Author + information
- Published online April 30, 2018.
- Marvin A. Konstam, MD∗ ()
- ↵∗Address for correspondence:
Dr. Marvin A. Konstam, The CardioVascular Center, Tufts Medical Center, 800 Washington Street, Box 108, Boston, Massachusetts 02111.
Health care costs represent a major concern for our public health and for our economy. Annual U.S. direct costs for managing heart failure (HF) are currently estimated in the neighborhood of $30 billion (1,2). With an expanding HF population, expected to reach a prevalence of at least 3% by 2030, and assuming no major change in the way we care for patients, direct costs have been projected to reach $53 billion by that same year (2). Such a rise is not sustainable and will surely be mitigated by some combination of initiatives that will force and/or incentivize major changes in practice. It will be important to prioritize targeting the greatest opportunities to achieve this goal. But discussions of cost too often omit consideration of the purpose of health care: improved outcomes. As we drive changes in our practice, it is essential to target health status and cost-effectiveness, not utilization or cost alone. One example of a nationwide effort, designed exclusively to reduce cost, that may have inadvertently worsened survival is the Center for Medicare and Medicaid Services’ 30-day readmission penalty, which targets utilization, without regard to health status or clinical outcomes, and ignores the competing risk of death (3,4).
Two studies published in this issue of JACC: Heart Failure provide needed information to help us achieve these 2 goals: targeting opportunities for cost savings, while preserving or improving health care outcomes. Ziaeian et al. (5), examined fee-for-service Medicare hospital payments during the initial hospitalization for HF with preserved left ventricular ejection fraction, and over 1 year thereafter, and observed larger payments among underserved minorities, compared with other groups. Wadhera et al. (6), examined the relationship between mortality and fee-for-service Medicare payments during a 30-day episode including and following hospitalization for HF. They found that patients treated at hospitals with higher payments tended to show reduced mortality. Despite the limitations of each study, taken together they suggest that racial and ethnic minorities are a good place to start for driving improvements in health status, potentially resulting in reduced cost, while cautioning that initiatives focusing on reduced utilization alone may adversely impact clinical outcomes and cost-effectiveness.
In this issue of JACC: Heart Failure, the primary outcome of interest in the study by Ziaeian et al. (5) was unadjusted Medicare payments, limited to part A, or hospital payments, for acute care services during the index hospitalization, at 30 days, and at 1 year. Minority populations—particularly blacks and Hispanics—had higher hospital payments than whites did. Following adjustment for patient and hospital factors, and for regional socioeconomic status, there were no significant differences across racial/ethnic groups for index hospital payments. However, at 30 days, among patients who were readmitted, blacks had significant, 9% higher relative adjusted payments, than did whites, and at 1 year, blacks (14%), Hispanics (7%), and other non-Asian minorities (24%) had significantly higher payments than did whites.
A few cautions are needed in interpreting this report. First, these findings relate to hospital fee-for-service payments only and do not include Medicare Choice plans or costs for clinician-providers, outpatient visits, medications, or testing. As the investigators (5) point out, these other items may drive approximately 40% of overall direct expenditure for HF (7). Second, these analyses are of Medicare hospital payments, which may be higher or lower than actual hospital costs. Third, the adjusted cost ratios beyond the index hospitalization apply only to patients who were rehospitalized, so do not fully reflect differences in rehospitalization rates. However, acute care service utilization, including readmission rates, cumulative length of stay, and medical procedure rates, were higher among non-Asian minorities than among whites.
It should be no surprise that difference in hospital-related factors, patient-related factors, and socioeconomic status account for a substantial degree of variation in payments across racial and ethnic groups. Hospitals generally receive a single case-based payment for a given hospitalization, based on coding for the severity-adjusted diagnosis-related group. A given hospital’s relative payment for any given diagnosis-related group is affected by a number of “hospital-related” factors, specifically the following: 1) relative wages in the geographic region (roughly, county by county); 2) “disproportionate share reimbursement,” based on the hospital’s percentage of low-income patients, relative to national average; and 3) for teaching hospitals, “indirect” payments for medical education (based on the ratio of residents to beds). These factors, not adjusted for in the primary analysis, likely aggregate by the proportion of underserved minority patients treated, particularly in urban areas. So, a substantial proportion of the increase in payments observed is likely not directly linked to the patient’s minority status. All 3 sets of covariates within the investigators’ adjustment models (5) may not have fully accounted for extrinsic factors in the analysis of payment rate ratios. There are likely unaccounted-for nuances in hospital and patient variation, and socioeconomic status was accounted for, not individually, but on a zip code basis.
However, none of these caveats detract from the investigators’ conclusions (5) of potential value of interventions to improve health status among underserved minorities. Regardless of causal basis, patients within these groups—at least those within the Medicare fee-for-service system—are at risk for greater hospital-based health care costs. Although data are provided only for a subset of patients with HF, it seems likely that the same factors are in play for multiple other disease states.
The investigators (5) appropriately call out HF disease management services and improved management of elements of the metabolic syndrome as interventions with a high likelihood of reducing cost. The first published randomized, controlled trial of HF disease management (8) showed a powerful treatment effect—31% relative reduction and 13.2% absolute reduction in hospitalized patients—within a population dominated by elderly, black women with a 76% prevalence of hypertension and a mean left ventricular ejection fraction of approximately 43%. One possible reason for the strong benefit is that patients probably had relatively little connection to primary care, so that the intervention provided many patients with their best chance of receiving optimal medication treatment. Long-term benefits of disease management appear to be linked to patient adherence (9,10), suggesting that relatively low-cost educational initiatives may go a long way. The dominant form of HF with preserved left ventricular ejection fraction, which may be referred to as metabolic-senile cardiovascular disease (11), appears closely linked to risk factors such as diabetes, hypertension, and obesity. Although more prospective data are needed, aggressive management of these conditions within minority populations may well reduce disease progression and adverse clinical outcomes, including hospitalization for HF.
The analysis by Wadhera et al. (6) generates a cautionary note regarding the approach to cost reduction in populations with HF. The observed inverse correlation between 30-day Medicare HF episode payments and patient mortality suggests that overly aggressive efforts to reduce utilization may adversely affect clinical outcomes. The findings may be linked to observations that aggressive efforts to prevent rehospitalization within 30 days of discharge may be associated with excess mortality (4). However, unlike the study by Ziaeian et al. (5), Wadhera et al. (6), who did not restrict their analysis to HF with preserved left ventricular ejection fraction, also did not limit themselves to hospital (part A) payments, but rather included payment across multiple settings, service providers, and supplies. So their findings are not impacted exclusively by hospital utilization.
Association does not prove causality, although some aspects of the study design strengthen causal inference. In particular, the observed relationship between higher payments and better outcomes was derived by relating hospital-level payments to patient-level outcomes. When the investigators (6) examined both payments and mortality at a patient level, the finding reversed; that is, higher payments for a particular patient were associated with a worse outcome, presumably because sicker patients incur more costly services. Examining payments at the hospital level mitigates some of the confounding from these associations at the patient level, because payment variation at the hospital level is much less strongly associated with individual patients’ prognosis (12). Although observational analyses are always subject to confounding, the fact that the individual-level associations go in the reverse direction diminishes concern of confounding of the primary analysis by differences in patient prognosis. Although the analysis was adjusted for hospital-specific factors, including regional costs, indirect payments for medical education, and disproportionate share reimbursement status, hospitals with higher payments likely differed in additional ways, particularly the nature of available services, such as advanced HF therapies. Hospitals offering such services tend to draw patients with different clinical characteristics and demographics, including perception of candidacy for advanced therapies, than those without such services. These population differences may have influenced the investigators’ findings (6).
There are challenges in attempting to link the findings of these 2 studies. Ziaeian et al. (5) did not examine clinical outcomes, whereas Wadhera et al. (6) did not examine differences across racial and ethnic groups. Nevertheless, each study challenges us to be more thoughtful in considering ways to reduce cost—one pointing to specific populations for which intervention might be most fruitful; the other cautioning us not to cut utilization for the sake of cutting. Combining the 2 sets of findings, we can conclude that efforts to reduce service utilization should be undertaken with careful consideration of the value of that service to the population being served. Rather than focusing on service utilization per se, interventions should be directed primarily toward improving health—as by providing disease management services and by addressing the underlying drivers of disease. If we follow this course, we will secondarily reduce utilization of costly services, while deriving the greatest value for our patients.
The author gratefully acknowledges the expert advice and assistance of David M. Kent, MD, MS, Kathleen M. Farren, and Kristine M. Hanscom, CPA, MST, CGMA, in the development of this manuscript.
↵∗ Editorials published in JACC: Heart Failure reflect the views of the authors and do not necessarily represent the views of JACC: Heart Failure or the American College of Cardiology.
Dr. Konstam has reported that he has no relationships relevant to the contents of this paper to disclose.
- 2018 American College of Cardiology Foundation
- Heidenreich P.A.,
- Albert N.M.,
- Allen L.A.,
- et al.,
- for the AHA Advocacy Coordinating Committee, Council on Arteriosclerosis, Thrombosis and Vascular Biology, Council on Cardiovascular Radiology and Intervention, Council on Clinical Cardiology, Council on Epidemiology and Prevention, Stroke Council
- Konstam M.A.
- Fonarow G.C.,
- Konstam M.A.,
- Yancy C.W.
- Ziaeian B.,
- Heidenreich P.A.,
- Xu H.,
- et al.
- Wadhera R.K.,
- Joynt Maddox K.E.,
- Wang Y.,
- Shen C.,
- Yeh R.W.
- Echouffo-Tcheugui J.B.,
- Bishu K.G.,
- Fonarow G.C.,
- Egede L.E.
- Ferrante D.,
- Varini S.,
- Macchia A.,
- et al.
- Konstam M.A.,
- Konstam V.
- Konstam M.A.,
- Abboud F.M.