Self-Reported Health and Outcomes in Patients With Stable Coronary Heart Disease

Background-—The major determinants and prognostic importance of self-reported health in patients with stable coronary heart disease are uncertain.
Methods and Results-—The STABILITY (Stabilization of Atherosclerotic Plaque by Initiation of Darapladib Therapy) trial randomized 15 828 patients with stable coronary heart disease to treatment with darapladib or placebo. At baseline, 98% of participants completed a questionnaire that included the question, “Overall, how do you feel your general health is now?” Possible responses were excellent, very good, good, average, and poor. Adjudicated major adverse cardiac events, which included cardiovascular death, myocardial infarction, and stroke, were evaluated by Cox regression during 3.7 years of follow-up for participants who reported excellent or very good health (n=2304), good health (n=6863), and average or poor health (n=6361), before and after adjusting for 38 covariates. Self-reported health was most strongly associated with geographic region, depressive symptoms, and low physical activity (P<0.0001 for all). Poor/average compared with very good/excellent self-reported health was independently associated with major adverse cardiac events (hazard ratio [HR]: 2.30 [95% confidence interval (CI), 1.92–2.76]; adjusted HR: 1.83 [95% CI, 1.51–2.22]), cardiovascular mortality (HR: 4.36 [95% CI, 3.09–6.16];adjusted HR: 2.15 [95% CI, 1.45–3.19]), and myocardial infarction (HR: 1.87 [95% CI, 1.46–2.39]; adjusted HR: 1.68 [95% CI, 1.25–2.27]; P<0.0002 for all).Conclusions-—Self-reported health is strongly associated with geographical region, mood, and physical activity. In a global coronary heart disease population, self-reported health was independently associated with major cardiovascular events and mortality beyond what is measurable by established risk indicators.Clinical Trial Registration-—URL: http://www.ClinicalTrials.gov. Unique identifier: NCT00799903. ( J Am Heart Assoc. 2017;6: e006096. DOI: 10.1161/JAHA.117.006096.)

General health is usually considered to be a patient-centered measure.1 It can be assessed by asking patients about their physical, mental, and social functioning and other aspects of quality of life using questionnaires such as the Short Form–362 and the EQ-5D.3 The simplest assessment of health can be obtained by asking a single question: “How is your overall health?” In large epidemiological studies, the response to this simple question predicted both total and cardiovascular mortality.4–6 Few studies, however, have evaluated the clinical importance of this simple indicator of overall health in CHD patients.7–9 In addition, the primary determinants of self- reported health in patients with stable CHD are uncertain.The aims of this study were to describe the most important factors associated with self-reported health and to evaluate its independent association with outcomes in patients with stable CHD who participated in the global STABILITY (Stabi- lization of Atherosclerotic Plaque by Initiation of Darapladib Therapy) trial.10,11The STABILITY trial (ClinicalTrials.gov identifier NCT007- 99903) was a global outcomes trial designed to determinewhether darapladib, a specific inhibitor of lipoprotein- associated phospholipase A2, would reduce the risk of cardiovascular death, myocardial infarction, and stroke in patients with chronic CHD.10 In total, 15 828 participants from 39 countries were randomized. All patients had chronic stable CHD, defined as prior myocardial infarction, prior coronary revascularization, or multivessel CHD confirmed by coronary angiography.

In addition, patients had to meet at least one of the following cardiovascular risk criteria: age≥60 years; diabetes mellitus requiring pharmacotherapy; high-density lipoprotein cholesterol <1.03 mmol/L; current or previous smoker, defined as ≥5 cigarettes per day on average; significant renal dysfunction (estimated glomerular filtration rate ≥30 and <60 mL/min per 1.73 m2 or urine albumin:creatinine ratio ≥30 mg albumin/g creatinine); or polyvascular disease (CHD and cerebrovascular disease or CHD and peripheral arterial disease). Darapladib did not influence the risk of major adverse cardiac events (MACE).10,11 More detailed descriptions of the study design and population were published previously.10–12 All patients provided written informed consent, and the relevant ethics committees in each participating country approved the study, in accordance with the Helsinki Declaration.At baseline, 15 528 participants (98%) completed a lifestyle questionnaire. This included the following question: “Overall, how do you feel your general health is now?” Possible responses were excellent, very good, good, average, and poor. Because only a small proportion of participants responded with excellent and poor, results are presented for 3 groups (1) excellent or very good, (2) good, and (3) average or poor.In addition to the cardiovascular risk criteria needed for study inclusion, the following information was recorded at baseline: age, sex, diagnosis of hypertension, history of congestive heart failure, prior myocardial infarction, prior percutaneous coronary intervention or coronary artery bypass grafting, and New York Heart Association functional class. Body mass index was calculated in kg/m2. Geographic regions of enrollment were North America, South America including Mexico, Western Europe including Australia and New Zealand, Eastern Europe and Asia including South Africa.

These country groupings were chosen primarily by geographic location but also considered the similarity of race, culture, and gross domestic product of countries.On the lifestyle questionnaire, patients indicated whether they lived alone and the number of years of education completed, classified as <8 years, 8 to 12 years, trade school, or college or university. Financial stress and home- and work- related stress were graded as never or rarely, sometimes, often, or always. Depressive symptoms were assessed from the questions “How is your current mood?” (denoted as “depressed mood”) and “Have you lost interest in activities or hobbies that normally give you pleasure?” (denoted as “loss of interest”), with possible responses of never or rarely, some- times, often, or always.13 Physical activity was assessed from the total hours of moderate-intensity (4 metabolic equiva- lents) and vigorous-intensity (8 metabolic equivalents) phys- ical activity during an average week.14,15 A healthy diet was classified using a Mediterranean diet score based on responses to a simple food-frequency questionnaire.16 The number of remaining teeth was recorded. Tooth loss was classified as ≤25 teeth.17Hemoglobin, white blood cell count, creatinine, low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, trigylcerides, and C-reactive protein were measured at a Quest Diagnostics Clinical Laboratory. In addition, blood samples were obtained from 14 577 patients at baseline and stored until biochemical analysis.

Lipoprotein-associated phospholipase A2 was measured at diaDexus Inc. High- sensitivity troponin T, NT-proBNP (N-terminal pro-B-type natriuretic peptide), cystatin C, interleukin 6, and growth differentiation factor 15 were measured by an electrochemi- luminescence immunoassay using a Cobas e 601 analyzer (Roche Diagnostics) at the Uppsala Clinical Research Center Laboratory in Uppsala, Sweden.The primary end point of MACE was the composite of cardiovascular death, myocardial infarction, or stroke. Second- ary end points were total mortality, cardiovascular mortality, noncardiovascular mortality, myocardial infarction, stroke, major coronary events, and hospital admission for heart failure. Clinical outcomes were adjudicated by a committee. The event definitions and main results of the study were presented elsewhere.10,11 The median follow-up time was 3.7 years (interquartile range: 3.5–3.8 years). Vital status was complete for 99.3% of patients and 99.6% of total possible follow-up time.The baseline characteristics were summarized, with categor- ical variables presented as count and proportion and contin- uous variables presented as mean and standard deviation. To investigate differences across the 3 groups of patients, the categorical variables were compared with the v2 test. Continuous variables were compared with Mann–Whitney nonparametric tests.To investigate the associations between different covari- ates and self-reported health, multivariable ordinal logistic regressions were used due to the ordinal nature of the dependent variable (excellent or very good, good, average orpoor). The results of the regression analyses are presented as adjusted odds ratios (ORs) and 95% confidence intervals. The ORs describe the associations of different covariates with decreasing levels of self-reported health. Standardized ORs are presented for all continuous variables. All biomarkers were based on a 1-SD increase, ORs for body mass index and Mediterranean diet score were based on a 1-U increase, and ORs for physical activity (metabolic equivalent in hours per week) is based on a 50% increase.

To evaluate the relative strength of association of different covariates with self- reported health, v2—df was plotted.18 This statistic allows comparison of predictors with different parameters, with a higher number indicating a stronger association.Associations between self-reported health and the study outcomes were assessed using Cox proportional hazards regression models and expressed with hazard ratio and 95% confidence interval. The underlying proportional hazards assumptions of the Cox proportional hazards models were verified by Schoenfeld residual tests. Using a multivariate model, we adjusted for demographic variables (age at randomization, sex, geographic region), psychosocial mea- sures (depressed mood, loss of interest in hobbies, financial stress, stress at work or home, years of education), lifestyle risk factors (body mass index, smoking status, physical activity,15 Mediterranean diet score,16 attendance at cardiac rehabilitation), disease markers at baseline (diagnosis of hypertension, congestive heart failure, significant renal dys- function, prior myocardial infarction, prior coronary revascu- larization [percutaneous coronary intervention or coronary artery bypass grafting]), prior multivessel chronic heart disease, diabetes mellitus, polyvascular disease, tooth loss, New York Heart Association functional class), and biomarkers (low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, hemoglobin, high-sensitivity troponin T, inter- leukin 6, growth differentiation factor 15, triglycerides, estimated glomerular filtration rate [using the Chronic Kidney Disease Epidemiology Collaboration formula], creatinine, white blood cell count, high-sensitivity C-reactive protein, NT-proBNP, cystatin C, and lipoprotein-associated phospholi- pase A2 activity). The covariates included in the model were prespecified based on previous analyses from the STABILITY trial.16,17,19,20 Kaplan–Meier curves were constructed for MACE by self-reported health groups.All analyses were performed using SAS software version9.4 (SAS Institute). For all statistical analyses, a 2-sidedP<0.05 was considered statistically significant.

Results
The baseline characteristics of the study population are presented for participants reporting excellent or very good, good, and average or poor self-reported health in Table 1. Abroad range of demographic, geographic, psychosocial, life- style-related, and conventional cardiovascular risk factors, disease markers, and biomarkers were associated with self-reported health. Thirty-two of these 38 covariates wereassociated with self-reported health in univariate analysis, with P<0.0001.Patients who self-reported poorer health were slightly more likely to be on an angiotensin-converting enzymeinhibitor or angiotensin II receptor antagonist, a beta-blocker, or a P2Y12 antagonist compared with those with better self- reported health (Table 1). Aspirin use was slightly less in patients reporting poorer general health. Statin use was similar by self-reported health. There was no difference in self- reported health at baseline according to allocation to darapladib or placebo, as expected by randomization.Variables independently associated with self-reported health in a multivariable adjusted model are displayed in Table 2. The statistical strength of associations between different variables and self-reported health is compared in Figure 1. The strongest independent associations were with geographic region, depressed mood, and low physical activity. The geographic region with the best self-reported health was North America, followed by South America, Western Europe, Eastern Europe, and Asia (Table 1). After adjusting for other covariates associated with self-reported health, persons from South America and North America had the best self-reported health, participants from Western Europe reported slightly worse general health, and those from Eastern Europe and Asia were much more likely to report poorer health (Table 2).

A less healthy diet, obesity, smoking, diabetes mellitus, hyper- tension, exertional dyspnea, cardiovascular disease markers, and tooth loss were each associated with poorer self-reported health. Blood markers independently associated with poorer self-reported health were high-sensitivity cardiac troponin T, growth differentiation factor 15, interleukin 6, low-density lipoprotein cholesterol, and low hemoglobin. Age and sex were not associated with self-reported health in the fully adjusted model. Attendance at cardiac rehabilitation was not associated with self-reported health after adjusting for covariates (Table 2).There was a stepwise increase in the risk of MACE withworsening self-reported health (Figure 2). The hazard ratios for adverse outcomes by self-reported health before and after adjustment for covariables are displayed in Figure 3A and 3B, respectively. Poorer self-reported health was associated with increased cardiovascular and total mortality, myocardial infarction, stroke, and hospital admission for heart failure (Figure 3A). After adjustment for all covariates, these asso- ciations were maintained for all outcomes except heart failure hospitalization and stroke (Figure 3B).

Discussion
In this study, which evaluated a global population of patients with stable CHD on optimal secondary prevention treatment, average or poor self-reported health was independently associated with a 2- to 3-fold increased risk of cardiovascular mortality and myocardial infarction compared with patients reporting very good or excellent health. These observations indicate that self-reported health is an important incremental risk indicator of myocardial infarction and cardiovascular mortality in patients with stable CHD, despite optimal secondary prevention treatment. The association with a large number of prognostically important variables is consistent with the conclusion that self-reported health is a global health measure that both reflects the cumulative effects of a broad range of known risk indicators and indicates the importance of additional risk indicators not measurable by conventional methods.A number of large studies have evaluated associations between self-reported health and mortality in general popu- lations.4,6,21 In the UK Biobank cohort, which included nearly 500 000 volunteers and evaluated multiple clinical, biomar- ker, and genetic risk factors, self-reported health was the strongest single predictor of all-cause mortality in men and the third strongest mortality predictor in women after a cancer diagnosis and illness or injury.4 Meta-analyses of smaller studies have also been consistent in reporting associations between poorer self-reported health and cardio- vascular and all-cause mortality.5,22 In a large Swedish general population cohort, poorer self-reported health was associated with a higher prevalance of multiple cardiovascular risk factors, and with an increased risk of myocardial infarction during follow-up over ≈13 years.21
A systematic review of studies reporting the relationship between self-reported health and fatal and nonfatal cardio- vascular outcomes22 identified 3 studies8,9,23 including 10 648 patients with known cardiovascular or ischemic heart disease. In this meta-analysis, patients with poor health compared with good or excellent health had a ≈2.4 times higher risk of cardiovascular death, consistent with the current study. However these studies have limitations, including poor measurement of baseline risk factors and cardiovascular disease status, lack of detail on study meth- ods, and poor ascertainment of disease status or severity.

Most previous studies reporting self-reported health have been undertaken in a single country. In the current study, which included patients from 39 countries and multiple regions of the world, self-reported health was strongly associated with geographic region and country of residence. These observations suggest that cultural or regional norms need to be considered when interpreting self-reported health. In addition, the large geographic differences in self-reported health changed after adjustment for covariates, suggesting that geographic differences in self-reported health reflect both differences in the burden of disease or symptoms and different perceptions of their impact on health. Socioeco- nomic24 and international gradients25 in adverse outcomes for patients with CHD persist despite adjustment for conventional cardiovascular risk factors and are well described. It is possible these could be explained in part by differences in general health.

Both psychosocial and conventional cardiovascular risk factors have been associated with self-reported health in previous studies.7,21,22 In these studies, however, information on multiple covariates was more limited, and a detailed analysis of the relative importance of different factors was not undertaken. In the current study, depressive symptoms were strongly associated with self-reported health, consistent with an effect of mood on the perception of health and the impact of poorer health on mood. Stress from various causes and fewer years of education were also associated with poorer self-reported health. Other indicators of socioeconomic status were not assessed in this study. Lifestyle risk factors, including current smoking, physical activity, diet, and body weight, were relatively strong predictors of self-reported health.The presence of clinical markers of cardiovascular diseaseor risk, including prior myocardial infarction, stroke, polyvas- cular disease, diabetes mellitus, and renal dysfunction, was associated with poorer self-reported health; however, the strength of these associations, as assessed by the v2 statistic, was generally less than that of psychosocial and lifestyle measures. Associations between self-reported health and blood biomarkers, some of which are powerful risk predictors, were generally weaker than those of clinical variables. Associations of self-reported health with low- density lipoprotein cholesterol, hemoglobin, interleukin 6, growth differentiation factor 15, and troponin T are consistent with an influence of multiple pathophysiological pathways.

There were modest differences in use of several evidence- based medications by self-reported health, but the reasons for these differences cannot be determined reliably in this observational study. Differences in medication use could beexplained by treatment indication, such as impaired left ventricular function, which is a stronger indication for beta blockers and angiotensin-converting enzyme inhibitors or angiotensin receptor antagonists, or effects of treatment on general health (eg, if beta blockers caused fatigue). A common association between medication use and other factors that influence health, such as socioeconomic condi- tions, geography, or medical care, is also possible.12The current study has a number of limitations. The degree to which the observed associations are causal cannot be established in this observational study. Despite the availability of a wide array of clinical and biochemical risk indicators and risk factors, there are still additional unmeasured risk factors, for example, details on all comorbidities, extent of coronary lesions, and genetic factors that were not included in the database. Furthermore, participants selected to participate in a clinical trial may not be representative of all CHD patients; however, the large geographically and culturally diverse study population and the internal consistency of the data suggest that results are likely to be broadly applicable. Additional strengths of the current analysis include the standardized assessment of multiple clinical, psychosocial, and lifestyle variables and near-complete ascertainment of outcomes.

This study described associations between self-reportedhealth and a broad range of prognostically important variables. Our observation that many of these associations remained statistically significant after multivariable adjust- ment indicates that self-reported health captures a broad range of conditions affecting sense of well-being.Broad geographic regions were prespecified for the analysis, but chosen groupings are relatively crude indicators of the importance of cultural, medical, and socioeconomic differences between and within countries—a limitation that is likely to diminish estimates of the importance of geographic factors. Despite this, geographic region was a strong independent predictor of both self-reported health and MACE.An assessment of self-reported health can be undertaken simply as part of almost any consultation. Poor or worsening self-reported health may occur for many reasons, both cardiac and noncardiac. This study suggests that poorer general health, whatever the reason, is associated with a higher risk ofMACE in patients with stable CHD. This raises the possibility that interventions that improve overall health could reduce cardiovascular risk.Disease-specific patient-centered outcome measures, such as the Quality of Life After Myocardial Infarction Questionnaire26 or the Seattle Angina Questionnaire,27 may be more sensitive to the impact of cardiovascular disease on quality of life than on overall health.1 Self-reported overall health is influenced by many factors. Compared with disease-specific tools, general health may be less sensitive to changes in symptoms directly related to, for example, CHD. Nevertheless, disease-specific measures have some disadvantages. They generally include more questions, are less comparable for patients with different medical prob- lems, are more time consuming to administer, and may notcapture the impact of non–disease-related factors on general health.Because self-reported overall health is a powerful risk marker, considering it could better inform decisions about treatment or delivery of health care and discussions about risks to future health. Few studies have evaluated whether and how response to different treatments relates to self- reported health. Further research is needed to evaluate the effectiveness of targeting health care based on assessment of self-reported general health.

Conclusions
In a global stable CHD population, self-reported health was strongly associated with geographical region, psychosocial variables, and lifestyle risk factors. Self-reported health was independently associated with MACE when adjusting for baseline characteristics including a wide array of prognostic biomarkers. These data support the conclusion that self- perceived health and psychosocial and lifestyle-related factors contribute to cardiovascular events beyond what is measur- able by established risk Darapladib indicators.