessels grafted (median score = 63, interquartile range = 11) and those who had <3 vessels grafted (median score = 70, interquartile range = 8; p<0.05). Depression was not associated with other biomedical and sociodemographic factors.

Univariate analyses demonstrated a significant relation between preoperative depression and LOS (r=0.326; p<0.001), but a non-significant relation with incidents of early perioperative complications. Preoperative depression, however, was significantly higher in patients with late complications (mean depression T score=67.4, SD=8.0), than in patients without late complications (mean=62.4, SD=8.1) (t(107)=2.28, p<0.05). When depression scores were analyzed in relation to separate complications using the method of logistic regression, it was found that each increase in depression score almost doubled the possibility of leg wound infection (OR=1.99; 95% CI 0.99–1.46, p=0.066). The occurrence of leg infection was found in 8% of patients with high depression compared with 1.2% of patients with low depression. Both of these results, however, and associations between depression and other complications, were non-significant.

When relations between LOS and biomedical and sociodemographic factors were investigated, univariate analyses revealed a significant relation with age (r=0.304, p<0.01). Although patients disabled by their illness experienced longer LOS when compared with employable patients, the difference was non-significant. LOS was unrelated to the other biomedical and sociodemographic variables.

A hierarchical regression analysis was performed to determine whether depression as a risk factor for LOS was independent of other characteristics of the patients. This method allowed us to control for the effect of characteristics that might contribute to LOS before examining the effect of preoperative depression. Sociodemographic covariates (age, education, smoking, disability) were entered as a block in the first step of the regression model, followed by biomedical indicators (diabetes, ejection fraction, AH, history of MI, surgery group, NYHA and PTCA) as a block in the second step, and by depression in the third step. The results of the hierarchical regression analysis of the impact of depression on natural logarithm LOS are presented in Table III. A beta coefficient indicates the difference in LOS that is associated with an increase of one standard deviation in independent variable when holding constant the effects of the other independent variables. R² indicates the percent of variance in LOS that is explained by each regression model. Covariates entered at the first step of the analysis accounted for 10.9% of the variance with all covariates identified as independent predictors of postoperative LOS. Older age was significantly associated with longer postoperative LOS. Biomedical indicators identified as independent predictors entered at the second step accounted for an additional 8.6% of the variance. A relationship of borderline significance was found between PTCA and postoperative LOS (p=0.05). Entry of the depression variable at the third step accounted for an additional 10.5% of the variance. At that step, higher level of depression, older age, and PTCA were significantly associated with longer postoperative hospital stay.

Table III. Results from hierarchical linear regression analysis predicting postoperative length of hospital stay following CABG (N=109).

	Standardized beta	p	Standardized beta	p	Standardized beta	p
Covariates
age	0.256	0.01	0.311	0.003	0.335	0.001
education (years)	0.097	0.33	0.081	0.43	0.105	0.28
smoking	−0.057	0.56	−0.068	0.49	−0.047	0.61
disability	0.1	0.30	0.073	0.49	−0.021	0.84
diabetes			0.018	0.86	0.063	0.52
ejection fraction			−0.114	0.25	−0.078	0.40
hypertension			−0.084	0.42	−0.126	0.20
history of MI			−0.182	0.07	−0.146	0.12
surgery group			−0.065	0.51	−0.004	0.96
NYHA Class			−0.043	0.67	−0.034	0.73
PTCA			0.192	0.05	0.206	0.026
depression score					0.345	0.001
R^2	10.90%		19.50%		30%
Adjusted R^2	7.30%		10%		20.80%

MI, myocardial infarction; NYHA, New York Heart Association; PTCA, percutaneous transluminal coronary angioplasty.

When late complications were analyzed in relation to biomedical and sociodemographic factors, results revealed an occurrence of late complications in 40% of patients with diabetes compared with 12.2% in those without diabetes; the difference between these subgroups was statistically significant (χ²=5.636, p<0.05). Relationships with other factors were not significant. Univariate logistic regression analysis showed that each increase in depression T score increased the odds of occurrence of late complications by 9% (p=0.028, CI 95% 1.051–1.187). When controlling for diabetes, each depression T score increased the odds of presence of late complications by 10% (p=0.018, CI 95% 1.02–1.19) (see Table IV).

Table IV. Results from logistic regression analysis of late perioperative complications in relation to diabetes and depression (N=109).

Variables	p-value	OR	95% CI
Diabetes	0.017	6.40	1.39–29.42
Depression	0.018	1.10	1.02–1.19

CI, confidence interval; OR, odds ratio.

Discussion

The results of this study show that almost one quarter of the CABG patients had a high level of preoperative depression, and those with high depression scores stayed longer in hospital following surgery and were more likely to have late perioperative complications. The prevalence of preoperative depression in our sample is comparable to that reported by Connerney et al. (3) who found that 20% of CABG patients met DSM-IV criteria for major depression and 28% scored high on the Beck Depression Inventory. Our results also support findings from several other studies showing an association between depression in patients undergoing heart surgery and LOS (11,17,18).

It is of interest that dispositional pessimism, which is related to depression, has also been shown to influence LOS. In an investigation of patients undergoing elective CABG, Halpin and Barnett (18) found, after adjusting for age, gender, and severity of disease, that the average LOS for pessimistic patients was 1.3 days longer than for optimistic patients. Although Contrada et al. (17) found that preoperative depressive symptoms were associated with longer LOS in patients undergoing CABG or valve replacement or both, stronger religious beliefs were associated prospectively with fewer surgical complications and shorter hospital stays. The effect of religious beliefs was stronger among women than men and was independent of biomedical and other psychosocial predictors. In that study, dispositional optimism and social support were unrelated to outcomes (17). In another study of patients undergoing heart surgery, however, Contrada et al. (11) found that low social support showed an independent prospective association with longer LOS.

Although there are no other studies of cardiac patients with which to compare our findings on the relation of depression to incidents of early and late perioperative complications, some studies have investigated the relation between depression and specific perioperative complications. Green et al.(19), for example, found that preoperative depression in patients undergoing major, elective noncardiac surgery was an independent predictor of postoperative delirium.

One limitation of our study was the reduction of all possible outcomes to two yes/no variables which were selected because of the small sample size and relative infrequency of individual outcomes. However, individual outcomes differ from one another in their incidence, importance, and known risk factors. Further studies are needed to explore the relation of depression to each individual outcome.

The association of depression with adverse postoperative outcomes may be accounted for by several mechanisms and consequences of depression including pathophysiological changes, such as increased heart rate and reduced heart rate variability (20), lower adherence to medication and follow-up care (21,22), higher sympathetic adrenergic stimulation (23), or an unknown common antecedent factor that has a role in the development of both depression and cardiac disease (24). In a recent study of CABG patients, those with depression who carried the long allele of the serotonin-related gene polymorphism (5HTTLPR) were more likely to have a new cardiac event within two years of surgery than depressed patients with the short allele (25). This finding led the investigators to suggest that combining genetic and psychiatric profiling may be a way to identify patients at the highest risk for adverse outcomes following CABG (25). Depression might also influence post-CABG inflammation or infection, since it has been associated with decreased natural killer cell cytotoxicity and with a higher incidence of in-hospital fevers as well as infectious illness during the first six months after CABG (26).

Given the impact of depression on perioperative outcomes, and the relative ease with which depression can be assessed preoperatively, it seeems likely that psychological inteventions to ameloriate depressive symptoms could reduce LOS and late complications following CABG, and thereby not only improve patient care but also have the economic benefit of reducing medical costs.

Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.