The Heart and Soul Study is a prospective cohort study of psychosocial factors and health outcomes in patients with coronary disease. Data collection methods have been described in detail elsewhere (Bibbins-Domingo et al., 2007
; Ruo et al., 2003
; Ruo, Rumsfeld, Pipkin, & Whooley, 2004
). Briefly, administrative databases were used to identify patients with one of the following eligibility criteria: a history of myocardial infarction, angiographic evidence of at least 50% stenosis of one or more coronary vessels, evidence of ischemia by treadmill or nuclear stress testing, or a history of coronary revascularization. The exclusion criteria were an intention to move out of the area within three years, a history of myocardial infarction within the last six months, or exercise tolerance less than 1 block, all of which precluded safe completion of the study.
Between September 2000 and December 2002, we recruited 1,024 participants with CHD from 2 Department of Veterans’ Affairs Medical Centers, 1 university hospital, and 9 public health clinics in the San Francisco Bay Area. These are the subjects of the current analysis. The baseline appointment included a medical history interview, a physical examination, an exercise treadmill test with a stress echocardiogram, and a comprehensive health status questionnaire. The protocol was approved by the institutional review boards at all participating facilities.
Outcomes: HF Hospitalization
We conducted annual telephone follow-up interviews with participants (or their proxy) to ask about death or hospitalizations. If any participant or proxy reported a hospitalization, we retrieved and reviewed the medical records for that hospitalization. Two independent and blinded adjudicators retrieved and reviewed medical records, electrocardiograms, death certificates, and coroner’s reports. If both adjudicators agreed on the outcome classification, it was binding. If there was disagreement, they conferred, reconsidered their classification, and requested consultation from a third, blinded adjudicator as necessary.
The outcome of interest, HF hospitalization, was defined as hospitalization for a clinical syndrome with ≥2 of paroxysmal nocturnal dyspnea, orthopnea, increased jugular venous pressure, pulmonary rales, third heart sound, cardiomegaly on chest X-ray, or pulmonary edema on chest X-ray, as determined by the adjudicators from reviewing medical records. These clinical signs and symptoms must have represented a clear change from the normal clinical state of the patient and been accompanied by either failing cardiac output, determined as peripheral hypoperfusion (hypotension in the absence of other causes, such as sepsis or dehydration) or peripheral or pulmonary edema. In addition to all the hospital records, supportive documentation of decreased cardiac index, increasing pulmonary capillary wedge pressure, decreasing oxygen saturation, and end-organ hypoperfusion, if available, were included in adjudication.
Outcome: All-Cause Mortality
For any reported death, medical records, coroner’s reports, and death certificates were reviewed. For this analysis, we include all causes of death.
Predictor: Cardiac Self-Efficacy
The main predictor of interest was cardiac self-efficacy, defined as participants’ confidence in their ability to take care of their health (Bandura, 1977
). We measured cardiac self-efficacy using Sullivan’s validated 5-item summative “Maintain function” scale (Arnold et al., 2005
; Berkhuysen et al., 1999
; Salamah, Wahl, & Abriam-Yago, 2003
; Sullivan et al., 1998
). We elected to use a disease-specific measure because we hypothesized patients’ cardiac disease would be a strong driver of their self-efficacy, and that would be best captured by the well-validated, widely used Sullivan scale. Each item begins with the stem, “How confidant are you that you know or can,” and assesses an aspect of daily life-function, such as work and social activities (See Appendix
). The responses are a 5-level Likert scale from 0 = “not at all confident” to 4 = “completely confident.” We did not include a “not applicable” response option. The self-efficacy scores range between 0 and 20, with a higher score indicating better self-efficacy to maintain function.
To evaluate whether differences in baseline cardiac function accounted for the relation between self-efficacy and HF, we performed three baseline measures of cardiac function: echocardiographic assessment of resting left ventricular ejection fraction (LVEF), exercise treadmill test for exercise capacity, and a stress echocardiogram for assessment of fixed and inducible ischemia (wall motion abnormalities). All of these variables are well-established measures of cardiac function (Braunwald et al., 2002
A complete resting two-dimensional echocardiogram was performed on each participant. To determine the LVEF, standard two-dimensional parasternal short-axis and apical two-chamber and four-chamber views were used. Before and after exercise, we obtained apical two-chamber, four-chamber, and precordial long-and short-axis views to detect changes in wall motion or ventricular dilatation with exercise. To account for fixed and exertional wall motion defects (our measure of ischemia), we calculated the wall motion score at peak exercise. Each of 16 wall segments was evaluated for contractility at peak exercise, as follows 1 = normal, 2 = hypokinetic, 3 = akinetic, 4 = dyskinetic, 5 = aneurysm. The scores for each segment are averaged to create an index from 1–165 16, with a higher score indicating worse contractility.
Other Participant Characteristics
Participants reported demographic characteristics, including age, ethnicity, education, and marital status. Patients reported their annual household income, and because we were interested in low income as a risk factor, we dichotomized responses into less that $20,000 versus ≥$20,000 annual household income. Self-reported history of myocardial infarction, stroke, diabetes mellitus, or hypertension, as well as alcohol and tobacco use, were assessed by questionnaire. To account for other clinical characteristics that could affect HF hospitalization, we recorded use of medications such as beta-blockers, statins, renin-angiotensin system inhibitors, and antidepressants. Body-mass index (weight in kilograms divided by height in meters squared) was calculated for each participant.
Because we wanted to assess the independent contribution of self-efficacy, we chose to measure and adjust for several potentially related psychosocial variables (Cossette, Frasure-Smith, & Lesperance, 2001
; Golden-Kreutz & Andersen, 2004
; Ruo et al., 2003
; Shyu, Tang, Tsai, Liang, & Chen, 2005
; Spertus, McDonell, Woodman, & Fihn, 2000
; Sullivan, LaCroix, Spertus, & Hecht, 2000
). We measured depressive symptoms using the Patient Health Questionnaire-9 (Spitzer, Kroenke, & Williams, 1999
), a validated measure in which a higher score indicates more depressive symptoms. We considered a score of 10 or higher as consistent with depressive symptoms (McManus, Pipkin, & Whooley, 2005
). To assess perceived stress, we used the 16-point, 4-item Perceived Stress Scale (Cohen, Kamarck, & Mermelstein, 1983
), in which experiencing at least one stressful symptom “fairly often,” or a score of 9 or higher, indicates stress. We asked participants “do you have as much contact as you like with someone you feel close to, someone in whom you can trust and confide (yes/no)?” to assess social support (Williams et al., 1992
We aimed to evaluate the associations between cardiac self-efficacy and baseline cardiac function, between cardiac self-efficacy and future HF hospitalization, and between cardiac self-efficacy and all-cause mortality. First, we examined the association of decreasing self-efficacy quartile with three measures of cardiac function at baseline. Next, we created Kaplan-Meier curves to depict heart-failure free survival, and overall survival, by quartile of cardiac self-efficacy.
To further evaluate the independent association of self-efficacy with HF hospitalization, we performed a sequential multivariate linear regression, per standard deviation (4.5-point decrease) in self-efficacy. Potential predictors were grouped a priori into five conceptually based blocks, and each block of variables was added sequentially into the model as follows: (a) demographic variables: age, gender, race, income, education, marital status; (b) medical history: hypertension, diabetes mellitus, prior myocardial infarction, prior coronary revascularization, prior stroke, body mass index; (c) medication use: beta-blocker, statin, renin-angiotensin system inhibitor, aspirin, antidepressant; (d) psychosocial characteristics: depressive symptoms, current tobacco use, poor social support, regular alcohol use, and perceived stress; (e) cardiac function variables: LVEF, wall motion index, and exercise capacity. We employed an identical procedure to elucidate the relationship between self-efficacy and all-cause mortality.
In all models, we tested for interactions between self-efficacy score and other psychosocial characteristics (depressive symptoms, social support, and perceived stress); and between self-efficacy and gender and nonwhite ethnicity and age. Results are reported as odds ratios with 95% confidence intervals. Analyses were performed using SAS version 9 (SAS Institute, Inc., Cary, NC).