Author + information
- Received April 6, 2011
- Revision received June 21, 2011
- Accepted June 27, 2011
- Published online September 27, 2011.
- John R. Kapoor, MD, PhD⁎,⁎ (, )
- Roger Kapoor, MD, MBA†,
- Anne S. Hellkamp, MS‡,
- Adrian F. Hernandez, MD, MS‡,
- Paul A. Heidenreich, MD, MS§ and
- Gregg C. Fonarow, MD∥
- ↵⁎Reprint requests and correspondence:
Dr. John R. Kapoor, University of Chicago, 5841 South Maryland Avenue, Chicago, Illinois 60031
Objectives The aim of this study was to analyze the relationship between payment source and quality of care and outcomes in heart failure (HF).
Background HF is a major cause of morbidity and mortality. There is a lack of studies assessing the association of payment source with HF quality of care and outcomes.
Methods A total of 99,508 HF admissions from 244 sites between January 2005 and September 2009 were analyzed. Patients were grouped on the basis of payer status (private/health maintenance organization, no insurance, Medicare, or Medicaid) with private/health maintenance organization as the reference group.
Results The no-insurance group was less likely to receive evidence-based beta-blockers (adjusted odds ratio [OR]: 0.73; 95% confidence interval [CI]: 0.62 to 0.86), implantable cardioverter-defibrillator (OR: 0.59; 95% CI: 0.50 to 0.70), or anticoagulation for atrial fibrillation (OR: 0.73; 95% CI: 0.61 to 0.87). Similarly, the Medicaid group was less likely to receive evidence-based beta-blockers (OR: 0.86; 95% CI: 0.78 to 0.95) or implantable cardioverter-defibrillators (OR: 0.86; 95% CI: 0.78 to 0.96). Angiotensin-converting enzyme inhibitors or angiotensin receptor blockers and beta-blockers were prescribed less frequently in the Medicare group (OR: 0.89; 95% CI: 0.81 to 0.98). The Medicare, Medicaid, and no-insurance groups had longer hospital stays. Higher adjusted rates of in-hospital mortality were seen in patients with Medicaid (OR: 1.22; 95% CI: 1.06 to 1.41) and in patients with reduced systolic function with no insurance.
Conclusions Decreased quality of care and outcomes for patients with HF were observed in the no-insurance, Medicaid, and Medicare groups compared with the private/health maintenance organization group.
Heart failure (HF) is a major cause of morbidity and mortality and places a significant economic drain on the health care system (1). Mortality from HF is higher in patients with reported lower socioeconomic status (2). However, there is a lack of evidence to explain this observation. Potential disparities in the quality of health care might account for these findings, but data from the few small studies that exist are inconsistent, with some studies showing a possible association with poorer quality of care and adverse outcomes (3) and other studies failing to replicate these findings (4). Prior studies also suggest that patients with HF with different types of health insurance may receive different treatment, which may result in differences in short- and long-term outcomes (2,3). Currently, there are no consistent data that differences in quality of care on the basis of a patient's health insurance exist in the realm of inpatient management and follow-up care in patients hospitalized with HF. Determining possible differences in quality of care and outcomes in patients by insurance type is warranted to develop interventions aimed at improving adherence to HF quality-of-care measures and outcomes.
The Get With the Guidelines Heart Failure (GWTG-HF) program prospectively tracks several performance measures and other quality-of-care indicators for patients hospitalized with HF (5,6). In this study, we sought to investigate HF quality-of-care measures, length of hospital stay, and in-hospital mortality stratified by type of health insurance. The goal was to determine the association of payment status with health care quality and in-hospital outcomes in a contemporary database to identify potential targets to achieve reductions in health disparities and improve outcomes.
As previously described (7,8), the GWTG-HF program is a national, prospective, observational and ongoing voluntary clinical registry and continuous quality improvement initiative. The registry enrolls adults hospitalized with episodes of new or worsening HF as the primary reason for admission or with significant HF symptoms that developed during hospitalization for which HF was the primary discharge diagnosis. Participating hospitals are from all census regions of the U.S. and include teaching and nonteaching, rural and urban, and large and small hospitals.
Clinical information about consecutive eligible patients is submitted by participating institutions online using an interactive case report form in compliance with Joint Commission and Centers for Medicare and Medicaid Services standards. Outcome Sciences, Inc. (Cambridge, Massachusetts), serves as the data collection and coordination center for GWTG. Clinical data are abstracted using standardized definitions; variables collected include demographic and clinical characteristics, medical history, previous treatments, contraindications to therapies, and outcomes. Checks are performed to ensure that the reported data are complete and that the accuracy of data quality is monitored. Participants' institutional review boards review and approve the GWTG protocol. Sites were granted a waiver of informed consent under the common rule because data were used primarily at the local sites for quality improvement. The Duke Clinical Research Institute serves as the data analysis center.
The population for this study consisted of 106,351 HF admissions from 249 fully participating sites between January 2005 and September 2009. Patients were excluded if they were missing data on payment source (n = 5,916) or mortality (n = 927). This left a study cohort of 99,508 at 244 sites. Data were stratified into groups by payment source (Medicare, Medicaid, no insurance, and private/health maintenance organization [HMO]). Patients with Medicare along with private/HMO insurance were classified as private/HMO. Patients with Medicaid and Medicare were classified as Medicaid.
The main pre-specified quality-of-care outcomes that were measured include the core measures used by the Centers for Medicare and Medicaid Services and the Joint Commission (8,9), as follows: 1) complete discharge instructions; 2) documented evaluation of left ventricular function before arrival, during hospitalization, or planned after discharge; 3) angiotensin-converting enzyme (ACE) inhibitors or angiotensin receptor blockers (ARBs) for patients with HF with left ventricular systolic dysfunction (LVSD) without contraindications or intolerance; and 4) adult smoking cessation advice or counseling for patients with histories of smoking cigarettes. An additional GWTG-HF measure of any beta-blocker use at discharge for patients with LVSD without contraindications or intolerance was also included. Patients with documented contraindications were excluded from quality measures. Two composite measures established by the GWTG-HF program were also assessed. One was a composite quality measure—an opportunity quality-of-care index—that was based on the number of therapeutic interventions in relation to the circumstances when those interventions were indicated for the 5 measures (number of successes with quality measure/number eligible for quality measure). The other composite measure was an all-or-none measure for 100% compliance with all 5 quality measures (whether eligible patients received 100% of guideline-based therapy, up to a maximum of all 5 measures). Additional quality measures of interest included: 1) anticoagulant therapy at discharge for patients with atrial fibrillation; 2) aldosterone antagonists prescribed at discharge for patients with LVSD; 3) hydralazine nitrates in African American patients with LVSD; 4) evidence-based specific beta-blocker (carvedilol, metoprolol succinate, or bisoprolol fumarate) prescribed at discharge for patients with LVSD; 5) implantable cardioverter-defibrillators (ICDs) placed or prescribed at discharge for patients with left ventricular ejection fractions (EFs) ≤30%; and 6) ACE inhibitors or ARBs and beta-blockers for patients with LVSD at discharge. In the subgroup from 2009, the use of deep venous thrombosis prophylaxis and administration of the influenza and pneumococcal vaccinations were also assessed. Finally, length of hospital stay (number of days from admission to discharge) and in-hospital deaths were also assessed.
Baseline characteristics were compared across payment source groups using the Pearson chi-square test for categorical variables and the Kruskal-Wallis test for continuous variables. Categorical variables are reported as percents and continuous variables as median (interquartile range [IQR]). Multivariate logistic regression was performed using the generalized estimating equations methods to adjust for clustering within hospitals to determine whether payment source independently influenced each outcome and quality-of-care measure. Private/HMO was the reference group. Log transformation was used for the length-of-stay analysis to achieve an approximately normal distribution. Models were adjusted for patient characteristics and medical history and hospital characteristics. A p value <0.05 was considered significant for all tests. All analyses were performed using SAS version 9.2 (SAS Institute, Inc., Cary, North Carolina).
The total study cohort consisted of 99,508 patients hospitalized with diagnoses of HF, of whom 45,353 (45.6%) had documented reduced EFs, 47,779 (48.0%) had preserved EFs, and 6,376 (6.4%) did not have EFs documented. There were 28,702 patients (28.8%) documented to have private/HMO insurance, 55,103 (55.4%) with Medicare, 10,684 (10.7%) with Medicaid, and 5,019 (5.0%) who were uninsured or without insurance type documented. The baseline characteristics of the overall population stratified by insurance group are presented in Table 1. The median age of the overall population was 75 years (IQR: 62 to 83 years), with the oldest population seen in the Medicare group (median 78 years; IQR: 70 to 85 years) and the youngest population seen in the group with no insurance (median 53 years; IQR 46 to 61 years). Overall, this was a predominantly white population (68%), with more black HF patients seen in the Medicaid and no-insurance groups. There were also equal proportions of female and male patients in the overall population, but across groups, there were differences, with higher relative proportions of female patients seen in the Medicaid and Medicare groups, whereas more male patients were seen in the private/HMO and no-insurance groups. Although the Medicare group had more anemia, coronary artery disease (including an ischemic etiology), cerebrovascular accident or transient ischemic attack, peripheral vascular disease, and renal insufficiency, those with Medicaid had more chronic obstructive pulmonary disease or asthma, diabetes, and hypertension. Overall, the no-insurance group had the lowest prevalence of comorbidities, except for hypertension. In addition, greater proportions of the Medicaid and no-insurance groups were treated at academic hospitals and at hospitals with intervention, percutaneous coronary intervention, or surgery availability (Online Table 1).
Payment source and quality of care
Adherence to performance and quality measures varied by payment source (Online Table 2). When the cohort with LVSD was considered, the group with no insurance had the highest proportions of prescriptions for many evidence-based therapies (e.g., ACE inhibitors or ARBs at discharge and beta-blockers), while the lowest proportions were seen in patients with Medicare. Similarly, high use of evidence-based management in patients with no insurance pertained to discharge instructions, smoking cessation, the use of aldosterone antagonists, anticoagulation for atrial fibrillation, deep venous thrombosis prophylaxis, and the 100% compliance measure compared with all other groups. However, the lowest proportions of either ICD placement or a prescription for an ICD at discharge was detected in the no-insurance and Medicaid groups (31% and 38%, respectively), and the highest proportions were detected in the Medicare (42%) and private/HMO (41%) groups (p < 0.0001 for all groups). Finally, the lowest rates of influenza or pneumococcal vaccinations were also seen in the Medicaid and no-insurance groups.
Adjusted quality of care according to payment source
Differences in quality of care according to payment source persisted after generalized estimating equations multivariate regression analyses accounting for common patient and hospital characteristics and clustering of patients within hospitals (Table 2). For example, compared with the private/HMO group, the no-insurance group was less likely to receive anticoagulation for atrial fibrillation (odds ratio [OR]: 0.73; 95% confidence interval [CI]: 0.61 to 0.87), evidence-based specific beta-blockers for LVSD (OR: 0.73; 95% CI: 0.62 to 0.86), or ICD implantation or planned implantation in appropriate candidates (OR: 0.59; 95% CI: 0.50 to 0.70). Patients with no insurance were more likely to receive discharge instructions and smoking cessation counseling. The Medicaid group was also less likely to receive ACE inhibitors or ARBs and beta-blockers (OR: 0.89; 95% CI: 0.79 to 0.99) and similarly less likely to receive evidence-based specific beta-blockers (OR: 0.86; 95% CI: 0.78 to 0.95) or ICDs for LVSD (OR: 0.86; 95% CI: 0.78 to 0.96). Finally, the Medicare group was less likely to receive ACE inhibitors or ARBs (OR: 0.85; 95% CI: 0.76 to 0.95) or the composite of an ACE inhibitor or ARB and a beta-blocker (OR: 0.89; 95% CI: 0.81 to 0.98).
Payment source and length of hospital stay
After multivariate adjustment, compared with the private/HMO group, there were longer associated hospital stays in the Medicare, Medicaid, and no-insurance groups (Table 3). There were also longer associated hospital stays after adjustment in patients on Medicaid, those on Medicare, and those with no insurance compared with the private/HMO group among patients with preserved systolic function and in patients on Medicare and those with no insurance among patients with LVSD (Online Table 4).
Payment source and in-hospital mortality
There were 2,944 in-hospital deaths (3.0%) during the study (Online Table 2). After multivariate adjustment for patient characteristics and laboratory values, a higher overall mortality rate was seen in the Medicaid group (OR: 1.22; 95% CI: 1.06 to 1.41) (Table 3). A higher rate of in-hospital mortality was again seen in the Medicaid group in patients with preserved EFs and in the no-insurance group in patients with LVSD (Online Table 4).
Among hospitals participating in GWTG-HF, we found significant differences in the application of evidence-based care and in-hospital outcomes by payment source in this large contemporary cohort of patients hospitalized with HF throughout the U.S. Specifically, higher adjusted rates of in-hospital mortality were seen in patients with Medicaid and in patients with no insurance (in the group with LVSD). Similarly, compared with patients with private/HMO insurance, longer HF hospitalization adjusted length of stay was seen in patients with Medicaid, those with Medicare, and those with no insurance. In addition, compared with patients with private/HMO insurance, we found that patients with no insurance, Medicaid, or Medicare less often received some of the guideline-recommended therapies that are currently included in the HF performance and quality measures. After adjustment for potential confounders, there was lower use of the composite of ACE inhibitors or ARBs and beta-blockers in patients on Medicare and Medicaid. Patients with Medicaid were also less likely to receive smoking cessation instructions, evidence-based specific beta-blockers for LVSD, and ICDs in eligible patients. Having no insurance was similarly associated with lower use of anticoagulation for atrial fibrillation or evidence-based beta-blockers for LVSD and a decrease in the odds of receiving ICDs in eligible patients. Interestingly, patients without insurance were more likely to receive discharge instructions and smoking cessation counseling. Medicare patients were most likely to receive ICD placement, which may reflect the financial impact of ICD placement in patients without insurance or with Medicaid on providers. These findings suggest that even among hospitals participating in a national quality improvement program, HF patients' insurance status is still associated with the care provided and clinical outcomes in the inpatient setting.
Despite major advances in HF treatments, access to evidence-based care may be limited by patients' insurance status. Limitations in or complete lack of health insurance may influence access to and delivery of care. However, there is a lack of studies assessing this in patients with HF. One study demonstrated a significantly higher admission rate in a managed care cohort compared with groups with other payment sources (10). However, there was no association between managed care and poor short-term outcomes of hospitalization. Our study adds significant insight demonstrating that insurance status is significantly associated with the use of guideline-recommended HF therapies and in-hospital clinical outcomes in a very large contemporary cohort of patients with HF.
Smaller studies of other populations have reported disparities in health care quality and outcomes on the basis of socioeconomic status (3). Studies have demonstrated a link between socioeconomic inequalities and cardiovascular disease mortality (11) and a decreased likelihood of receiving regular medical care (12). Although insurance status is not synonymous with socioeconomic status, they are inter-related. There is a lack of studies assessing the association of socioeconomic status and/or payer status with HF quality of care and outcomes, and a few small studies have yielded opposing findings. Higher associated readmission rates were demonstrated among lower socioeconomic groups in other studies (13). However, findings from other studies on HF are inconsistent, with some failing to demonstrate an association between socioeconomic status and outcomes (4).
The prevalence of HF and mortality has been reported to be higher in those with lower socioeconomic status, but there is little evidence to explain this observation (2). It has been proposed that uncorrected risk factors such as smoking, hypertension, coronary artery disease, and diabetes mellitus in the lower socioeconomic groups may account for this finding (13). Indeed, several of these factors are seen more often in lower socioeconomic groups, supporting this notion (14). However, differences among groups were seen after multivariate analyses in our study adjusting for many of these factors. Another possibility is that differences in the implementation of guideline-endorsed HF therapy, as we saw in this study, may in part explain this observation. Our data demonstrate that significant differences exist in the implementation of guideline-endorsed HF therapy and outcomes according to payer status, with decreased quality and outcomes in patients with no insurance, Medicaid, and Medicare compared to the private/HMO group. A bias not to prescribe drugs or therapies with lifesaving benefits to certain groups can certainly perpetuate the observed increases in HF prevalence and poor outcomes. The reasons for disparate prescribing behavior on the basis of payer status are unknown. Access inequalities to specialist care during hospital admissions may explain some of the differences (3). One study demonstrated an increased rate of rehospitalization in patients with HF with lower socioeconomic status that was independent of disease severity, suggesting that socioeconomic status may influence the clinical management of HF (13). The same study also showed that the increased rate of rehospitalization was independent of noncompliance with diuretic agents, arguing, at least in the case of diuretic agents, against medication noncompliance as a reason for increased morbidity in patients with lower socioeconomic status. The reasons behind the disparities warrant further investigation to help mitigate associated poorer outcomes in patients with lower socioeconomic status. Finally, it is unclear how the quality of health care and outcomes will be affected in the era of health care reform that is expected to expand insurance coverage to more Americans, including an expansion of Medicaid eligibility.
First, although insurance status was collected and analyzed, there were no direct measures of socioeconomic status or mechanism to conclude why certain higher priced interventions were not advocated. For example, refusal because of inability to pay may have been present and might have affected the results. Second, the lack of follow-up after discharge does not allow assessment of long-term outcomes. Third, data were collected by medical chart review and depend on the accuracy and completeness of documentation and abstraction. Although contraindications and intolerance to medications were recorded as documented in the medical record, there may have been patients with contraindications or intolerances to treatments that were present but not documented, particularly for less well established quality measures. Given the observational nature of the study, residual measured and unobserved variables may have confounded the results. Although the generalized estimating equations multivariate analyses adjusted for multiple baseline differences, selection bias influencing physicians' and patients' decision making may influence these findings. Furthermore, although this was a registry-based study with an opportunity to study patients in the real-world setting, data collection was dependent on the voluntary participation of hospitals, such that findings may not be generalizable to hospitals that differ in care patterns or patient characteristics. Additionally, although it is likely that socioeconomic status correlates with insurance type, this study could not distinguish whether these findings were influenced by payment model, socioeconomic status, or both. Finally, because of the large number of patients in this study, small differences might have led to statistical significance but lack clinical relevance.
Using data from the GWTG-HF quality program, the results of this study suggest that the implementation of guideline-endorsed HF therapy and in-hospital outcomes are associated with payer status, with decreased quality and outcomes in patients with no insurance and those on Medicaid and Medicare compared with those with private/HMO insurance. Addressing these differences in care and outcomes will require additional efforts.
For supplemental tables, please see the online version of this paper.
The Get With the Guidelines Heart Failure (GWTG-HF) program is provided by the American Heart Association (AHA). The GWTG-HF program is currently supported in part by Medtronic, Ortho-McNeil, and the AHA Pharmaceutical Roundtable. GWTG-HF has been funded in the past through support from GlaxoSmithKline. Dr. Hernandez has received research support from Johnson & Johnson, Proventys, and Amylin Pharmaceuticals. Dr. Heidenreich has received grant support from Medtronic. Dr. Fonarow has received research support from the National Heart, Lung, and Blood Institute (significant), is a consultant for Novartis (significant) and Scios (modest), and has received an honorarium from Medtronic (modest). All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- angiotensin-converting enzyme
- angiotensin receptor blocker
- confidence interval
- ejection fraction
- Get With the Guidelines Heart Failure
- heart failure
- health maintenance organization
- implantable cardioverter-defibrillator
- interquartile range
- left ventricular systolic dysfunction
- odds ratio
- Received April 6, 2011.
- Revision received June 21, 2011.
- Accepted June 27, 2011.
- American College of Cardiology Foundation
- McCullough P.A.,
- Philbin E.F.,
- Spertus J.A.,
- Kaatz S.,
- Sandberg K.R.,
- Weaver W.D.
- Fonarow G.C.,
- Abraham W.T.,
- Albert N.M.,
- et al.
- Mackenbach J.P.,
- Cavelaars A.E.J.,
- Kunst A.E.,
- et al.
- Struthers A.D.,
- Anderson G.,
- Donnan P.T.,
- et al.
- Connolly V.,
- Urwin N.,
- Sherriff P.,
- Bilous R.,
- Kelly W.