Charity Care and Community Benefit in Non-Profit Hospitals: Definition and Requirements.

Policymakers are using different ways to measure the community benefit provided by non-profit hospitals because different policy makers have different policy objectives. We compare 3 commonly used measures of community benefit; examine the correlation across the 3 measures; examine how the distribution of community benefits varies across non-profit hospitals; and compare the factors associated with the level of community benefit for each definition. The main dataset for this study is the Schedule H of IRS Form 990 data for 2017. We merged this data with the 2017 American Hospital Association (AHA), the 2017 CMS Hospital Cost Report, and the 2018 American Community Survey data. The final sample consists of 1904 non-profit hospitals. We define 3 measures of community benefit: (1) Total community benefits: combining all 17 possible measures in the 990 data; (2) Total community benefits less unreimbursed Medicaid care because it reflects a policy choice made by the state; and (3) only charity care. We also subdivided the community benefits into individual and service-based benefit. Gini Coefficients and descriptive analysis show the distribution of 3 types of community benefit measures. On average, hospitals spent 8.1% of their expenses on all community benefits; 4.3% on community benefits less unreimbursed Medicaid; and 1.7% on charity care. The provision of charity care showed more variation (Gini coefficient) than the other 2 measures. Different hospital and geographic characteristics were associated with each definition, suggesting that different types of hospitals place emphasis on different community benefits. When policy makers choose among different definitions of community benefit, they should consider what incentives they want to instill.

What do we already know about this topic?

Previous studies showed that some states have laws and regulations requiring hospitals to provide free or discounted care to vulnerable populations that meet minimum eligibility requirements.

How does your research contribute to the field?

Different types of hospitals emphasize different types of community benefits. This study subdivides the 17 possible types of community benefit items into 2 main categories—individual based and service-based categories. Individual-based categories provide care directly to a specific person while service-based categories offer services to the community.

What are your research’s implications toward theory, practice, or policy?

Policy makers need to pay attention to the components of community benefits that they want hospital to prioritize.

Introduction

States and localities have begun comparing the level of charity care and community benefit that non- profit hospitals provide to the level of tax benefits they receive.[1,2] Non-profit hospital systems are often the largest employer in their community and most do not pay state and local taxes. This increases the tax burden on others in the community. Some non-profit hospitals have been earning substantial annual operating surpluses and accumulating substantial financial reserves, however, on average, non-profit hospitals do not earn positive surpluses on patient care. Many of the profits are earned on non-patient care activities such as office buildings, investment income, and other related activities.

As state and local governments see ever-increasing hospital prices, operating surpluses and accumulated reserves, and larger tax needs, policymakers have begun to look at the level and types of community benefits non-profit hospitals were providing, and some wondered how they should assess community benefit. Several questions policy relevant questions emerge: (1) Do all non-profit hospitals spend similar percentages of total revenues on community benefit? (2) Does it matter which measure of community benefit is used? (3) What factors are associated with varying levels of community benefit? and (4) Which measures of community benefit are also likely to be provided by for-profit hospitals?

The legal authority governing non-profit organizations requires them to retain or reinvest or distribute any operating surplus to community, and to report data, but federal laws do not specify a definition of community benefit or a minimum amount to be provided. Despite increased policy attention to community benefit in recent years, policy makers can reach different conclusions on whether a hospitals’ community benefit exceeds their tax benefits depending on the definition of community benefit.

The Internal Revenue Service (IRS), and Centers for Medicare and Medicaid Services (CMS) require hospitals to complete forms detailing the different types of possible community benefits they provide. The forms collect slightly different data and for-profit hospitals are not required to submit 990 data to the IRS since the 990 forms are only for non-profit organizations. However, neither organization defines what a hospital must do to satisfy community benefit obligations.

Instead, there is a list of possible community benefits that the policy makers can use in evaluating the community benefits of a hospital. While the forms and questions differ slightly between the 2, empirically the differences between the 2 approaches are quite small for most hospitals. The slight differences involve how to define a hospital’s community benefit; instructions on how to measure community needs; and requirements to report financial assistance policies. Our analysis focuses on the IRS methodology because IRS provides a more comprehensive list of possible community benefits than CMS. However, we use the CMS data to get information on the community benefits provided by for-profit hospitals.

Policy makers typically choose from the 17 different types of community benefits (CB) when they develop their own measure of community benefit. The 3 most common groupings are: (1) all 17 components; (2) all 17 type of CB excluding unreimbursed Medicaid costs; and (3) only charity care and free and discounted community care. Depending on the definition chosen a specific hospital may provide more or less community benefits than its tax benefits.[5,9]

In this study, we use these 3 alternative definitions to explain the variances between hospital community benefits and hospitals characteristics. We show how they are correlated and examine how they are distributed across hospitals using a Gini Coefficient. We show how certain factors (eg, beds, revenues, racial, and ethnic makeup of community) associated with each of the 3 measures.

Method

Data

Hospital organizations are required to complete Schedule H of IRS Form 990 to show the community benefit provided by its hospital facilities and the other non-hospital health care facilities it operates.

For policymakers, the main decision is which components of community benefits to include. The IRS list includes 17 possible community benefits. We divided the 17 possible items into 2 main categories—individual-based and service-based categories. Individual-based categories provide care directly to a specific person, while service-based categories offer services to the community. An example of an individual-based category is charity care, and an example of a service-based category is unreimbursed education. Table 1 shows the different 17 types of community benefits. In the analysis, we created 3 groupings (1) all 17 types of CB, (2) 16 types of CB which eliminated unreimbursed Medicaid costs from the first group, and finally (3) only included charity care and free and discounted community care. As reported in Table 1 we categorized the 17 types of CB to 2 main categories: individual-based category and service-based category.

Table 1.

Nonprofit Hospitals’ Net Community Benefit Expense as a Percent of Total Expenses in 2017 (N = 1904).

	Non-for-profit				For-Profit
	CB, % of total hosp. expenses		CB expenses ($1 000 000)		CB, % of total hosp. expenses		CB expenses ($1 000 000)
	Mean	SD	Mean	SD	Mean	SD	Mean	SD
Individual-based service
Charity care (CHR)	1.5	2.16	4.8	11.70	0.78	1.14	1.33	2.54
Cash and in-kind contributions for community benefit (CBC)	0.17	0.74	0.94	6.66	—	—	—	—
Unreimbursed costs from other means-tested programs (UOM)	0.23	1.87	0.56	5.83	0.01	0.11	0.01	0.05
Community health improvement services and operations (CBS)	0.34	0.83	1.1	3.30	—	—	—	—
Subsidized health services, not means-tested (SHS)	1.47	3.10	2.74	7.22	—	—	—	—
Unreimbursed costs from medicaid (UMD)	3.78	4.93	12.34	31.06	1.14	1.31	0.94	1.45
Total individual-based service	7.49	6.76	22.48	48.92	1.93	1.82	2.28	3.04
Service-based benefits
Unreimbursed health professions education (UED)	0.55	1.19	4.29	17.31	0.14	0.29	0.35	0.87
Unfunded research (URS)	0.08	0.80	1.04	8.79	0.003	0.01	0.003	0.01
Building activities
Physical improvements and housing—net building expense	0.01	0.45	0.01	0.27	—	—	—	—
Economic development—net building expense	0.004	0.37	0.02	0.36	—	—	—	—
Community support—net building expense	0.02	0.16	0.04	0.35	—	—	—	—
Environmental improvements—net building expense	0.001	0.01	0.003	0.04	—	—	—	—
Leadership development and training—net building expense	0.001	0.01	0.002	0.01	—	—	—	—
Coalition building—net building expense	0.004	0.03	0.01	0.14	—	—	—	—
Community health improvement advocacy—net building expense	0.01	0.06	0.02	0.13	—	—	—	—
Workforce development—net building expense	0.02	0.10	0.05	0.35	—	—	—	—
Other—net building expense	0.004	0.04	0.01	0.10	—	—	—	—
Total building activities	0.07	0.55	0.17	0.79
Total system-based service	0.70	1.59	5.49	21.60	0.14	0.29	0.35	0.87
Total
Total charity (CHR+CBC)	1.67	2.33	5.74	16.00	0.78	1.14	1.13	2.54
Total CB without unreimbursed costs from medicaid	4.41	4.76	15.63	39.09	0.93	1.18	1.69	2.82
Total CB	8.19	7.07	27.97	65.17	2.07	1.86	2.63	3.03

Source. IRS 990, 2017; CMS, 2017.

Note.

(1) There are significant differences (P < .001) between FP and NP hospitals for all community benefits at percentage and dollar values level. (2) For for-profit hospitals, we used worksheet S-10: line 23 for CHR, lines 8 and 12 for Medicaid shortfall (UMD), line 16 for UOM. We used worksheet A line 21 (interns and residents service-salary and fringes (approved)), line 22 (interns and residents service-salary other programs (approved), and line 100 (interns and residents service-salary not approved (teaching program) for UED and worksheet A’s line 191 for URS.

In order to assess whether the same set of factors are associated with the all 17 types of community benefits, we merged the 2017 IRS data with hospital characteristics from the 2017 American Hospital Association Annual Survey Database (AHA) and 2017 Hospital Cost Report Data from CMS. Because there is no crosswalk identifier between IRS and AHA or CMS data, we merged the hospital data in 3 steps. First, we merged the data using organization name, state and city, then on state level, city and street address, and finally performed a manual review by 2 data analysts. We supplemented these data with county-level information on demographic characteristics using the 2018 Census American Community Survey. We identified 17 types of community benefits in non-for-profit hospitals using IRS-990 data. We used AHA data for measures of hospital characteristics.

For-profit hospitals also provide community benefits. To examine the level and types of community benefits used by for profit hospitals of community benefit provided by for-profit hospitals, we used CMS data to examine 5 categories of community benefits in for-profit hospitals: charity care, unreimbursed Medicaid, unreimbursed Other Means-Tested (UOM), unreimbursed education and health professions education, and unfunded research (URS) expenses. These are the only community benefits reported in the Medicare cost report.

IRS reported information for some local hospital systems as a single entity and others reported each member hospital separately, so 1 IRS identification number could correspond to several hospitals. To include each hospital in our analysis, we disaggregated the IRS 990 data for these system hospitals based on the percentage of cost that each specific hospital represents to estimate their hospital-level charity care. From reported 2380 listed non-profit organizations in IRS data we were able to match 1904 hospitals with AHA data. The remaining 476 (2380 − 1904) hospitals could not be linked with AHA and ACS data, our findings based on the IRS data.

Statistical Analysis

We examined the association between certain hospital and demographic characteristics and each measure of community benefit. To understand that how spending on charity care depends on hospital revenues, we categorized the percent of hospital revenues for each of the 3 measures into quartiles to determine if there were differences in the characteristics of hospitals by the percentage of community benefit, they provide.

We compared the level of community benefits between non-profit and for-profit hospitals. The idea was to assess which of the 17 services were also likely to be provided by for-profit hospitals. There could be a possible business reason for some of these services in addition to the community benefit. The assumption is that for-profit hospitals would provide services that benefit the hospital. We ran sets of unequal variances t-tests to compare the level of community benefits between non-for-profit and for-profit hospitals. As reported in Table 1, non-for-profits have provided higher community benefits for all types of community benefits than for-profits (P < .001).

We estimated the product-moment correlation coefficient between all 17 types of CB and the 16 types of CB (community benefits’ minus “Medicaid shortfall” and all 17 CB and charity care measures. We reported an unadjusted significance level.

Finally, To show the inequality in the level of the 3 measures of community benefit across hospitals we calculated the Gini Coefficient (GC) with jackknife standard errors, (See following equation ):

The jackknife estimate typically gives a satisfactory approximation where that analytical standard errors may not exist. The Gini Coefficient is calculated as A/(A + B). When the Gini Coefficient is 0 there is perfect equality and when the Gini Coefficient is 1 there is complete inequality. The Gini coefficient has been used in many health-related studies[13-16] to examine the level of disparity.

All data analysis was performed using Stata version 15 (StataCorp).

Results

Of the 1904 non- profit hospitals 17% were church operated; 69% were non-teaching; 63% had a system affiliation and 39% operated a trauma center. The average hospital had 191 hospital beds, 57 physicians with admitting privileges; 51% of admissions were for Medicare beneficiaries; 21% were for Medicaid beneficiaries; and 6% of hospitals were sole community providers. About 17% of hospitals were in non-concentrated markets (Herfindahel index <0.15) and they are located in communities with an average 13% poverty ratio (see Supplemental Appendix 2 for more details).

In 2017, hospitals spent an average of 8.2% of their expenses using all 17 community benefit measures. The largest portion was for unreimbursed Medicaid costs (3.8%). Hospitals spent an average of 4.3% of all of their expenses on community benefits when not including unreimbursed Medicaid costs. Hospitals spent an average of 1.7% on charity care and free and discounted community care.

The hospitals that could not be matched (20% of hospitals) had higher levels of community benefit. These hospitals spent 11.3% of their expenses on all types of community benefit, 7.1% of their expenses on community benefit less unreimbursed Medicaid, and 4.2% on charity care. These hospitals were spread across the United States; 53.5% of these hospitals are in states with Medicaid expansion programs and about half of them adopted the Federal Poverty Guidelines (FPG) equal or more than 200% FPG as a factor in determining eligibility for providing free care. These hospitals had __significantly__ lower income than matched hospitals ($370.3 vs $599.1 million; P = .0004).

In Table 1, we compared the levels of community benefits in certain categories between non-for-profit and for-profit hospitals. Across all 17 types of CB, non-profit hospitals spent on average $23.0 million annually on community benefits compared with only $2.6 million in for-profit hospitals. Comparing the spending differences across the 17 types of CB shows that the greatest differences occurred in the individual categories and the least differences in the service-related categories. Service-related categories were therefore deemed to be more likely to provide benefit to the hospital as well as the community.

The strongest correlation was between all 17 types of community benefit and the 16 measures not including unreimbursed Medicaid (.712). The correlation was lower when charity care was compared. It was only .386 between charity care and the 17 measures and .548 between charity care and the 16 measures.

Figure 1 shows the Gini coefficient for the 3 different categories of community benefits. Charity care has the highest distance to the perfect equality line suggesting that charity care is the most unevenly distributed across non-profit hospitals. The Gini coefficient was 0.546 (SE: 0.015) for charity care; 0.459 (SE: 0.009) for the 16 measures and 0.398 (SE: 0.008) in all 17 measures.

Figure 1.

Distribution of different community benefits in non-profit hospitals.

Table 2 displays the characteristics of hospitals that provide varying levels of charity care. Hospitals, which spent more on community benefit and charity care, were average size hospitals with lower revenues, located in urban areas. These hospitals were more likely located in the communities with higher black and Hispanic population, higher food stamp recipients, and higher percentages of uninsured and people 65 years and older.

Table 2.

Association between Community Benefit Spending and Selected Hospitals and Communities Characteristics.

	First quartile	Second quartile	Third quartile	Fourth quartile
	Mean/SD	Mean/SD	Mean/SD	Mean/SD
	n = 469	n = 482	n = 477	n = 476
Community benefit spending
Charity care (%)	0.15¶ (0.11)	0.65 (0.18)	1.44β (0.31)	4.44α (3.25)
CB without medicaid reimbursement	3.20¶ (3.94)	3.80 (3.63)	4.65β (3.69)	6.94α (6.94)
Community benefits (%)	5.41¶ (7.32)	7.63 (6.01)	8.57β (5.34)	10.93α (7.46)
Hospital and SES characteristics
Hospital beds	151.09¶ (170.79)	182.49 (217.06)	203.59β (226.91)	227.77α (235.90)
Not Hispanic Black (%)	8.29¶ (10.74)	7.95 (11.37)	9.58 (13.10)	13.62α (14.56)
Population without health insurance coverage (%)	7.53¶ (3.50)	7.58 (4.45)	7.62 (3.61)	10.34α (3.81)
Hispanic (%)	11.35 (13.33)	10.27 (12.87)	10.52 (11.63)	12.96 (13.36)
People with income below poverty line in last 12 months (%)	10.20≠ (4.82)	9.43 (4.33)	9.73 (4.27)	10.67 (3.83)
All household with Food Stamp/SNAP benefits in the past 12 months (%)	12.55≠ (5.90)	11.69 (5.01)	12.52 (5.37)	13.10 (5.15)
Population 65 years old and above (%)	17.89≠ (3.87)	18.31 (4.19)	18.12 (4.17)	17.16¥ (4.26)
Rural population (%)	37.03 (34.66)	36.24 (31.26)	33.30 (29.41)	28.54α (27.95)

Source. IRS 990, 2017; CMS, 2017, ACS, 2018.

Note.

We defined quartile based on the distribution of charity care spending, <25%, 26%-49%, 50%-75%, and >75%.

Unequal t-test shows a significant difference between Q1 and Q2 (¶P < .001; ≠P < .05).

Unequal t-test shows a significant difference between Q1 and Q3 (βP < .001).

Unequal t-test shows a significant difference between Q1 and Q4 (αP < .001; ¥P < 0.05).

Discussion

Non-profit status requires hospitals to retain or reinvest or distribute profits to community, but there are no specific definitions of community benefit and policy makers use a variety of definitions. The federal government also does not require hospitals to provide a specific level of community benefit to maintain their tax exemption. We show that depending on the definition of community benefit chosen, hospitals provide very different levels of community benefit.

Including all possible measure of community benefit on the IRS list shows that hospitals on average provide 8.2% of their total expenses for community benefits, other studies reported a similar number, for example, Leider et al with using 2009 to 2012 IRS-990 estimated the CB as 8.0% of total hospital expense and Young et al and Singh et al reported 7.5% and 7.2% by using 2009 IRS-990 data. Unreimbursed Medicaid costs represent almost half of this expenditure. Excluding the unreimbursed Medicaid costs drops the percentage of revenues spent on community benefit to 4.3%. State officials determine the level of Medicaid payment and some policy makers may choose not to recognize costs in excess of the Medicaid payment to be a community benefit, since the state has determined that the Medicaid payment is adequate.

Some policy makers only want to count charity care and free and discounted care as community benefits. Some might consider this to be “purest” measure of community benefit since it is care provided directly to individuals without any benefit to the hospital aside from any good will and the possibility of additional charitable contributions. The other 16 types of community benefits either reflects policy choices made by government or services that may generate benefit to the hospitals as well as the community. For example, providing continuing medical education to staff will benefit the hospital if the staff remains in the hospital. In this regard, they are not significantly different from the for-profit company having employee-training programs. Individual categories showed greater differences with for-profit hospitals than service-related categories. In addition, the Gini coefficient was greater for charity care suggesting that making the decision to provide free care was different than the decision to provide other community benefits that could also benefit the hospital.

The IRS’s definition of charity care is “free or discounted health services provided to people who meet the hospital’s financial assistance criteria and are unable to pay for the services.” CMS requires hospitals to “provide all or a portion of services free of charge to patients who meet the hospital’s charity care policy or financial assistance policy (FAP).” These 2 federal documents agree that charity care is largely determined by an individual hospital’s charity care or financial assistance policy.

In addition, some states have laws and regulations requiring hospitals to provide free or discounted care to vulnerable populations that meet minimum eligibility requirements (eg, income below specific poverty line, not being eligible to participate in any other government program; lack of private insurance or inadequate private insurance and do not living in a public institution), or resources less than $2000 and not being eligible for Medicaid. A few states___Illinois, Nevada, Pennsylvania, Texas, and Utah___require hospitals to provide a certain mandatory level of charity services for certain categories of patients.

Comparing the level of the 17 types of community benefits between non-profit and for-profit hospitals shows that for-profit hospitals provide a considerable amount of these community services in some categories and less in others. While it is not possible to know their motives, it is likely that the for-profits also generate some benefit from providing these services. Typically, for-profits and non-profits hospitals are closer in the percentages of service-related community benefit and further away on individual benefits. If the service-related benefits have a return to the hospital, then service-related services might be considered less philanthropic than individual services.

Charity care represents only 20% of the broadest definition of community benefit but represents what some policy makers may consider to be purest category of community benefit, since the other items on the list also benefit the hospital to some extent. Charity care is less correlated than the other broader definitions of community benefit and shows greater inequality than the other measures. Analysis of specific hospitals (data not shown) suggests there are significant outliers—hospital that may provide a high percentage of charity care but relatively low levels of overall community benefits. These are hospitals that devote resources without any expectation of a benefit. It is critical to examine each hospital’s level of charity care, unreimbursed Medicaid costs and all community benefits to understand their motivations.

Which of the 17 Categories Should be Included as a Community Benefit?

By law, the purpose of tax exemption is to ensure that non-profit hospitals offer community benefits. The 17 possible categories of community benefits are included on the IRS 990 Schedule H and___as explained in the method section___5 categories are reported in the Medicare Cost Report. Policy makers often consider it to be a “laundry list” of possible measures of community benefits and can choose individual items from this list. States and localities use different definitions of community benefits. For instance; the laws of California, Illinois, Indiana, Maryland, New Hampshire, New York, Rhode Island, Texas, Vermont, and Washington require tax-exempt hospitals to develop state implementation strategies (also referred to as community benefit plans, community service plans, or implementation plans) to satisfy state regulatory requirements. States standards do vary with different requirements that could be standardized.

One criterion for policy makers to consider is whether providing the service primarily benefits the hospital, the patient, or the community. One possible way is to compare the provision of community benefits between non-profit and for-profit hospitals. While this may not be a perfect measure of what constitutes community benefit, it is a measure of what for-profit hospitals perceive as a providing a financial return to the hospital.

Another approach is to consider the hospital’s responsibility for population health outcomes.[29,30] This community-based approach suggests including all 17 categories of community benefits. There is a movement to expand the list to include community building activities that “protect or improve the community’s health or safety, and that aren’t reportable in service-based activities [IRS’s Part I of schedule H].” The idea is that hospitals should promote a “culture of health” that enables all individuals to lead the healthiest possible lives. For example, hospitals providing meals to “food insecure” residents in Ohio reduces chronic disease rates by increasing participation in federally funded meal programs, showing that hospitals have become engaged in improving socioeconomic factors in their community.

Some policy makers have suggested minimum community benefit thresholds that hospitals should meet. Establishment of an explicit threshold is controversial and there is no conclusive evidence that implementing minimum community benefit thresholds results in increased community benefit investments. Another approach is a more hospital-specific requirements requiring the community benefit (or community benefit minus compensated care or charity care) to be less than the profit margin of the specific hospital.

With the recent outbreak of COVID-19, it is unclear how the provision of community benefits will change. There could be a greater need for charity care; however, the Congress has chosen to pay for the COVID 19 care for the uninsured. There is a greater need for community services such as testing facilities, but it is unclear how many hospitals have provided these services and how much they cost.

Conclusion

Policy makers are choosing what measure of community benefit to use and the metric for comparison.

The study has 2 recommendations concerning community benefit obligations for policy makers to consider. First, the components of the definition that offer the less tangible benefit to the hospital should be given priority in developing a community benefit. Second, policy makers should consider the relative benefits of an overall standard or a hospital specific standard based on the hospital’s profit and other characteristics.

Study Limitations

First, this study is based upon what hospitals have reported to IRS. Researchers have identified some limitations in the IRS data,[37,38] There are some hospitals that have not reported information to IRS. However, our current analysis and our previous analysis of the 2 possible data sets (CMS cost report and IRS-990 data) suggest that the IRS-990 data provides the most comprehensive list of 17 types of community benefits.

A previous study showed that hospital more likely to report slightly higher values on average to the IRS than Medicare to take advantage of tax exemption,[8,39] but this was not consistent across all hospitals. We were able to match above 80% of non-for-profit hospitals on the IRS list and this could affect the results.

Click here for additional data file.

Supplemental material, sj-pdf-1-inq-10.1177_00469580211028180 for Charity Care and Community Benefit in Non-Profit Hospitals: Definition and Requirements by Hossein Zare, Matthew Eisenberg and Gerard Anderson in INQUIRY: The Journal of Health Care Organization, Provision, and Financing