ACADEMIC EMERGENCY MEDICINE 2011; 18:15–21 © 2011 by the Society for Academic Emergency Medicine
Objectives: It has been well documented that screening, prevention, and treatment disparities in cardiovascular care exist. Most studies have focused on the outpatient setting. The purpose of the present analysis was to assess if a disparity of care exists in the care of emergency department (ED) patients with acute heart failure in a secondary analysis of the Heart Failure and Audicor Technology for Rapid Diagnosis and Initial Treatment (HEARD-IT) multinational study.
Methods: Only patients with an adjudicated diagnosis of acute heart failure were included in this analysis. Racial groups included in this analysis were limited to white and African American or black patients, due to their predominance in the cohort. Logistic regression including clinically relevant demographics, past medical history, exam, diagnostic tests, and adjudicated diagnosis of acute heart failure as covariates was performed to assess the association of race with treatment with a diuretic or nitroglycerin and 30-day death or readmission.
Results: Of the cohort, 418 of 1,076 (38.8%) were included in the analysis. Median age was 69 years (interquartile range [IQR] = 55–79 years), 49% were white, and 51% were African American or black. There was no difference in the correct admitting diagnosis in the two groups (p = 0.83). Multivariate adjustment revealed that African American or black race was not associated with treatment with diuretics (adjusted odds ratio [OR] = 1.00, 95% confidence interval [CI] = 0.55 to 1.82) or nitrates (adjusted OR = 1.27, 95% CI = 0.76 to 2.13) in the ED. In a separate regression analysis there was no association with African American or black race with 30-day adverse events (adjusted OR = 1.22, 95% CI = 0.68 to 2.16).
Conclusions: This secondary analysis of HEARD-IT data did not identify racial disparities in the treatment of adults with acute heart failure in the ED.
The management of acute heart failure is centered on the relief of patients’ symptoms through treatment with diuretics, vasodilators, or both. In the acute setting, treatment is often initiated without knowledge of the patient’s underlying cardiac function. Treatment algorithms recommend that the selection of therapeutic agents chosen should be based on the signs of fluid overload and congestion in conjunction with the initial blood pressure. This approach suggests that treatment in the acute setting should be consistent across race and sex.
Prior studies evaluating racial disparities have reported that heart failure is more prevalent in African Americans, a finding largely attributed to the higher prevalence of underlying hypertension.1 In addition, in patients with systolic dysfunction, mortality and functional limitations are greater in African Americans.2,3 Studies of inpatient treatment have reported varying results, with some demonstrating a disparity of care.4–11 The majority of these studies used large databases or registry data to detect disparities in care over the course of a hospitalization and did not focus on the acute setting.
There are few data regarding the variation in care patterns based on race in the initial management of patients presenting with suspected acute heart failure. It has been reported that African American patients seek care more often in the emergency department (ED) than white patients. It has also been documented that African American patients experience a lower rate of hospitalization.10 Therefore, evaluating the potential for care disparity in only the inpatient setting may underidentify areas of concern. Using data from a large prospective trial of patients with suspected heart failure presenting to the ED, we compared treatment of patients with acute heart failure based on white or African American or black race.
This was a secondary analysis of the Heart Failure and Audicor Technology for Rapid Diagnosis and Initial Treatment (HEARD-IT) multinational study that evaluated the use of automated heart sound detection in the evaluation of patients presenting to the ED with dyspnea suspected to be from heart failure at presentation.12 The study was approved by the institutional review boards at all participating centers.
Study Setting and Population
A convenience sample of patients was enrolled at nine sites (seven in the United States, one in Switzerland, and one in Taiwan) from March to October 2006.12 All of the participating institutions were academic medical centers, and both emergency medicine house staff and board-certified attending physicians were responsible for patient care. Eligible patients were at least 40 years old, with dyspnea as the chief complaint. Patients with dialysis-dependent renal failure, or whose dyspnea was clearly not related to acute heart failure (e.g., penetrating chest injury or blunt trauma), were excluded. In this secondary analysis of the HEARD-IT trial data, only patients with an adjudicated diagnosis of acute heart failure were included.
Patients fulfilling inclusion and exclusion criteria who supplied informed consent were enrolled. ED data collected prospectively included race, demographics, medical history, physical examination, and electrocardiogram findings, as documented by the treating emergency physician. Race was coded as African American or black, white, Asian, and other. For the purpose of this analysis, black and African American were analyzed as a single cohort. Medications administered in the ED and before arrival were also recorded.
All patients were followed by chart review throughout their index stay to document in-hospital events. Thirty-day follow-up was obtained from a telephone interview by study personnel, and medical records were reviewed for all patients at follow-up. Adverse events were defined as all-cause mortality or cardiac-related hospital admission (acute coronary syndrome [ACS] or acute heart failure).
Two cardiologists at each institution with access to data collection forms, clinical reports from the index hospitalization, and follow-up data, but not the acoustic cardiography results, independently reviewed all medical records pertaining to the patient and categorized the patient’s acute presentation as symptoms resulting from: 1) acute heart failure, 2) a cause other than acute heart failure, or 3) a cause other than acute heart failure in a patient with a history of left-sided ventricular dysfunction. If the two cardiologists disagreed, results were adjudicated by a third cardiologist. There was no adjudication for any other cardiac diagnosis, including ACS.
Only racial groups white and African American or black were included in this analysis due to their predominance within the cohort. The sample was described with medians and interquartile ranges (IQRs) for continuous variables and frequencies and percentages for categorical variables. Measures of association are presented as odds ratios (ORs) with 95% confidence intervals (95% CIs). The Shapiro-Wilk test was used to assess for normality for all continuous variables entered into the regression models. The continuous variable B-type natriuretic peptide (BNP) failed to conform to a normal distribution and therefore was log-transformed to achieve a parametric distribution. Multivariable logistic regression used generalized estimating equations analysis to adjust for clustering within enrollment site, and was performed to assess the effect of site on the use of ED diuretic or vasodilators. No association was found.
Standard logistic regression, including covariates identified by prior studies, or clinical relevance that affect treatment decisions at the time of presentation, was performed to determine the association with race and early treatment. Relevant covariates to determine association of race with use of ED diuretic or vasodilator included: race, sex, BNP log-transformed, history of renal insufficiency, history of coronary artery disease, history of congestive heart failure (CHF), history of hypertension, history of pulmonary disease (chronic obstructive pulmonary disease [COPD] or asthma), home medication of diuretics, presence of jugular venous distention, presence of lower extremity edema, presence of rales or wheezing on exam, initial systolic blood pressure (mm Hg), initial pulse oximetry, and chest radiograph showing cardiomegaly or signs of congestion (cephalization, pleural effusion, or interstitial edema) at presentation.13–17 All covariates were entered into the regression model.
A separate standard logistic regression model using all candidate variables was created to determine the association between race and adverse events at 30 days. Candidate variables were those previously reported to be associated with acute heart failure prognosis including age, sex, race, blood urea nitrogen, sodium, hemoglobin, creatinine, log-transformed BNP, systolic blood pressure (mm Hg), respiratory rate, ED diagnosis of myocardial infarction, a history of CHF, and intensive care unit admission.18–23 Adjusted OR and 95% CI are presented.
Colinearity was assessed by evaluation of the tolerance and variance inflation factor of covariates included in both models. Interactions were assessed between sex and history of CHF, race and sex, and history of hypertension and race. The discriminatory capacity of the final models was assessed using the area under the receiver-operating characteristic curve. The goodness of fit was evaluated with the Hosmer-Lemeshow test. Linearity of the logits was ascertained using the Box-Tidwell procedure and the logits of all continuous variables in the models were found to be linear.24 Analyses were conducted with Stata, Version 10.0 (Stata Inc., College Station, TX).
The overall HEARD-IT study consisted of 1,076 patients, of whom 418 (38.8%) had an adjudicated diagnosis of acute heart failure and were included in this analysis (Figure 1). The median age of the study cohort was 69 years (IQR = 55–79 years); 49% of patients were white, and 51% were African American. Patients were predominantly male (58.6%), with a history of hypertension (84%) and history of CHF (73%) (Table 1). In the overall cohort, 295 (70.6%) received diuretics in the ED and 170 (40.7%) received nitroglycerin.
|White (n = 204)||African American (n = 214)|
|Age (years), median (IQR)||75 (64-82)||61 (51-73)|
|Male, n (%)||127 (62.5)||118 (55.1)|
|History of CHF||141 (69.1)||164 (76.4)|
|History of coronary artery disease||110 (53.9)||84 (39.3)|
|History of COPD||44 (21.7)||41 (19.3)|
|History of diabetes||82 (40.2)||97 (45.3)|
|History of asthma||11 (5.4)||46 (21.6)|
|History of hypertension||166 (81.8)||186 (86.9)|
|History of chronic renal insufficiency||53 (36.1)||58 (27.1)|
|Home diuretics||149 (75.6)||154 (73.7)|
|Home ACEI/ARB||116 (58.6)||100 (48.8)|
|Home beta-blocker||134 (67.7)||112 (54.4)|
|Jugular venous distention||64 (31.4)||57 (26.4)|
|Rales||101 (49.5)||117 (54.7)|
|Lower extremity edema||132 (64.7)||130 (60.8)|
|New York Heart Association Class III or IV||123 (60.3)||125 (58.4)|
|sBP (mm Hg), median (IQR)||136 (115–158)||154 (132–177)|
|Pulse (beats/min), median (IQR)||88 (73–106)||88 (75–103)|
|Respiratory rate (breaths/min), median (IQR)||20 (18–24)||20 (18–24)|
|Oxygen saturation (%), median (IQR)||96 (94,98)||97 (94–99)|
|CXR cardiomegaly||118 (62.1)||143 (68.8)|
|CXR signs of congestion*||120 (62.8)||135 (64.9)|
|Creatinine level (mg/dL), median (IQR)||1.4 (1.1–2.6)||1.3 (0.9–1.8)|
|BUN level (mg/dL), median (IQR)||27 (18,40)||19 (14–30)|
|Hemoglobin (gm/dL), median (IQR)||12.3 (10.9–13.9)||11.9 (10.3–13.5)|
|BNP (pg/mL), median (IQR)||846.7 (437.9–1,664)||764 (401–1,774)|
|Prehospital diuretic||5 (2.3)||10 (4.7)|
|Prehospital nitroglycerin||10 (4.9)||7 (3.3)|
|Diuretic||132 (64.7)||163 (76.2)|
|Nitroglycerin||68 (33.3)||102 (47.7)|
|Final ED diagnosis of heart failure||169 (82.8)||200 (94.8)|
|ED diagnosis of myocardial infarction||43 (21.2)||69 (32.7)|
|Discharged home from ED||18 (8.8)||25 (11.8)|
|Admitted to a medical unit||40 (19.6)||54 (25.5)|
|Admitted to a telemetry unit||87 (42.6)||90 (42.5)|
|Admitted to an intensive care unit||12 (5.9)||19 (8.9)|
|30-day readmission||32 (16.3)||56 (26.4)|
|30-day death||9 (4.6)||9 (3.2)|
|30-day composite events||40 (20.3)||63 (29.6)|
Multivariate analysis demonstrated that African American race was not associated with treatment with a diuretic in the ED (adjusted OR = 1.00, 95% CI = 0.55 to 1.82; Table 2). Similar results were noted for the association with race and ED administration of nitroglycerin (adjusted OR = 1.27, 95% CI = 0.76 to 2.13; Table 3). Hosmer-Lemeshow goodness-of-fit testing provided no basis for rejecting the models as poorly fit (p = 0.88 and p = 0.39, respectively).
|Covariates||Adjusted OR||95% CI|
|BNP (pg/mL), log-transformed||1.95||1.09–3.50|
|History of CAD||0.70||0.56–1.65|
|History of COPD or asthma||0.88||0.49–1.58|
|History of CHF||1.88||0.98–3.61|
|History of CRI||0.70||0.39–1.26|
|History of HTN||0.72||0.35–1.52|
|Home diuretic use||1.51||0.81–2.84|
|Rales on exam||1.36||0.80–2.30|
|Elevated JVP on exam||2.90||1.50–5.62|
|Lower extremity edema on exam||1.55||0.91–2.66|
|Systolic blood pressure on arrival (mm Hg)||1.02||1.01–1.03|
|Oxygen saturation at presentation (%)||0.96||0.91–1.02|
|CXR with congestion*||1.02||0.59–1.76|
|CXR with cardiomegaly||1.12||0.65–1.92|
|ED diagnosis of heart failure||4.15||1.85–9.36|
|Covariates||Adjusted OR||95% CI|
|BNP (pg/mL), log-transformed||1.41||0.85–2.36|
|History of CAD||1.40||0.86–2.26|
|History of COPD or asthma||0.76||0.45–1.27|
|History of CHF||0.97||0.54–1.74|
|History of CRI||1.03||0.61–1.74|
|History of HTN||0.86||0.45–1.67|
|Home diuretic use||0.72||0.41–1.25|
|Rales on exam||1.26||0.79–2.03|
|Elevated JVP on exam||0.87||0.51–1.47|
|Lower extremity edema on exam||0.64||0.39–1.04|
|sBP on arrival (mm Hg)||1.02||1.01–1.03|
|Oxygen saturation at presentation (%)||0.95||0.91–1.00|
|CXR with congestion*||1.42||0.87–2.32|
|CXR with cardiomegaly||1.18||0.72–1.93|
|ED diagnosis of heart failure||2.33||1.01–5.42|
Thirty-day follow up was complete in 410 of 418 patients (98.1%). Readmission or death occurred in 103 patients (25.1%). In an unadjusted analysis, African American patients were more likely to be readmitted than white patients (OR = 1.84, 95% CI = 1.1 to 2.98) although there was no difference in mortality between the groups. After adjustment of baseline covariates, the association of African American race with 30-day adverse events was no longer significant (adjusted OR = 1.22, 95% CI = 0.68 to 2.16, Hosmer-Lemeshow goodness of fit = 0.56; Table 4).
|BNP (pg/mL), log-transformed||1.83||1.05–3.19|
|ED diagnosis of myocardial infarction||0.88||0.20–1.65|
|sBP arrival (mm Hg)||1.00||0.99–1.00|
|Respiratory rate (beats/min)||0.99||0.94–1.03|
|Admission sodium level (mEq/L)||0.99||0.94–1.05|
|Admission BUN level (mg/dL)||0.99||0.98–1.01|
|Admission creatinine level (mg/dL)||0.99||0.98–1.01|
|Admission hemoglobin level (g/dL)||0.89||0.79–1.01|
|ICU admission from ED||0.57||0.19–1.65|
In our study of patients who presented to the ED and were enrolled in the HEARD-IT trial who were diagnosed with decompensated heart failure, we noted no disparity in care, as defined by the use of heart failure-specific therapies. Our results differ from those of the prior studies, which largely evaluated disparities after patients were admitted to the hospital.4–11
The existence of disparity of care among racial groups has been reported for a number of cardiovascular conditions, including acute myocardial infarction (AMI) and CHF. Most of the initial studies demonstrating disparity of care between white and African American or black patients were reported in patients admitted for AMI. For example, Bradley et al.25 used the National Registry of Myocardial Infarction to show that door-to-drug and door-to-balloon times were longer for patients identified as African American when compared to those identified as white. However, a large portion of the difference was related to hospital features; after accounting for these, much of the difference was removed. In a registry study of chest pain patients presenting to 12 academic EDs, African Americans and other nonwhite patients with ACS underwent cardiac catheterization less frequently than white patients (OR = 0.45, 95% CI = 0.29 to 0.71; and OR = 0.53, 95% CI = 0.40 to 0.68, respectively).26 Data from another large observational study also noted significant disparity in acute treatment (the first 24 hours) among African Americans with ACS when compared to white patients.27
Similarly, studies evaluating the disparity in the treatment of African Americans with heart failure have found varying results. Ahmed et al.7 reported that nonwhite patients were less likely to receive care from a cardiologist when admitted for heart failure. In a retrospective review of over 2,000 Medicare beneficiaries with heart failure or pneumonia, African American patients with heart failure were noted to receive lower quality of care than other heart failure patients by both explicit process criteria and implicit review.4 However, in a retrospective analysis of medical record data systematically collected for the National Heart Failure Project, African American patients were noted to receive a quality of care similar to that of white patients. Although the African Americans had a higher rate of readmission, mortality was lower.6 In contrast, Yancy et al.11 reported that African Americans received higher quality of care with increased use of evidence-based therapies. These studies have focused largely on inpatient and discharge therapies and did not specifically evaluate ED management. Differences between studies may be reflective of the variable entry criteria for each study or the use of heart failure-specific registries that were not robust enough to allow for adjustment for treatment variations among institutions.
Identification of potential disparities in care is important as appropriate ED treatment may affect clinical outcome. A prehospital study that evaluated the use of nitroglycerin, furosemide, and/or morphine reported that mortality was lower when compared to a cohort receiving no heart failure treatment (survival OR = 2.51, 95% CI = 1.37 to 4.55).28 Peacock et al.17 compared early to delayed administration of nesiritide and found that early treatment was associated with a shorter adjusted mean total hospital length of stay (5.4 days vs. 6.9 days; p < 0.001), less likelihood to require transfer to the intensive care unit from another inpatient unit (OR = 0.301, 95% CI = 0.206 to 0.440), and increased probability of discharge to home (OR = 1.154, 95% CI = 1.005 to 1.325). Although these studies suggest that early treatment is beneficial, there are few prospective data describing the effect of ED-based acute heart failure therapy.
The lack of difference we noted in 30-day events differs from prior reports.5,6,29 It has been suggested that ED utilization varies by race, as access to care may be limited.6 In our study, we noted that African American race was not associated with a significantly increased likelihood of adverse events at 30 days. Auble et al.,29 in an observational study using single state data, reported that African American patients had a lower 30-day mortality compared with white patients (OR = 0.34, 95% CI = 0.27 to 0.49). Secondary analysis of a large heart failure registry has found results similar to ours and showed no difference in hospital readmission or mortality when adjusting for other covariates.11
This is a secondary analysis, and the primary study was not designed for this analysis or designed to be able to detect differences in mortality between groups of patients. In addition, there are limited evidence-based treatment recommendations for the management of acute heart failure in the ED. Acute therapy was at the discretion of the treating physician. All the patients were treated in academic centers, which may have standardized protocols for the management of heart failure.
The lack of disparity of treatment effect we noted may be biased by the lack of treatment options in the dyspneic patients with presumed heart failure, in addition to the small sample size studied. There is no documentation of time of medication administration and dose of agents used. Because data regarding hospital course or outpatient treatment were not collected, covariates associated with 30-day events may not be adequately accounted for in the regression model. In addition, we did not include the interventions of bilevel positive airway pressure use or intubation in our model. We therefore cannot account for their association with the use of nitrates, use of diuretics, or 30-day events. The HEARD-IT study did not require cardiac marker testing, and therefore we used the unadjudicated diagnosis of myocardial infarction as a surrogate. Elevated troponin I and troponin T values in patients with heart failure have been associated with poor prognosis, and using an ED surrogate diagnosis of myocardial infarction does not fully account for the increased risk of 30-day adverse events. In addition, we did not adjust for known left ventricular systolic and diastolic function, as these values are often not available at the time of presentation, especially in those patients who present with de novo heart failure.
In this large multicenter study we noted no differences in treatment with diuretics or nitrates in the ED between white and African American or black patients. Further study is needed to determine if this finding reflects the lack of treatment bias or the limited treatment options available for therapy in patients with acute heart failure.