Hospitals face financial pressure from decreased margins from Medicare and Medicaid and lower reimbursement from consolidating insurers.
The objectives of this study are to determine whether hospitals that became more profitable increased revenues or decreased costs more and to examine characteristics associated with improved financial performance over time.
The design of this study is retrospective analyses of U.S. non-federal acute care hospitals between 2003 and 2013.
There are 2824 hospitals as subjects of this study.
The main measures of this study are the change in clinical operating margin, change in revenues per bed, and change in expenses per bed between 2003 and 2013.
Hospitals that became more profitable had a larger magnitude of increases in revenue per bed (about $113,000 per year [95% confidence interval: $93,132 to $133,401]) than of decreases in costs per bed (about − $10,000 per year [95% confidence interval: − $28,956 to $9617]), largely driven by higher non-Medicare reimbursement. Hospitals that improved their margins were larger or joined a hospital system. Not-for-profit status was associated with increases in operating margin, while rural status and having a larger share of Medicare patients were associated with decreases in operating margin. There was no association between improved hospital profitability and changes in diagnosis related group weight, in number of profitable services, or in payer mix. Hospitals that became more profitable were more likely to increase their admissions per bed per year.
Differential price increases have led to improved margins for some hospitals over time. Where significant price increases are not possible, hospitals will have to become more efficient to maintain profitability.
The online version of this article (10.1007/s11606-018-4347-4) contains supplementary material, which is available to authorized users.
KEY WORDS: profitability, prestige, market powerU.S. hospitals continue to face significant financial pressure. Consolidation of health insurers has put pressure on commercial prices. 1 , 2 Medicare margins remain negative, 3 and Medicaid margins have fallen substantially. 4 A recent paper suggests that the share of hospitals with negative margins could rise from one-quarter to 41% in a decade. 5 Despite these challenges, average hospital margins have remained steady, and at many institutions they have increased. The average hospital’s total operating margin rose from 4.3% in 2007, the year before the Great Recession, to about 6.2% in 2014. 3
How do hospitals deal with such financial pressures, and, in some cases, thrive? The literature suggests several strategies hospitals have used in the past. Some strategies may increase revenues. One such strategy is to use accounting changes; for example, hospitals might “upcode” a patient into a more severe, and thus more reimbursed, diagnosis related group (DRG). 6 , 7 Upcoding was relatively common in the 1990s, but evidence suggests that upcoding is less prevalent now, likely due to increased government focus. 6 A second such strategy is to provide more profitable services. 8 , 9 Technologically-intensive services are reimbursed better, and hospitals historically have been posited to adopt technology to raise revenue. 10 Hospitals have continued to acquire well-reimbursed technologies in recent years, 11 , 12 though these are moving to the outpatient setting. Third, hospitals can change the mix of patients served, away from less-reimbursed public insurance enrollees towards better-reimbursed privately insured patients. 13 – 15 A fourth revenue-raising strategy is to raise prices. Hospital consolidation has increased markedly, and a rich literature links greater concentration with higher prices. 16 – 18 Having greater prestige may also allow a hospital to negotiate higher prices. 19 Other strategies may decrease costs. One such strategy is reducing length of stay. Such reductions were prominent in the 1980s and 1990s when many payers switched to prospective payment systems for inpatient care, but further reductions may be difficult. 20
In this study, we use national data to answer three questions. First, how has variation in hospital operating margins changed over time? Second, what are the characteristics of hospitals that increased their operating margin over time? Third, what financial and clinical steps did hospitals take to become profitable?
We used data from the Medicare Cost Reports from 2003 to 2013, part of the Healthcare Cost Report Information System (HCRIS). The HCRIS data have been used by the Medicare Payment Advisory Committee (MedPAC) and others to calculate hospital margins. 3 , 21 We linked HCRIS data to the American Hospital Association (AHA) annual survey, which provides information on hospital characteristics such as teaching status, and the Center for Medicare and Medicaid Services Impact Files, which provide hospital case-mix information. For a subset of analyses examining length of stay and DRG weight, HCRIS data were also linked to the Healthcare Cost and Utilization Project (HCUP) State Inpatient Databases (SID); all other analyses used our full sample. The states for which SID data were available to us for the examined time period were Colorado, Florida, Kentucky, Maryland, Massachusetts, New Jersey, New York, and Washington.
Our sample included 2824 hospitals that met the following conditions: they existed in the AHA data from 2003 to 2013; they were hospitals serving medical and surgical patients or were hospitals specializing in cancer, cardiac, or orthopedic care; they were not federal government hospitals; they were not critical access hospitals; and they submitted HCRIS data from 2003 to 2013. Analyses using the SID data included 587 of the above 2824 hospitals.
Our primary measure of performance was clinical operating margin, defined as net patient revenues minus total operating expenses (e.g., staffing costs, medications) divided by net patient revenue. This measure excludes non-clinical sources of revenue such as donations and investment income. 22 While MedPAC uses a total operating margin measure that includes these other sources of revenue, our focus on clinical operations led us to remove them. These margin measures are highly correlated. In 2013, the correlation of clinical operating margin with total operating margin was 0.77, and the correlation of the difference between 2003 and 2013 in clinical operating margin with the difference in total operating margin was 0.78.
We dropped values of operating margin in 2003 or 2013 that were greater than 50% or less than − 50% (111 hospitals affected) as these were likely due to reporting errors, replacing missing values with the adjacent year (i.e., 2004 or 2012) when possible (63 hospitals affected).
To address our first question, variation in margins over time, we examined the distribution of margins in 2003 and in 2013. We also examined the percent of hospitals that were in a particular operating margin category in 2003 that remained in that same category in 2013 and the percent that moved to a higher or lower margin category.
We then calculated overall net patient revenues per bed and overall costs per bed as well as Medicare and non-Medicare (private, Medicaid, and uninsured) revenues per bed. We estimated multivariate linear regressions relating these financial variables (our main outcomes) to our main independent variable, which was a categorical variable that took the value of 1 for hospitals in the upper quartile of change in operating margin between 2003 and 2013, − 1 for hospitals in the lower quartile of change in operating margin, and 0 for hospitals with a change in operating margin in the middle two quartiles. This categorical variable simulates the change in going from the lowest quartile of change in operating margin (mean change of − 19.1%) to the middle two quartiles (mean change of − 1.6%) or from the middle two quartiles to the highest quartile (mean change of 12.8%), allowing us to ask whether moving to a higher category of change in operating margin was associated with higher revenues, lower costs, or both. Other hospital characteristics we control for include being a large hospital (> 300 beds), profit and ownership status, rural location, teaching status, percent of admissions from Medicare, percent of admissions from Medicaid, being a specialty hospital, case mix, and state, all of which were measured in 2003.
To address our second question about which hospitals had better financial performance changes, we ran multivariate linear regressions in which our main outcomes were 1) a continuous measure of change in operating margin between 2003 and 2013, and 2) changes between 2003 and 2013 in non-Medicare revenue per bed and in expenses per bed, and our independent variables of interest were hospital characteristics. We focused on two categories of hospital characteristics: hospital prestige and hospital concentration. We defined prestigious hospitals as (1) a teaching hospital (defined as a member of Council of Teaching Hospitals); and (2) a U.S. News and World Reports top hospital in 2003, separated into (a) U.S. News “Honor Roll” hospitals, (b) U.S. News Best Hospitals for cancer, heart and heart surgery, or orthopedics that were not also “Honor Roll” hospitals, and (c) all other hospitals ranked in U.S. News that year. We used two measures for hospital concentration: (1) whether a hospital that was not part of a hospital system in 2003 joined a system prior to 2013 and (2) a hospital’s or its hospital system’s share of beds in its hospital referral region (HRR) in 2003. We regressed the main outcomes above on these measures of hospital prestige and hospital concentration in one regression, controlling for other hospital characteristics noted above.
For our third question—changes in hospital clinical characteristics associated with operating margin increases—our main outcomes of interest were (1) changes in average DRG weight; (2) changes in the number of profitable services or technology; (3) changes in hospital capacity utilization and length of stay; and (4) changes in payer mix. In separate regressions, we regressed these main outcomes on our categorical variable of quartile of change in operating margin, controlling for other hospital characteristics noted above. For DRG weight, we used the SID data to calculate the difference in the average DRG weight of an admission between 2003 and 2013. For profitable services or technology, we identified 14 services or technologies present in the AHA data between 2003 and 2013 that have been previously identified in the literature as profitable 23 and calculated the difference in number of these services or technologies between 2003 and 2013. We measured hospital capacity utilization as the average number of admissions per bed per year. For length of stay, we used the SID data to calculate the difference in average length of stay between 2003 and 2013. For payer mix, we used AHA data to calculate the difference in percent of admissions that were from Medicare, Medicaid, and other sources (private and uninsured) between 2003 and 2013.
Values were normalized to 2013 dollars using the consumer price index. Analyses were conducted using Stata 15.0 and used robust standard errors. Study approval was obtained from the National Bureau of Economic Research.
In 2013, the average hospital had a clinical operating margin of − 2.7%. By comparison, total operating margin that year in our sample averaged 5.0%. (This differs from the margin reported by MedPAC because we examine a different set of hospitals.) The distribution of clinical operating margins in 2003 and in 2013 is shown in Figure 1 . The standard deviation of margins was significantly higher in 2013 (13.6%) than in 2003 (10.6%) (p value for difference < 0.001). One quarter of hospitals in 2013 had margins of more than 5%, and the top 5% of hospitals had margins of more than 18%. The average hospital had a decrease in margin of − 2.4% between 2003 and 2013.