In an analysis of state all-payor data we found that, on-average, not for-profit hospitals provide a similar level of uncompensated care as for-profit hospitals for patients hospitalized with three common diagnoses. These findings are noteworthy given the growing concern that certain NFP hospitals may not provide enough charity care to justify the generous tax exemptions that they receive. Alternatively, our finding that government hospitals provide significantly more uncompensated care than either NFP or FP hospitals is a testimony to the vital role that these hospitals play the U.S. healthcare system.
A number of aspects of our study merit further discussion. First, it is important to explain why the tax exempt status of not-for-profit hospitals is so important. America's approximately 4,200 NFP hospitals receive an estimated $6-$8 billion in tax exemptions per year (1995 dollars)-- an average of $1.6 million per hospital [35
]. It is presumed by policy makers that these tax savings will be used by NFP hospitals to support activities that provide so-called "community benefit" [36
]. Community benefit, while difficult to quantify, typically encompasses a range of activities including provision of care to the uninsured, community outreach activities, research, and teaching [36
]. Quantifying community benefit is methodologically challenging because many components (e.g., hospital outreach activities, research, and teaching) are not readily and reliably captured in commonly available data sources. Thus, in practice, most research and policy evaluations of the quantity of community benefit provided by hospitals typically relies upon study of the amount of charity care provided to the underinsured and uninsured. There are a relatively limited number of data sources that contain data on the quantity of charity care provided by hospitals; these sources include surveys collected by hospital organizations (e.g., the AHA annual survey) and financial reports derived from information that hospitals are required to submit to state and federal regulators (e.g., Internal Revenue Service Schedule H for hospitals)[2
]. While these data are certainly useful, there is emerging evidence that hospitals are not so different from private sector businesses when it comes to the pressures to manipulate financial data in unseemly ways [28
]. In particular, NFP hospitals have recognized that their financial statements are scrutinized by regulators and the media. Hospitals with excessive profitability or minimal amounts of charity care in these reports are increasingly targeted for financial audits and even penalties. Given these incentives to "manage" earnings, we believe we believe that our analysis-focusing on the actual insurance coverage of patients admitted to hospitals as opposed to hospital financial statements-provides an important new method for evaluating charity care provided by hospitals.
Second, it is important to discuss our finding that NFP hospitals do not appear to provide a greater level of uncompensated care than FP hospitals. We initially suspected that the lower proportion of uninsured patients admitted to NFP hospitals in unadjusted analyses might reflect the neighborhoods where not-for-profit hospitals were located [26
]. However, even after adjusting for patient, hospital, and market-level factors we found no evidence that NFP hospitals provide a greater level of uncompensated care than FP hospitals. Thus, other explanations must be explored. One potential explanation is that uninsured patients preferentially seek care from for-profit and government owned hospitals [40
]. It is easy to understand why uninsured patients might prefer government operated safety net hospitals that often have an explicit mission to provide medical care to vulnerable populations, but it is less clear why uninsured patients would prefer for-profit hospitals. Another possible explanation for our finding that not-for-profit hospitals perform less charity care comes from the work of Weiner et al. who have found evidence that hospitals have implicit policies and strategies to minimize the amount of care provided to uninsured populations [42
Third, it is important to consider our findings in the context of how we defined uninsured patients in this analysis. We defined uninsured patients as those patients whose payment source was categorized as either charity care or self-pay in the SID data in accordance with methods that have been employed in prior studies [43
]. There is some evidence that these two groups of patients differ with individuals classified as self-pay being more likely to represent either individuals who have declined employer sponsored health insurance or the so-called working poor while individuals categorized as charity care may be more likely to represent the truly indigent [44
]. That said, from a hospital perspective both charity care and self-pay patients constitute populations for whom hospitals are less likely to receive adequate reimbursement for services that are provided and thus combining these groups for analytic purposes seems justified [46
Finally, we would like to briefly comment on two additional findings. Our finding that for-profit hospitals had higher nurse staffing levels than both not-for-profit and government owned hospitals is important. While conventional wisdom and older studies have suggested that that for-profit hospitals may reduce nurse staffing as a strategy to maximize profits more recent studies have demonstrated a more complex picture with reduced levels of nurse staffing in both for-profit and not-for-profit hospitals. Our finding of higher nurse staffing levels in for-profit hospitals in combination with our principal findings that for-profit hospitals do not admit fewer uninsured patients adds to evidence that for-profit hospitals may not place profitability above other important aspects of hospital quality. It is also important to briefly comment on our finding that more than 40% of all women admitted for childbirth were primarily insured by Medicaid. While on first glance this appears extremely high, these data are consistent with prior studies and likely reflect a combination of the erosion of employer based health insurance combined with the socioeconomic challenges faced by younger mothers.
There are a number of limitations to our study that merit brief mention. First, our study relied upon administrative data and thus may have been subject to bias if insurance status was systematically miscoded more often by one group of hospitals, though we have no reason to suspect this to be the case. Second, our analysis was limited to patients admitted with three different diagnoses to hospitals in ten states and therefore should be extrapolated to other diagnoses and states with caution. Further study is needed to verify the generalizability of our study to other conditions. Third, our study focused on amount of community benefit provided by hospitals as measured by the proportion of admissions who were categorized as uninsured and did not assess other aspects of community benefit that are important to consider. Nevertheless, in practice community benefit is commonly assessed through the measurement of charity care or uncompensated care and thus we believe that our analysis is germane to the current political debate.
In summary, we found no evidence that not-for-profit hospitals admit a higher percentage of uninsured patients than for-profit or government-owned hospitals, for the three conditions studied. While we cannot comment on other types of community benefits that not-for-profit hospitals may provide, concerns about the tax exempt status of not-for-profit hospitals may be warranted.