Ethnic differences in quality of life and its association with survival in patients with heart failure

Abstract Background Optimizing quality of life (QoL) is a key priority in the management of heart failure (HF). Hypothesis To investigate ethnic differences in QoL and its association with 1‐year survival among patients with HF. Methods A prospective nationwide cohort (n = 1070, mean age: 62 years, 24.5% women) of Chinese (62.3%), Malay (26.7%) and Indian (10.9%) ethnicities from Singapore, QoL was assessed using the Minnesota Living with HF Questionnaire (MLHFQ) at baseline and 6 months. Patients were followed for all‐cause mortality. Results At baseline, Chinese had a lower (better) mean MLHFQ total score (29.1 ± 21.6) vs Malays (38.5 ± 23.9) and Indians (41.7 ± 24.5); P < .001. NYHA class was the strongest independent predictor of MLHFQ scores (12.7 increment for class III/IV vs I/II; P < .001). After multivariable adjustment (including NT‐proBNP levels, medications), ethnicity remained an independent predictor of QoL (P < .001). Crude 1‐year mortality in the overall cohort was 16.5%. A 10‐point increase of the physical component (of MLHFQ) was associated with a hazard (HR 1.22, 95% 1.03‐1.43) of 1‐year mortality (P = .018) in the overall cohort. An interaction between MLHFQ and ethnicity was found (P = .019), where poor MLHFQ score (per 10‐point increase) predicted higher adjusted mortality only in Chinese (total score: HR 1.18 [95% CI 1.07‐1.30]; physical: HR 1.44 [95% CI 1.17‐1.75]; emotional score: HR 1.45 [95% CI 1.05‐2.00]). Conclusions Ethnicity is an independent determinant of QoL in HF. Despite better baseline QoL in Chinese, QoL was more strongly related to survival in Chinese vs Malays and Indians. These findings have implications for HF trials that use patient‐reported outcomes as endpoints.


| INTRODUCTION
Heart failure (HF) is a debilitating condition and a leading cause of mortality worldwide. 1 Increasingly, symptomatic HF appears to affect patients in Southeast Asia disproportionately, with HF presenting at a much younger age, characterized by greater severity and poorer outcomes when compared to the rest of the world. 2 Within the Southeast Asian patient population with HF, there exist drastic ethnic differences in both hospitalization rates and mortality. 2 Understanding these differences and their underlying mechanisms-social and pathophysiologic-is necessary to effectively treat the growing HF disease burden in these countries.
Multi-morbidity is a hallmark of HF patients; their coexistence and interactions impact outcomes and functional status adversely in the elderly. 3 To effectively address HF, it is therefore imperative to gain a deeper understanding of how it affects patients' quality of life (QoL), as a cross-disease outcome. 4,5 For this study we focused on patientreported QoL, one aspect of patient reported outcome measures (PROMs). This choice builds on a growing recognition that appropriate decisions can only be made when informed by both biomedical factors and patient concerns. 6 The increasing importance of patient input has largely been fueled by the growing burden of chronic disease, for which care is often long-term and costly. While evidence remains inconsistent, some studies have found that the use of PROMs positively impacted health outcomes. 6 The integration of PROMs into clinical care represents a unique opportunity to improve the patient experience, the doctor-patient relationship, and ultimately, health outcomes.
HF has a notoriously negative impact on patient-reported QoL. [7][8][9][10] Patients must cope not only with their physical symptoms, including shortness of breath and fatigue, but also with the inability to do the things they once could, the emotional stress of being sick, and the financial burden of treatment. Accordingly, several HF-specific QoL measurement tools have been developed over the years. [11][12][13][14][15] show that a majority of HF patients place greater value in QoL than longevity. 16 Optimizing QoL, as a patient-centered outcome, must therefore become a key priority in the management of HF. In addition to being a significant treatment aim, QoL also has the potential to be a useful prognostic tool for HF, although past studies have reported inconsistent findings on the association between poor QoL and worse survival. 17 Despite a wealth of data on QoL for Western patients with HF, where prevalence is 1% to 2%, there is a notable lack of data for Southeast Asian patients with HF. 2,18 We sought to investigate ethnic differences in QoL by looking at QoL through three distinct lenses: descriptive, causal, and prognostic.
We investigated potential interethnic differences in QoL and assessed the relationship of QoL to mortality.

| Setting
Singapore is a highly developed city state. Its national healthcare expenditure constitutes approximately 4.9% of the GDP, which is considerably lower than other developed countries. 19 Singapore extends universal healthcare coverage to all the citizens; however, it has a mixed healthcare financing system. 20 While public expenditure on healthcare is partially financed through general revenues, the healthcare financing system has been layered with a more elaborate diversified system through legislating compulsory savings funds by private individuals to fund healthcare expenditure. 20 Consequently, out-of-pocket healthcare expense in Singapore is relatively high (54.8%) compared with other developed countries. 19 Nonetheless, multiple layers of healthcare financing and government subsidies (up to 80% of total bill) are in place to ensure that local citizens are not denied access to healthcare. 21

| Study design and study population
Data from the Singapore Heart Failure Outcomes and Phenotypes study (SHOP) study 18 were used to address the aims of this study. The SHOP study was a population-based study of patients with a validated diagnosis (clinician-judged) of HF, who were recruited from six centers in Singapore. Patients above 18 years of age who either presented to the hospital with a principal diagnosis of HF or attended an outpatient hospital clinic for the management of HF-within 6 months from an episode of acute decompensated HF that resulted in a hospital admission or was treated in the outpatient clinic-were included into the study and followed prospectively for 1-year. Patients with severe valvular heart disease, transient pulmonary oedema complicating acute coronary syndrome, or end-stage renal disease were excluded from the study.
The demographics of Singapore and HF admissions had been previously described. 2,18 The major ethnic groups comprised: Chinese (74%), Malay (13%), and Indian (9%), respectively, of the 5.5 million population. 18 Baseline patient demographic details and clinical data, such as vital signs and symptoms on physical examination, New York Heart Association (NYHA) functional status, serum biochemistry and hematology, comorbid conditions, medications, and interventions were recorded.
Comprehensive two-dimensional echocardiography was also performed on all eligible patients, using standardized machines at study sites.
Health-related QoL was assessed at baseline, 3 and 6 months following first admission/consultation with the Minnesota Living with Heart Failure Questionnaire (MLHFQ). 11 The MLHFQ 11 is a self-report questionnaire comprising 21 items, which assesses how HF affects the physical and emotional dimensions of the well-being of the patient.
These dimensions are combined into a total score that reflects a global assessment of the patient's well-being. Patients were asked to indicate how much HF prevented them from living, using an ordinal scale from 0 (not present or no effect), 1 (very little), up to 5 (very much). The MLHFQ score, computed by the summation of the scores to all the questions, ranges from a minimum of zero which equates to no impairment as a consequence of HF and 105 for maximum impairment. Lower MLHFQ scores correlate with better QoL. The questions cover signs and symptoms pertaining to physical activity, social interaction, sexual activity, work, and emotions. The MLHFQ was administered by the team of multi-ethnic clinical coordinators who translated it into Mandarin, Malay, Tamil (as the official language for South Asian language in Singapore) and other common dialects, for non-English speaking patients. The validity of the MLHFQ has been well documented. 22 For socioeconomic status, we used the small areal Socioeconomic disadvantage index (SEDI) described by Earnest et al. 23

| STATISTICAL ANALYSIS
Descriptive statistics were used to characterize the study population and ethnic groups. Categorical variables are described as percentages and continuous variables are described as a mean with standard deviations or median (interquartile range) if skewed. The relationship of MLHFQ scores (total, physical, and emotional) with independent risk factors was assessed using linear least-squares regression models.
Univariable analyses were first performed on all the baseline variables.
Covariates with P-values <.1 were then considered for multivariable adjustments, including any important clinical and demographic factors.
Multivariable Cox proportional hazards models were used to determine the association of MLHFQ scores (total, physical, and emotional) with 1-year all-cause mortality. Multivariable adjustments included ethnicity, age, sex, body mass index, systolic blood pressure, diastolic blood pressure, heart rate, biomarkers (Galectin-3, NT-proBNP), NYHA functional class, diabetes, coronary artery disease, atrial fibrillation, hypertension, prior stroke, liver disease, chronic respiratory disease, history of smoking or alcohol usage, evidence-based medications for HF, and areal socio-economic disadvantage index (SEDI), 23 as proxy for socioeconomic status. Interaction effects were checked. A two-tailed P value of less than .05 was considered statistically significant. All statistical analyses were performed with STATA/SE v14.0.
Ethics approvals were obtained from the relevant human ethics committees at the investigating sites. The study conforms to the Declaration of Helsinki. Comorbidities were common in the study population (Table 1): with high prevalence of coronary artery disease (CAD) (in 53.8%), hypertension (72.2%), diabetes (57.1%), chronic kidney disease (CKD, 50.5%) and smoking (53.8%). CAD, hypertension, and diabetes were significantly higher in Malays and Indians vs Chinese (P < .001).

| Subject characteristics
Comorbid atrial fibrillation was highest in the Chinese (26.5%), but notably the least in Indians (12.2%). CKD was more similarly prevalent in half of Chinese and Malays but less in Indians (44.6%). For lifestyle risk factors, Malays were more likely to be smokers than Chinese and Indians, but Indians were more likely to report alcohol intake than the other ethnicities (P < .001), Table 1.
Chinese also had lower BMIs than Malay and Indian patients (P < .001). Systolic and diastolic blood pressure, heart rate, eGFR, galectin-3 levels, NT-proBNP levels, sex, and NYHA class composition did not differ significantly across ethnicities, although more Malay (17.0%) and Indian (24.3%) patients had higher severity of HF (in NYHA class III/IV) compared to Chinese (14.8%).

| Responses to MLHFQ
Of the 21 MLHFQ questions, patients reported the greatest burden in response to "Costing you money for medical care?" (median: 3, Table 2). Patients also reported a substantial burden in response to questions focused strictly on physical symptoms: "making your walking about or climbing stairs difficult?", "making you short of breath?", and "making you feel tired, fatigued or low on energy?" (median: 2). The lowest scores were given in response to "making it difficult for you to concentrate or remember things?", "making you feel depressed?" and "making your sexual activities difficult" (median: 0).

| One-year mortality
Crude 1-year all-cause mortality was 16%, with no significant difference among the ethnicities. Of the three scores: MLHFQ total, physical, and emotional scores, the physical score was the strongest independent predictor of mortality ( Figure 1).  There has been a growing recognition that PROMs are a legitimate measure for monitoring health care outcomes. PROMs can provide insights that, while unobtainable through direct clinical measurements, are nonetheless consistent with those of clinicians. 24 Notably, NHYA class III/IV (vs class I/II) was observed to be the single most powerful independent predictor of MLHFQ total score, suggesting that the findings confirmed a certain level of alignment between physician reported and patient reported QoL, which had been previously documented. 24 Physical mobility contributed to a significant component of the overall MLHFQ score. In terms of its prognostic utility, we observed a strong (twofold) association of a high (fourth quartile vs first quartile) physical score with 1-year adjusted mortality in the overall cohort.
Findings from several studies with inclusion of the physical domain of the MLHFQ had been inconsistent in terms of its predictive association with mortality. [25][26][27][28][29] Our findings were consistent with three of these studies 25,28,29 which had similarly found physical mobility to be independently associated with mortality. Other remaining studies 26,27 where no association was found had relatively smaller sample sizes. To begin interpreting these results, we first looked to biochemical factors that may explain the observed ethnic differences in QoL.
Other statistically significant ethnic differences in baseline characteristics -rates of hypertension, coronary artery disease smoking, and diabetes-did not seem to account for the observed differences in QoL. The same is true for systolic blood pressure, levels of log NT-proBNP, use of alcohol, angiotensin II receptor blocker use, and sex distribution. We must then ask ourselves the following question: what is driving the observed ethnic difference in QoL?

| Socio-economic status and financial burden
In this analysis, areal SEDI-as proxy for socio-economic status-did not significantly differ among the ethnic groups and as such did not explain any of the variation in QoL seen. Notably, though, >70% of each ethnic group were in the lowest two SEDI (low, low to middle income) categories. This proportion is higher than would be expected in a random sample of the general population, which may reflect an association between low SES and incident HF. However, our data are limited to prevalent HF. Given the low SES of our cohort, the extreme expense of treating HF, and the high (54.8%) out-of-pocket expense for healthcare expenditures in Singapore, 19 "costing you money for medical care" posed the greatest QoL burden to our study cohort. The observation of interethnic differences in the relationship of QoL to prognosis in HF calls for an ethnicity-specific approach with respect to measures aimed at improving QoL. However, it is important to note that these associations may not reflect a fundamental ethnic difference in HF progression. Any discussion of self-reported QoL measures is incomplete without calling attention to the multitude of factors that influence QoL which could not be controlled for in our analysis. For instance, at the population-level, structural and historical ethnic and racial biases play a major role in QoL, which has been welldocumented in the literature on racial health inequities in the United States. 33 At the individual level, personality traits, such as negative affectivity, developmental experiences, and cultural factors-among others-also influence patients' perceptions of their illness. 11,34 In addition to these unmeasured biases, our study may be limited by the smaller numbers of Malays and Indians in our cohort.

| Historical and cultural factors
The study is limited in that global well-being was not measured with the use of a generic QoL instrument. Additionally, no individual patient socioeconomic variables (eg, household income, education) were available so areal SEDI had to be used as proxy. Despite these minor limitations, a diverse patient population, comprising both inpatients and outpatients, and the availability of comprehensive clinical information, including NT-proBNP for adjustment and the use of contemporary pharmacotherapy, enhance the generalizability of this population-based study.

| CONCLUSIONS
In this relatively young cohort of HF patients with high multi-morbidity, ethnic differences in QoL were seen between Chinese, Malay, and Indian patients. Ethnicity was an independent determinant of QoL. Poorer physical QoL strongly predicted 1-year survival in the overall cohort. Healthcare professionals should be mindful of such factors to educate patients and their family members so as to provide patients with coping skills to better manage HF. Finally, the findings have implications for an individualized approach to the management of HF patients of different ethnicities and for HF trials that use patient-reported outcomes as endpoints.

Patient-centered values and QoL should in essence be integrated
in clinical decision-making.

ACKNOWLEDGMENTS
The contributions of all site investigators and clinical coordinators are duly acknowledged. This study was supported by National Medical