Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to Ethnicity and Race
Summary
U.S. Department of Health and Human Services Office of Inspector General report OEI-05-20-00580, July 2023, Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of Populations Served, under Ann Maxwell, Deputy Inspector General for Evaluation and Inspections. The Report in Brief states that communities with greater concentrations of Hispanic/Latino residents were generally associated with less 2020 targeted provider relief funding for hospitals than communities with lower concentrations. It states that high concentrations of Non-Hispanic Black residents were sometimes associated with more funding and that funding levels were not found to be associated with poverty. It describes the method of translating hospital allocations into estimated PRF per person amounts for each U.S. census tract.
Summary drafted by a model from the document's text below and checked by script against that text before publication. It is a navigation aid, not a reading of what the document proves. Where AI is used
- Issuer
- Source documents
- Document type
- Targeted Prf Allocated To Hospitals Had Differences By Ethnicity Race
- Case
- Targeted Prf Allocated To Hospitals Had Differences By Ethnicity Race
Cite this document
“Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to Ethnicity and Race.”
Original at oig.hhs.gov. Archived at Pandemic Darlings, https://pandemicdarlings.com/source-documents/targeted-prf-allocated-to-hospitals-had-differences-by-ethnicity-race/
, SHA-256 5940ca2f854fc5c463e178d139671a69157d1e7395cec6fb8c5f85af8374ac54.Full text
U.S. Department of Health and Human Services
Office of Inspector General
Targeted Provider Relief Funds
Allocated to Hospitals Had Some
Differences with Respect to the
Ethnicity and Race of Populations
Served
Ann Maxwell
Deputy Inspector General
for Evaluation and Inspections
July 2023, OEI-05-20-00580
U.S. Department of Health and Human Services
Office of Inspector General
Report in Brief
July 2023, OEI-05-20-00580
Targeted Provider Relief Funds Allocated to Hospitals Had Some
Differences with Respect to the Ethnicity and Race of Populations
Served
Why OIG Did This Review
Key Results The COVID-19 pandemic highlighted longstanding inequities,
Communities with greater like disparities in funding levels for health care providers by
concentrations of Hispanic/Latino locations and populations served, as well as inequitable access
residents were generally associated to quality health care. 1 From the beginning of the pandemic,
with less 2020 COVID-19 targeted reports indicated that people of color (e.g., Hispanic/Latino
provider relief funding for hospitals and Black Americans) and people from economically
than communities with lower
disadvantaged communities were at greater risk of COVID-19
concentrations. This funding disparity
exposure, illness, hospitalization, and death than members of
raises concerns because
Hispanic/Latino Americans have been
predominantly Non-Hispanic White communities. 2, 3, 4, 5, 67 In
more likely to be hospitalized or die addition, the Centers for Disease Control and Prevention (CDC)
from COVID-19 than non-Hispanic considers Hispanic/Latino ethnicity and Black race to be
White Americans. associated with social vulnerability, along with external
stressors such as poverty and poor housing conditions. This
On the other hand, high concentrations
of Non-Hispanic Black residents were means that communities with greater concentrations of
sometimes associated with more Hispanic/Latino residents, greater concentrations of Black
funding, and funding levels were not residents, and/or higher rates of people experiencing poverty
found to be associated with poverty. may be at a greater risk of experiencing long-term financial
hardship due to disease outbreaks. 8, 9
In April 2020, the U.S. Department of Health and Human Services (HHS) began distributing Provider
Relief Fund (PRF) payments through the Health Resources and Services Administration (HRSA) to
support health care providers, including hospitals, on the front line of the pandemic response. To
respond to the urgent need for health care funding, Congress required HHS to make PRF payments
using the most efficient payment systems practicable. HHS had to make decisions quickly about how to
allocate money in accordance with statutory criteria associated with the funds.
HHS has placed a priority on promoting health equity and reducing health disparities during the
COVID-19 pandemic, including in the distribution of resources. 10, 11 While the PRF was not designed
with the goal of addressing health disparities, understanding how early PRF Targeted Distributions
(Targeted PRF) correlated with racial, ethnic, and economic characteristics of the communities providers
serve can help to inform decisions for future public health funding and the opportunities they present
to advance the health equity goals of HHS.
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 1
OEI-05-20-00580
How OIG Did This Review
To analyze hospital funding according to populations served, we took allocations to hospitals from the
Targeted PRF in 2020 and translated them into estimated “PRF per person” amounts for each U.S.
census tract (in this report, we also refer to census tracts as “communities”). To do so, we used
Medicare data about the census tracts served by each hospital, and assigned each hospital’s funding
allocations to those census tracts proportionately. We then determined whether there were statistically
significant correlations between PRF per person and the racial, ethnic, and economic composition of the
census tracts. To account for other community characteristics that could help explain differences in PRF
per person, we analyzed rural and nonrural census tracts separately. We conducted this analysis for the
approximately $44 billion in Targeted PRF allocated to hospitals in 2020 through the four allotments
designated for 1) COVID-19 High Impact Area Hospitals, 2) Safety Net Hospitals, 3) Rural Hospitals, and
4) Indian Health Service and Tribal Hospitals.
What OIG Found
Our analysis identified some differences in PRF per person across census tracts with respect to the race
and ethnicity of their residents, but not with respect to poverty rates. Specifically, when we analyzed all
four Targeted PRF allotments combined, we found that communities with greater concentrations of
Hispanic/Latino residents were associated with less PRF per person than communities with smaller
concentrations of Hispanic/Latino residents. In nonrural areas, communities with greater concentrations
of Non-Hispanic Black residents were associated with more PRF per person than communities with
smaller concentrations of Non-Hispanic Black residents, but this pattern did not occur in rural areas.
We did not find a meaningful association between PRF per person and the proportion of residents
experiencing poverty in the community.
When we analyzed the four Targeted PRF allotments individually, we found the most notable trends in
the allotment targeted to rural hospitals (about $9.7 billion): Communities with greater concentrations
of Hispanic/Latino residents or Non-Hispanic Black residents were associated with less PRF per person
than communities with smaller concentrations of Hispanic/Latino Residents or Non-Hispanic Black
residents.
Why This Matters
Differences in hospital funding with respect to the characteristics of the populations hospitals serve—
including race and ethnicity—could potentially exacerbate pre-existing disparities in health outcomes.
If hospitals that serve populations experiencing disparate health outcomes are under-resourced, those
populations may be left with less access to high-quality care, which could widen gaps in health
outcomes.
Health care funding is an important tool that can help HHS contribute to goals of reducing health
disparities, both in the context of COVID-19 and more broadly. We hope that this analysis is useful to
HHS in planning for future emergency funding scenarios and identifying opportunities to support these
goals, to the extent permitted by law.
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 2
OEI-05-20-00580
BACKGROUND
The Provider Relief Fund
The Provider Relief Fund (PRF) refers to $178 billion in funds appropriated by
Congress in 2020 to prevent, prepare for, and respond to COVID-19, and reimburse
health care providers for expenses or lost revenues related to COVID-19. Congress
did not require that PRF be allocated in a way that addressed health disparities or
build in considerations about which populations were impacted by funding
decisions. 12 Congress delegated PRF payment calculation and disbursement to HHS
and required HHS to make PRF payments using the most efficient payment systems
practicable. 13
HHS distributed the PRF through a series of payments (hereafter referred to as
allotments) to eligible hospitals and other health care providers with the requirement
that funds be used for health-care-related expenses or lost revenue attributable to
COVID-19. HHS designed the General Distributions to provide financial relief to
providers, including hospitals and other facilities, during the COVID-19 pandemic
based on provider characteristics such as Medicare reimbursements and Medicaid
and CHIP participation. 14 Targeted Distributions were designed to provide financial
relief and address COVID-19 challenges for specific types of hospitals and other
facilities based on characteristics such as high numbers of COVID-19 admissions and
rural or tribal designations. 15 HHS worked with The Office of the Assistant Secretary
for Planning and Evaluation and Health Resources and Services Administration
(HRSA), along with multiple stakeholders including third-party contractors, to
administer the PRF. In April 2020, HHS designated HRSA as the primary operating
division in charge of coordinating and overseeing the creation of payment files,
calculation of payment allocations for eligible health care providers and facilities, and
disbursement of funds for allotments of the General Distributions and Targeted
Distributions.
For additional details about the PRF, see Appendix A.
Scope and Limitations of Our Evaluation
We analyzed the following Targeted PRF allotments that were made to hospitals in
2020:
COVID-19 High Impact Area Hospitals (hereafter refered to as High Impact
Hospitals)
• HRSA determined eligibility for this allotment based on hospitals that had
high numbers of confirmed COVID-19 positive inpatient admissions in
January-June 2020.
Safety Net Hospitals
Rural Hospitals
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 3
OEI-05-20-00580
Indian Health Service and Tribal Hospitals (hereafter referred to as Tribal
Hospitals)
For our analysis, we excluded Targeted PRF amounts that were allocated to facilities
other than hospitals (such as nursing homes), as well as Targeted PRF amounts that
were allocated to children’s hospitals.
As a result of focusing on these specific allotments, we analyzed $43 billion of the
$46 billion total Targeted PRF allocations that went to hospitals in 2020.
We could not assess patterns
Exhibit 1: The allotments we reviewed
in allocations to hospitals
totaled over $40 billion in Targeted PRF,
from General Distributions.
Because these distributions and the allotment targeted at High Impact
were allocated to parent Hospitals accounted for almost half that
companies, data for individual amount.
hospital allocations do not
exist in the case of hospitals in
hospital systems. The extent
to which inclusion of General
Distribution allocations may
have resulted in different
outcomes than those we
identified is unknown.
We did not review other relief
funding that HHS is
distributing outside the PRF,
such as funds targeted to
rural communities through
the American Rescue Plan.
The extent to which additional
funding may have resulted in
Source: OIG analysis of subset of Targeted PRF allocations after
different outcomes than those
exclusions.
we identified is unknown.
We were only able to analyze the amounts HRSA allocated to hospitals. Therefore,
for the subset of hospitals in our analysis that are owned by a parent organization,
we are not able to see where the funds were ultimately distributed. After HRSA
calculated allocations for each hospital, HRSA then combined the allocations and
distributed a single payment to parent organizations for all of their hospitals. Parent
organizations were given flexibility to re-distribute funding to the facilities under
their umbrellas at their discretion.
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 4
OEI-05-20-00580
This report describes correlations between Targeted PRF per person and census tract
characteristics. Our analysis should not be read to imply that differences in census
tract characteristics caused differences in PRF per person. Additionally, we did not
assess compliance with Federal requirements.
Methodology in Brief
To compare funding across populations, we took Targeted PRF allocations to
hospitals and translated them into estimated “PRF per person” amounts for each U.S.
census tract a based on data about the census tracts served by each hospital. We used
Medicare hospital claims to identify the census tracts where each hospital’s patients
live. b We attributed each hospital’s Targeted PRF allocations to census tracts
proportionally to the number of the hospital’s Medicare claims for patients living in
each census tract. Finally, we divided the Targeted PRF amount we attributed to each
census tract (i.e., from allocations for all hospitals that served the tract) by the
population of the tract. See Exhibit 2 for a diagram illustrating the hospital service
area methodology.
_____________________________________________________________________________________________________________________________
a After excluding tracts in the territories and with data anomalies, we examined PRF per person for 70,342
census tracts (96.2% of tracts in the 50 states + DC).
b We linked hospital allocations to census tract populations based on Medicare hospital utilization
patterns because we do not have access to comprehensive nationwide utilization data from other payers.
For Medicaid, although we have access to hospital claims, we were not able to match these to patients’
residence census tracts. We conducted a sensitivity test to ensure that this approach did not
meaningfully impact our results. See Methodology for more information.
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 5
OEI-05-20-00580
Exhibit 2: Model to illustrate the hospital service area methodology we used
to calculate each census tract’s PRF per person
After calculating PRF per person for each census tract, we determined whether PRF
per person correlated with these demographic and economic characteristics of each
census tract: racial and ethnic composition and poverty rates. We performed this
analysis for the four Targeted PRF allotments combined, as well as for each of the
allotments individually.
To account for differences between rural and nonrural communities, we analyzed rural
and nonrural census tracts separately based on HRSA’s designations. We examined
PRF per person for 14,814 rural census tracts and 55,528 nonrural census tracts.
Finally, we created graphs to illustrate the correlations that we found between PRF per
person and certain demographic characteristics (e.g., race and ethnicity). We split
census tracts into four groups based on the shares of specific races or ethnicities and
then compared the PRF per person for each group.
See Methodology section for additional details.
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 6
OEI-05-20-00580
RESULTS
Across the Targeted PRF allocations we analyzed, both rural and
nonrural communities with more Hispanic/Latino residents had
less PRF per person
We analyzed Targeted PRF amounts across all four allotments to
We analyzed the
identify whether there were any overall funding differences when the
relationship between
allotments were combined and considered together. We identified
community
some notable patterns in PRF per person related to race and ethnicity,
characteristics and OIG’s
which we describe below in detail. Refer to Appendix C for detailed
estimate of PRF per
results of all community characteristics that we analyzed.
person, which is based
on Targeted PRF Rural and Nonrural Communities with Greater Concentrations of
allocations to hospitals Hispanic/Latino Residents
in 2020 and data about
the census tracts served We found census tracts with greater concentrations of Hispanic/Latino
by each hospital. residents were associated with less PRF per person than census tracts
with smaller concentrations of Hispanic/Latino residents. We
identified the same trend among both rural and nonrural census tracts, but these
differences in PRF per person were larger among rural census tracts.
Exhibit 3: In both rural and nonrural census tracts, we found that
communities with more Hispanic/Latino residents had less PRF per person.
Notes: The bar graphs illustrate the correlations described in the main text of the Results. “Pop” represents the total
population that lives in each group of census tracts.
Source: OIG analysis of PRF allocations, HHS Protect Data, ACS Data, Medicare hospitals claims and enrollment data,
USA Facts COVID-19 data, 2021
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 7
OEI-05-20-00580
In both groups of census tracts, we identified a negative correlation between the
proportion of Hispanic/Latino residents in a census tract and the PRF per person (rural
correlation coefficient = -0.154, p < .0001; nonrural correlation coefficient = -0.088 c, p
< .0001 d,), which means census tracts with greater concentrations of Hispanic/Latino
residents were associated with less PRF per person than census tracts with smaller
concentrations of Hispanic/Latino residents. These patterns are clear when we split
census tracts into groups by their share of Hispanic/Latino residents to represent the
correlation results visually, as shown in Exhibit 3. In rural census tracts, census tracts
with greater than 15 percent Hispanic/Latino residents were allocated about 16
percent less PRF per person than census tracts with less than 5 percent
Hispanic/Latino populations. We found a funding difference among nonrural census
tracts but with the same notable pattern: nonrural census tracts with greater than 15
percent Hispanic/Latino residents were allocated about 6 percent less PRF per person
than nonrural census tracts with less than 5 percent Hispanic/Latino residents.
See Appendix C for further statistical analysis.
Across the Targeted PRF allotments we analyzed, nonrural
communities with more Non-Hispanic Black residents had more
PRF per person
When we analyzed Targeted PRF amounts across all four allotments, we identified a
meaningful association for nonrural census tracts, but we did not identify a
meaningful association between PRF per person and the concentration of Non-
Hispanic Black residents for rural census tracts.
Nonrural Communities with Greater Concentrations of Non-Hispanic Black
Residents
We found nonrural census tracts with greater concentrations of Non-Hispanic Black
residents were associated with more PRF per person than census tracts with smaller
concentrations of Non-Hispanic Black residents.
_____________________________________________________________________________________________________________________________
c OIG identified correlations with coefficients less than -0.05 or greater than 0.05 as representing
meaningful associations. For example, some researchers may consider only stronger correlations, such as
those with coefficients less than -0.10 or greater than 0.10, as representing meaningful associations.
d Generally, a p-value less than 0.05 indicates a statistically significant correlation. Correlation coefficients
represent the strength of the correlation: coefficients closer to -1.0 indicate a stronger inverse
relationship, a coefficient of 0 indicates no relationship, and coefficients closer to 1.0 indicate a stronger
direct relationship.
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 8
OEI-05-20-00580
Exhibit 4: In nonrural census
tracts, we found that
communities with more Non-
Hispanic Black residents had
more PRF per person.
We identified a positive
correlation between the
proportion of Non-Hispanic Black
residents in a nonrural census
tract and PRF per person
(correlation = 0.136, p < .0001).
Nonrural census tracts with
greater than 15 percent Non-
Hispanic Black residents were
allocated about 32 percent Notes: The bar graph illustrates the correlations described in the
main text of the Results. “Pop” represents the total population that
more PRF per person than lives in each group of census tracts.
census tracts with less than 5
percent Non-Hispanic Black Source: OIG analysis of PRF allocations, HHS Protect Data, ACS
Data, Medicare hospitals claims and enrollment data, USA Facts
populations. COVID-19 data, 2021
Rural Communities and Concentrations of Non-Hispanic Black Population
When we analyzed the Targeted PRF amounts across all four allotments for rural
census tracts, we did not identify a meaningful association between the PRF per
person and concentrations of Non-Hispanic Black residents. The correlation between
the proportion of Non-Hispanic Black residents and PRF per person is close to zero,
indicating no meaningful association (correlation = 0.021, p = 0.01).
Across the Targeted PRF allotments we analyzed, there was no
meaningful association between the proportion of residents
experiencing poverty and PRF per person
We did not find a meaningful association between PRF per person and the proportion
of people experiencing poverty in rural or nonrural communities across the country.
The correlation coefficients between poverty rates and PRF per person are very small
for both rural and nonrural census tracts (nonrural census tracts: correlation = -0.030,
p < .0001; rural census tracts: correlation = 0.040, p < .0001).
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 9
OEI-05-20-00580
Findings Related to Additional Race Groups
For Non-Hispanic White residents, we found results that were opposite the
results for Non-Hispanic Black residents. Specifically, census tracts with greater
concentrations of Non-Hispanic White residents were associated with less PRF
per person than census tracts with smaller concentrations of Non-Hispanic
White residents in both nonrural and rural census tracts, though the correlation
was stronger in nonrural census tracts. These results logically make sense as
proportion of Non-Hispanic White and Non-Hispanic Black populations in
census tracts likely have an inverse relationship.
We found correlations between PRF per person and concentrations of
additional race groups. We did not focus our major takeaways on these groups
because they comprise relatively small proportions of the population in our
analysis. For example, we found a negative correlation between the proportion
of Non-Hispanic residents that identify as “Other Race” in a census tract and the
PRF per person, and this correlation was stronger in nonrural census tracts.
Additionally, in our analysis of the Tribal Hospitals allotment specifically, we
found a positive correlation between the proportion of Non-Hispanic American
Indian/Alaska Native (AI/AN) residents in a census tract and the PRF per person,
indicating that greater concentrations of Non-Hispanic AI/AN residents were
associated with higher PRF per person from that allotment.
See Appendix C for additional details.
In the PRF allotment targeted to rural hospitals, communities
with more Hispanic/Latino and Non-Hispanic Black Residents
had less PRF per person
In addition to analyzing Targeted PRF amounts across all four allotments, we analyzed
each allotment individually. Targeted PRF was allocated with different funding
formulas in the different allotments, so understanding the results of specific
allotments can help inform future funding decisions.
We found the most notable trends in the Rural Hospitals allotment. We found two
similar trends in this allotment: Communities with greater concentrations of
Hispanic/Latino residents were associated with less PRF per person than communities
with smaller concentrations of Hispanic/Latino residents, and communities with
greater concentrations of Non-Hispanic Black residents were associated with less PRF
per person than communities with smaller concentrations of Non-Hispanic Black
residents.
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 10
OEI-05-20-00580
In the sections below, we only present the results related to rural census tracts, and
not nonrural census tracts, because there was only a small percentage of the Rural
Hospitals allotment that was allocated to rural hospitals that served nonrural
communities. However, we did see similar trends in nonrural census tracts as well.
See Appendix C for detailed analysis.
The Rural Hospitals Allotment and Hispanic/Latino Residents
When looking within each individual Targeted PRF allotment, the overall trend of less
PRF per person to census tracts with greater concentrations of Hispanic/Latino
residents (as described in the first result) was also evident in the Rural Hospitals
allotment.
Exhibit 5: Among rural census
tracts, we found that
communities with more
Hispanic/Latino residents had
less PRF per person in the Rural
Hospitals allotment.
The overall pattern we identified in
the analysis across all four Targeted
PRF allotments (i.e., less PRF per
person allocated to hospitals that
served communities with greater
concentrations of Hispanic/Latino
residents) appears to be driven
primarily by the differences within Notes: The bar graph illustrates the correlations described in the
the Rural Hospitals Allotment, main text of the Results. “Pop” represents the total population
that lives in each group of census tracts.
specifically. In the Rural Hospitals
allotment, we identified a negative Source: OIG analysis of PRF allocations, HHS Protect Data, ACS
Data, Medicare hospitals claims and enrollment data, USA Facts
correlation between the
COVID-19 data, 2021
concentration of Hispanic/Latino
residents in a rural census tract and the PRF per person (rural correlation = -0.149, p <
.0001). Among the three other individual Targeted PRF allotments (High Impact
Hospitals, Safety Net Hospitals, Tribal Hospitals), the correlations between the
concentration of Hispanic/Latino residents and PRF per person were smaller or close
to zero. See Appendix C for specific correlation coefficients for each allotment.
The Rural Hospitals Allotment and Non-Hispanic Black Residents
In the Rural Hospitals allotment, rural communities with greater concentrations of
Non-Hispanic Black residents were associated with less PRF per person than rural
communities with smaller concentrations of Non-Hispanic Black residents. For the
Rural Hospitals allotments, we found a negative correlation between PRF per person
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 11
OEI-05-20-00580
and concentration of Non-Hispanic Black population in rural census tracts (correlation
= -0.117, p < .0001). e
Exhibit 6: Among rural census
tracts, we found that communities
with more Non-Hispanic Black
residents had less PRF per person
in the Rural Hospitals allotment,
but this trend was offset by the
opposite pattern in two other
allotments.
Notes: The bar graphs illustrate the correlations described in the main text of the Results. “Pop” represents the total
population that lives in each group of census tracts.
Source: OIG analysis of PRF allocations, HHS Protect Data, ACS Data, Medicare hospitals claims and enrollment data,
USA Facts COVID-19 data, 2021
We found that this trend related to Non-Hispanic Black residents was offset by the
opposite pattern in other individual allotments (see Exhibit 6), which may explain why,
as noted previously, we did not find a meaningful association between PRF per person
and concentration of Non-Hispanic Black residents in rural communities when looking
across all four Targeted PRF allotments. When we analyzed the High Impact Hospitals
_____________________________________________________________________________________________________________________________
e The correlation in the Tribal Hospitals allotment was also negative (correlation = -0.067, p < .0001).
We
did not include this allotment in Exhibit 6 because the dollar amounts per person across all rural census
tracts are too small to be visible on the same scale. The total money targeted to Tribal Hospitals was
much smaller than the other allotments.
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 12
OEI-05-20-00580
allotment and the Safety Net Hospitals allotment individually, we found a positive
correlation between PRF per person and concentration of Non-Hispanic Black
population in rural census tracts (for rural census tracts, High Impact Hospitals
allotment correlation = 0.286, p < .0001; Safety Net Hospitals allotment correlation =
0.234, p < .0001).
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 13
OEI-05-20-00580
CONCLUSION
While the PRF was not specifically intended to address health disparities, health care
funding, in general, is an important tool that can help HHS contribute to goals of
reducing health disparities by supporting communities and populations with greater
risk of poor health outcomes, both in the context of COVID-19 and more broadly.
HHS has prioritized health equity and reducing health disparities, especially during
the COVID-19 pandemic. 16, 17 Many HHS programs, including those that target
funding and resources, are meant to reach underserved populations and address
disparities. Further, HHS’s emergency preparedness goals include responding to
public health emergencies in ways that meet the needs of at-risk populations, which,
in the case of COVID-19, may include people of color and people from communities
experiencing higher rates of poverty. 18, 19
Understanding the relationship between the PRF per person and community
characteristics such as race, ethnicity, and income can help to inform HHS on
decisions about allotting future emergency funding. For example, in the PRF
allotment targeted to rural hospitals, communities with more Hispanic/Latino and
Non-Hispanic Black Residents had less PRF per person. However, Hispanic/Latino
Americans have been more likely to be hospitalized or die from COVID-19 than non-
Hispanic White Americans. Thus, it may be especially important to consider HHS’s
goal to reduce health disparities when allocating emergency funding to rural
communities in the future.
We hope that this analysis is useful to HHS in planning for future emergency funding
scenarios and identifying opportunities to support these goals, to the extent
permitted by law. Given that emergency funding, like PRF, is designed to respond to
an unexpected and urgent need for health care funding and the Department will have
to make decisions quickly about how to allocate money, effective planning could help
ensure HHS emergency funding is allocated in ways that meet the needs of at-risk
populations and further reduce health disparities. Because statutory parameters can
impact HHS’s ability to direct funding, it may also be important for HHS to engage
with Congress regarding the authority needed to most effectively consider health
equity, alongside other priorities like efficiency and expeditiousness, in emergency
funding scenarios.
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 14
OEI-05-20-00580
METHODOLOGY
Data Sources
PRF allocations
HRSA provided OIG with data on all Targeted PRF allocations by provider CMS
Certification Number (CCN) that were allocated during calendar year 2020.
Hospital data
We used data related to individual hospitals from HHS Protect. HHS Protect is a
COVID-19 data hub managed by HHS. 20 We used this data source to identify
hospitals and their locations.
Data from HHS Protect rely on reports submitted by hospitals. HHS Protect was
rolled out for use on April 10, 2020. 21
Medicare hospital claims and enrollment data
We used Medicare Part A and B claims data and Part C encounter data from calendar
years 2020 and 2021 to identify the service area of each hospital. We used hospital
inpatient and outpatient claims of discharges to identify the Medicare patients who
were served by each hospital, and then used Medicare enrollment files to identify the
census tracts where those patients live.
Census tract characteristics
We used Census American Community Survey (ACS) 2019 5-year estimates for census
tract social factors: percent of each self-identified race and ethnicity and percent of
residents with income less than the poverty level (poverty rate) Census. 22
The ACS data are self-reported, based on a sample, and are subject to sampling
variability. A relatively small sample of each census tract’s population responded to
the ACS survey, and respondents were instructed to choose which category best fits
with how they self-identify.
In the ACS survey, respondents are asked to self-identify their race. In 2019,
respondents were presented the following options and told to “check all that apply”:
White, Black or African Am., American Indian or Alaska Native, Asian Indian, Chinese,
Filipino, Japanese, Korean, Vietnamese, Native Hawaiian, Guamanian or Chamorro,
Samoan, Other Asian, Other Pacific Islander, Some other race. 23 The ACS 5-year
estimates data combine multiple Asian races into one category and multiple Native
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 15
OEI-05-20-00580
Hawaiian/Pacific Islander races into one category. f For this study, we further
combined Asian and Native Hawaiian/Pacific Islander to one category. We considered
respondents as “Other Race” by combining the count of respondents who selected
“Some other race” with the count of respondents who checked more than one race.
In the ACS survey, respondents were asked about their ethnicity separate from race.
Respondents were presented five options: 1) not Hispanic, Latino, or Spanish origin, 2)
Yes, Mexican, Mexican Am., Chicano, 3) Yes, Puerto Rican, 4) Yes, Cuban, 5) Yes,
another Hispanic Latino, Spanish origin (with a text box below to Print origin). Option
1 categorizes individuals into not Hispanic, Latino, or Spanish origin while options 2-5
categorize individuals into Hispanic, Latino, or Spanish origin. Because ethnicity is
separate from race, respondents of Hispanic, Latino, or Spanish origin could select any
of the races presented.
To simplify our analysis, we combined ACS respondent data about races and
ethnicities to create the following groups: Hispanic (of any race); non-Hispanic White;
non-Hispanic Black; non-Hispanic Asian or Native Hawaiian/Pacific Islander (Asian);
non-Hispanic American Indian/Alaska Native (AIAN); non-Hispanic Other Race (other
than White, Black, Asian, AIAN); or more than one Race (other Race).
To define rural and nonrural census tracts, we used HRSA’s list of rural tracts. HRSA
identifies rural vs. nonrural census tracts using Rural-Urban Commuting Area (RUCA)
codes created by the U.S. Department of Agriculture’s Economic Research Service. 24
To determine the distance from each census tract to the nearest hospital, which we
used as a control in our regression models, we used Medicare enrollment data for
enrollee addresses and the HHS Protect data above for hospital locations. For each
Medicare enrollee, we calculated the straight-line distance to each hospital and
determined the minimum distance. We then assigned the census-tract minimum
distance as the median of all the minimum distances for beneficiaries living in that
census tract.
COVID-19 deaths
We used county COVID-19 death count estimates from calendar year 2020 from
USAFacts.org. USAFacts is a nonpartisan non-profit organization that collects and
shares public data reported by state and local governments. 25 The USAFacts data
provide the cumulative cases and deaths by county from January 22, 2020, through
December 31, 2020. Because census-tract level data on COVID-19 deaths were not
available, we assigned data from each county to the individual census tracts located in
that county. Data from USAFacts rely on measurement and reporting from state- and
local-level public health agencies. 26 As such, data collection and reporting
methodologies may vary.
_____________________________________________________________________________________________________________________________
f We used ACS Table B03002, “Hispanic or Latino Origin By Race.”
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 16
OEI-05-20-00580
Data Analysis
We analyzed the relationship between PRF per person and the racial, ethnic, and
economic composition of each census tract. Specifically, we analyzed these variables:
concentration of Hispanic/Latino residents, concentration of each race group (e.g.,
Non-Hispanic Black), and concentration of residents experiencing poverty. For our
regression modeling, we also analyzed COVID-19 deaths per capita and distance to
the nearest hospital.
Calculating PRF allocations per person
We used Targeted PRF allocation data that we received directly from HRSA to
calculate the total amount of Targeted PRF for each hospital. We summed all
Targeted PRF allotments that were allocated in 2020 except for the allotment targeted
to children’s hospitals. We excluded this allotment because we used Medicare data to
assign PRF dollars to census tracts, as described below.
We analyzed allocation amounts to most hospitals that were allocated Targeted PRF,
including acute care hospitals, tribal hospitals, and critical access hospitals. We
excluded 20 hospitals that were allocated Targeted PRF but were not required to
submit COVID-19 data in HHS Protect, for example, rehabilitation and psychiatric
hospitals. 27
After calculating the Targeted PRF amounts across all four allotments for each
hospital, we calculated “PRF per person” for each census tract using data about
hospital service areas. We used Medicare claims and enrollment data to identify the
census tracts of the Medicare patients who had Medicare inpatient or outpatient
claims from each hospital during calendar year 2020 or 2021. For each hospital, we
then assigned PRF dollars to each census tract in its service area, proportional to the
number of claims for patients who live in each census tract. For each census tract, we
summed all the money assigned from all hospitals that served that census tract.
Lastly, we divided Targeted PRF amounts across all four allotments for each census
tract by the population of that census tract. See Exhibit 7 (same as Exhibit 2 from
Background) for a diagram of the hospital service area methodology using a very
simple example to illustrate. We repeated this same methodology for each of the
individual allotments of funding (High Impact Hospitals, Safety Net Hospitals, Rural
Hospitals, and Tribal Hospitals), using the same hospital service areas.
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 17
OEI-05-20-00580
Exhibit 7: Model to illustrate the hospital service area methodology we used
to calculate each census tract’s PRF per person
Excluded census tracts
We calculated PRF per person for all census tracts in which Medicare beneficiaries
served by hospitals live. However, we then excluded a small number of census tracts
from the analyses. We limited our analysis to tracts in 50 States and DC, excluding
tracts in U.S. territories. We also excluded census tracts with missing population
characteristics, especially low population estimates, and data anomalies. Specifically,
we excluded tracts with fewer than 100 residents or fewer than 10 Medicare enrolled
residents per ACS estimates, tracts with greater than 50% of the residents living in
institutional settings per ACS, and tracts that were outliers when we compared
Medicare enrollment counts and the ACS estimates of the number of people with
Medicare coverage. We excluded 5% of tracts for final analysis of 70,342 tracts.
Statistical analysis
For all allotments combined, and separately for the four allotments, we performed the
following statistical analysis to understand the relationships between PRF per person
and census tract community characteristics.
Correlation analysis. To examine the relationships between PRF per person and
census tract population characteristics, we calculated Pearson correlation coefficients.
We did this separately for rural and nonrural census tracts. We used log transformed
variables in the correlation analysis to reduce the influence of outliers.
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 18
OEI-05-20-00580
We reported a correlation between variables as those with a Pearson correlation
coefficient greater than 0.05 or less than -0.05 and a p-value less than .05. We
reported “no meaningful association” between variables when Pearson correlation
coefficients were between -0.05 and 0.05, even if those correlations were statistically
significant.
Regression analysis. To understand relationships between multiple variables, we ran
linear regressions controlling for COVID-19 deaths in January 22-June 10, 2020, and
June 11-December 31, 2020, distance to the nearest hospital, and the census tract
social factors (proportion of each race and ethnicity group and proportion of
population experiencing poverty) to understand which factors were associated with
PRF per person while controlling for the other factors in the model. Specifically, we
used the following independent variables: proportion of Hispanic/Latino, Non-
Hispanic Black, Non-Hispanic Asian/PI, and Non-Hispanic AI/AN residents, residents
who identify as Non-Hispanic Other Race, proportion of people experiencing poverty,
2020 COVID-19 deaths per capita, and distance to the nearest hospital. We used the
same set of independent variables across all regression models so that it would be
more possible to compare the estimated parameters among models. We log
transformed both the dependent and independent variables in all regression models
to reduce the influence of outliers.
Graphics. To highlight the notable correlations that we discuss in the main report, we
summarized the correlation analysis by creating bar graphs. We subdivided the
census tracts into groups based on the population characteristics (percent
Latino/Hispanic and percent non-Hispanic Black). For each of these groups, we
calculated the total PRF divided by the total population and then compared the PRF
per person across the groups. We created the groups by using interval cutoffs. For
example, we compared PRF per person for tracts with 0-<5% Hispanic/Latino, 5%-
<10% Hispanic/Latino, 10%-<15% Hispanic/Latino, and 15% or higher
Hispanic/Latino. The group labeled 0-<5% Hispanic/Latino was created by including
census tracts with equal to or greater than 0% of Hispanic/Latino residents and less
than 5% of Hispanic/Latino residents. 5%-<10% Hispanic/Latino was created by
including census tracts with equal or to greater than 5% Hispanic/Latino residents and
less than 10% Hispanic/Latino residents. We continued to use this method for the last
two interval cutoff groups.
Scope
For this evaluation, we analyzed the Provider Relief Funds allocated in Targeted
Distributions to hospitals during calendar year 2020. We did not examine General
Distributions because allocations could not be always linked to individual hospitals.
We excluded the Targeted Distribution targeted to children’s hospitals because the
hospital service area distribution limited to people covered by Medicare may not be
representative of children’s hospital service areas.
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 19
OEI-05-20-00580
Our analysis was limited to Targeted PRF allocated to hospitals in calendar year 2020.
This means our analysis does not provide information regarding any PRF allocated to
non-hospital providers, or about any PRF that were distributed after the end of 2020.
Our analysis should not be read to imply that differences in census tract
characteristics caused differences in PRF per person. Additionally, we did not assess
compliance with Federal requirements.
Limitations
We could not assess the patterns in allocations from the General Distribution with
respect to county population characteristics because data for individual hospital
allocations do not always exist. General Distribution payments were not always
allocated at the facility level because, in the case of hospitals in hospitals systems,
they were instead allocated directly to parent companies. Further, we were only able
to analyze Targeted Distribution amounts HRSA allocated to hospitals initially, not
where the funds were ultimately distributed, because parent organizations, which can
include multiple hospitals located in different areas, were given flexibility to re-
allocate and distribute funding to facilities under their umbrella at their discretion.
Therefore, we could not assume parent organizations disbursed the money in any
certain way. It is unknown the extent to which General Distribution allocations may
have changed the outcomes we identified, particularly for hospitals that did not
receive any funding through the Targeted Distribution but did receive funding
through later phases of the General Distribution.
However, researchers have raised concerns about the equity of the earlier phases of
General Distributions. A study published in the Journal of the American Medical
Association suggested that the formulas used to allocate money may have favored
hospitals that primarily serve White and higher-income populations. 28 Specifically,
initial distributions were allocated proportionally to hospitals’ past revenues.
Hospitals that primarily serve White populations tend to have higher revenues, on
average, than hospitals that primarily serve people of color and whose patients are
more likely to be covered by Medicaid or underinsured. 29, 30 However, because
General Distribution payments were often calculated and reported at the parent
organization level, there is not comprehensive data on how much individual facilities
and providers qualified for.
We used Medicare data in our calculation of PRF per person, rather than information
on all hospital patients, because we do not have access to comprehensive nationwide
utilization data from other payers that could be matched with residence census tracts.
Medicare claims data might not always be representative of the full population that
receives care at any given hospital. Although we do have hospital utilization from
Medicaid, we could not effectively match Medicaid enrollees to their residence census
tracts so we could not use this data.
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 20
OEI-05-20-00580
In this study, we used demographic data about characteristics of people living in
census tracts. Geography is a proxy that can help us understand the potential impact
on certain individuals, but it is not a substitute for lived experiences of individuals.
Additionally, we used county-level data for 2020 COVID-19 deaths because 2020
COVID-19 deaths data were not available at the census tract level.
Standards
We conducted this study in accordance with the Quality Standards for Inspection and
Evaluation issued by the Council of the Inspectors General on Integrity and Efficiency.
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 21
OEI-05-20-00580
APPENDIX A
Detailed information about Provider Relief Fund and related
relief funding
Provider Relief Fund
The PRF refers to relief funds appropriated by Congress to the Public Health and
Social Services Emergency Fund based on the Coronavirus Aid, Relief, and Economic
Security (CARES) Act in March 2020, the Paycheck Protection Program and Health
Care Enhancement Act (PPHCEA) in April 2020, and the Coronavirus Response and
Relief Supplemental Appropriations (CRRSA) Act in December 2020. The CARES Act
provided $100 billion, the PPHCEA added $75 billion, and the CRRSA Act added an
extra $3 billion. 31, 32, 33
Through these Acts, Congress delegated PRF payment calculation and disbursement
to HHS and required HHS to make PRF payments using the most efficient payment
systems practicable. 34 Congress appropriated these funds “to prevent, prepare for,
and respond to coronavirus, domestically or internationally, for necessary expenses to
reimburse health care providers for health care related expenses or lost revenues that
are attributable to coronavirus.” 35 Additionally, some of the PRF appropriations were
used for claims reimbursement programs and COVID-19 Coverage Assistance Fund. 36,
37
HHS worked with The Office of the Assistant Secretary for Planning and Evaluation
and Health Resources and Services Administration (HRSA), along with multiple
stakeholders including third-party contractors, to administer the PRF. In April 2020,
HHS designated HRSA as the primary operating division in charge of coordinating
and overseeing the creation of payment files, calculation of payment allocations for
eligible health care providers and facilities, and disbursement of funds for allotments
of the General Distributions and Targeted Distributions. g
General Distributions
General Distributions were intended to provide financial relief to providers, including
hospitals and other facilities, during the COVID-19 pandemic based on characteristics
such as participation in Medicare or Medicaid. 38 General Distributions consisted of
four phases of allotments. 39, 40, h Phase 1 consisted of a total of $41.9 billion to
_____________________________________________________________________________________________________________________________
g According to HRSA, third-party contractors developed the methodology and determined the calculated
payments files for hospitals and providers. Then, HRSA and the third-party contractors identified
ineligible hospitals and providers. Lastly, the United Health Group allocated final calculated distributions
to the eligible providers.
h Dollar amounts for General Distribution Phases 1-3 allotments are current as of January 2023.
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 22
OEI-05-20-00580
providers who billed Medicare fee-for-service in calendar year 2019. 41 Phase 2
consisted of $4.5 billion to providers that participated in state Medicaid/Children’s
Health Insurance Plan or Medicaid managed care plans, dental providers, assisted
living facilities, and certain Medicare providers. 42 Phase 3 consisted of $19.5 billion to
providers previously eligible for earlier phases, while also expanding to additional
provider types such as behavioral health providers. 43 Phase 3 payment calculations
included a deduction of all prior PRF payments for providers that received any PRF
funds. 44 Phase 4 has consisted of $15.4 billion to providers previously eligible for
earlier phases. Phase 4 payments consisted of 1) base payments based on providers’
reported changes in revenues and expenses from July 1, 2020, to March 31, 2021; and
2) bonus payments based on providers’ level of participation in Medicaid, the
Children’s Health Insurance Program, and Medicare. 45, i
According to HRSA, General Distribution payments were sometimes allocated and/or
disbursed to parent organizations, rather than individual facilities or providers. At the
beginning of the pandemic, health care facilities experienced immediate financial
hardship. 46 HHS made payments quickly to provide financial support to ensure
providers such as hospitals could sustain services. 47 The CARES Act required HHS to
use the most efficient payment systems practicable to provide PRF payments. 48
According to HHS guidance, any parent organization with more than one facility was
allowed to allocate and distribute funding to the facilities under its umbrella at its
discretion. 49
Targeted Distributions
Targeted Distributions were intended to provide financial relief to address added
COVID-19 challenges, specific types of hospitals, and other facilities based on certain
metrics and characteristics (e.g., high numbers of COVID-19 admissions, rural or tribal
designations). Targeted Distributions consisted of several allotments that each had
different formulas and methodologies to determine PRF payment amounts. 50
Targeted Distributions consisted of several allotments, some of which included
payments to hospitals. j, 51 The High Impact allotment consisted of $20.7 billion to
hospitals that had a high number of confirmed COVID-19 positive inpatient
admissions. 52 The Rural allotment consisted of approximately $11 billion to rural
hospitals including rural acute care general hospitals and Critical Access Hospitals
(CAHs), primarily based on facility operating expenses. 53 The IHS and Tribal Hospitals
allotment consisted of $494 million to IHS and tribal facilities based on facility or
program operating expenses and estimated service population. 54 The Safety Net
Hospitals allotment consisted of $13.9 billion to Safety Net Hospitals, acute care
hospitals, and children’s hospitals based on profit margin, proportion of Medicare
_____________________________________________________________________________________________________________________________
i Dollar amount for General Distribution Phase 4 allotment is current as of January 2023.
j Not all $178 billion of the PRF has been allocated to hospitals and providers.
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 23
OEI-05-20-00580
patients, and proportion of Medicaid patients for the children’s hospitals. 55 Exhibit 8
describes the metrics used for each PRF Targeted Distribution allotment.
Exhibit 8: 2020 PRF Targeted Distribution allotments that included hospitals
Targeted PRF
Allotments Total Amounts Metric Used to Target Allotments
COVID-19 High $20.7 billion Number of COVID-19 positive inpatient admissions
Impact Area
(High Impact
Hospitals)
Safety Net $12.8 billion Proportion of the individual facility score (number of facility
Hospitals: Acute beds multiplied by Medicare Disproportionate Patient
Care Facilities Percentage for an acute care facility) to the cumulative
facility scores
Rural Hospitals $10.7 billion Base payments plus operating expenses (including lost
and Clinics revenue) of rural acute care hospitals and CAHs, operating
expenses of Sole Community Hospitals, Medicare
Dependent Hospitals, & Rural Referral Center Hospitals in
Small Metro Areas
Safety Net $1.1 billion Proportion of the individual facility score (facility beds
Hospitals: multiplied by Medicaid only ratio) to the cumulative facility
Children’s scores
Hospitals
Indian Health $0.49 billion Base payments plus operating expenses of IHS and Tribal
Services and Hospitals and estimated service population
Tribal Hospitals
Notes: These amounts represent the total number of dollars distributed in the listed Targeted PRF allotments. They do
not represent the subset of Targeted PRF allotments that were in OIG’s analysis.
Source: HRSA, Provider Relief Fund Past Payments Targeted Distributions website. 56 Dollar amounts are current as of
November 2022.
According to HRSA, Targeted Distribution allotments were allocated for each
individual facility and were either distributed to individual facilities or rolled up and
distributed to parent organizations.
Related COVID-19 Relief Funding
In addition to PRF, HHS, through HRSA, began distributing American Rescue Plan
payments to rural health care providers in November 2021. These funds were
intended to address the disproportionate impact of COVID-19 on rural health care
providers and rural communities. 57 These payments were based on characteristics
such as Medicare reimbursements for services to rural beneficiaries. 58
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 24
OEI-05-20-00580
Related OIG Work
The Office of Inspector General (OIG) is conducting strategic oversight of HHS’s
pandemic response and recovery efforts, including additional reviews related to the
PRF. For example, OIG found that HHS and HRSA developed controls related to
selected PRF program requirements to ensure that providers received the correct PRF
payments from the Phase 1 General Distribution in a fast, fair, and transparent manner
but that some of those controls could be improved. 59
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 25
OEI-05-20-00580
APPENDIX B
Rural and Nonrural Census Tracts
Hospitals serving rural census tracts were allocated more PRF per person than
hospitals serving nonrural census tracts. The PRF per person in rural census tracts was
$218 per person, which is 81% higher than the PRF per person in nonrural census
tracts.
The COVID-19 pandemic has impacted urban and rural populations differently. The
disease initially spread most rapidly in large metropolitan areas (nonrural), which
aligns with the fact that most of the U.S. population lives in dense, urban regions. By
late October 2020, the number of cumulative COVID-19 cases in rural areas exceeded
those in urban areas. 60
In addition, the racial and ethnic makeup of rural and nonrural census tracts differs
considerably. Exhibit 9 summarizes many of the differences in specific racial and
ethnic group populations between rural and nonrural census tracts.
Because of these differences, we analyzed rural and nonrural census tracts separately
based on HRSA’s designations.
Exhibit 9: Statistics on nonrural and rural census tracts
Mean in Nonrural Mean in Rural
Census Tracts Census Tracts
(n = 55528 tracts) (n = 14814 tracts)
2020 COVID-19 106.87 108.92
Deaths/100,000
residents
% Poverty 13.96% 15.99%
% Non-Hispanic Black 14.86% 7.80%
% Non-Hispanic Asian/PI 6.21% 0.95%
% Non-Hispanic AI/AN 0.39% 1.71%
% Non-Hispanic Other 2.81% 2.08%
Race
% Non-Hispanic White 57.05% 78.98%
% Hispanic/Latino 18.68% 8.48%
Note: 2020 COVID-19 deaths represent the total deaths during the range of dates included in our study, January 22,
2020, to December 31, 2020.
Source: OIG Analysis of USAFacts COVID-19 data, ACS data, HRSA Rural Census Tract Designations.
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 26
OEI-05-20-00580
APPENDIX C
Detailed Statistical Analysis for All PRF and Individual Waves
We found that the median PRF per person across the four allocations was $106, and
PRF per person ranged between $49 per person and $203 per person.
Correlation Analysis Results
In the main report, we reported on the correlation coefficients between PRF per
person and proportion of Hispanic/Latino ethnicity, Non-Hispanic Black, and
proportion of people experiencing poverty for all Targeted PRF allotments. We also
performed the same analysis with regards to the following census tract characteristics:
proportion of Non-Hispanic Asian/PI residents, Non-Hispanic AI/AN residents,
residents who identify as Non-Hispanic Other Race, 2020 COVID-19 deaths per capita,
and distance to the nearest hospital. Exhibit 10 shows the correlation coefficients for
each of these characteristics in all Targeted PRF allotments. We calculated two sets of
correlation coefficients for all Targeted PRF allotments: 1) using only nonrural census
tracts and 2) using only rural census tracts.
The purpose of this report was to analyze correlations between PRF per person and
community demographic characteristics, including race, ethnicity, and economic
status. Because of this, we did not focus on the correlations between PRF per person
and 2020 COVID-19 deaths or PRF per person and distance to the nearest hospital,
even though these correlations were often stronger than that of the variables we
discussed in our main findings.
Please note that most of the census tracts have small percentages of residents that
identify as Non-Hispanic “Other Race” (a race not listed or more than one race), Non-
Hispanic Asian/PI, and Non-Hispanic AI/AN (please see Appendix B for further
details).
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 27
OEI-05-20-00580
Exhibit 10: Results for correlation analysis for all Targeted PRF allotments in
nonrural tracts and rural tracts
Note: Census tract characteristics were log transformed in all regression models.
All Targeted PRF Allotments
Pearson Correlation Coefficients (p-values)
Nonrural Tracts (n = 55528 tracts)
Census
Tract 2020 2020 % Non-
Charact- COVID COVID % Non- % Non- % Non- Hispanic % Non- Distance
eristics Deaths Deaths Hispanic Hispanic Hispanic Other % Hispanic % from
(Jan–Jun) (Jun–Dec) AIAN Asian Black Race Hispanic White Poverty hospital
PRF per 0.637 0.015 -0.097 -0.034 0.136 -0.142 -0.088 -0.192 -0.030 -0.125
Person
p-value <.0001 0.0006 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001
Rural Tracts (n = 14814 tracts)
Census 2020 2020 % Non-
Tract COVID COVID % Non- % Non- % Non- Hispanic % Non- Distance
Charact- Deaths Deaths Hispanic Hispanic Hispanic Other % Hispanic % from
eristics (Jan–Jun) (Jun–Dec) AIAN Asian Black Race Hispanic White Poverty hospital
PRF per 0.021 0.082 0.072 -0.101 0.021 -0.072 -0.154 -0.060 0.040 -0.251
Person
p-value 0.0113 <.0001 <.0001 <.0001 0.0096 <.0001 <.0001 <.0001 <.0001 <.0001
We also calculated correlation coefficients between PRF per person and community
characteristics for each individual Targeted PRF allotment. Exhibits 11-14 show those
correlation coefficients for High Impact Hospitals, Safety Net Hospitals, Rural
Hospitals, and Tribal Hospitals allotments. We calculated correlation coefficients
separately for nonrural and rural census tracts.
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 28
OEI-05-20-00580
Exhibit 11: Results for correlation analysis for High Impact Hospitals
allotments in nonrural tracts and rural tracts
Note: Census tract characteristics were log transformed in all regression models.
High Impact Hospitals Allotments
Pearson Correlation Coefficients (p-values)
Nonrural Tracts (n = 55528 tracts)
Census
Tract 2020 2020 % Non-
Charact- COVID COVID % Non- % Non- % Non- Hispanic % Non- Distance
eristics Deaths Deaths Hispani Hispanic Hispanic Other % Hispanic from
(Jan–Jun) (Jun–Dec) c AIAN Asian Black Race Hispanic White % Poverty hospital
PRF per 0.795 -0.024 -0.159 0.060 0.156 -0.142 -0.009 -0.186 -0.104 -0.139
Person
p-values <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 0.0343 <.0001 <.0001 <.0001
Rural Tracts (n = 14814 tracts)
Census 2020 2020 % Non-
Tract COVID COVID % Non- % Non- % Non- Hispanic % Non- Distance
Charact- Deaths Deaths Hispani Hispanic Hispanic Other % Hispanic from
eristics (Jan–Jun) (Jun–Dec) c AIAN Asian Black Race Hispanic White % Poverty hospital
PRF per 0.514 0.096 -0.102 -0.029 0.286 -0.101 -0.029 -0.164 -0.046 0.007
Person
p-values <.0001 <.0001 <.0001 0.0004 <.0001 <.0001 0.0004 <.0001 <.0001 0.3739
Exhibit 12: Results for correlation analysis for Safety Net Hospitals allotments
in nonrural tracts and rural tracts
Note: Census tract characteristics were log transformed in all regression models.
Safety Net Hospitals Allotments
Pearson Correlation Coefficients (p-values)
Nonrural Tracts (n = 55528 tracts)
Census
Tract 2020 2020 % Non-
Charact- COVID COVID % Non- % Non- % Non- Hispanic % Non- Distance
eristics Deaths Deaths Hispanic Hispanic Hispanic Other % Hispanic % from
(Jan – Jun) (Jun – Dec) AIAN Asian Black Race Hispanic White Poverty hospital
PRF per 0.264 0.044 -0.029 -0.018 0.153 -0.068 0.017 -0.224 0.093 -0.133
Person
p-values <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001
Rural Tracts (n = 14814 tracts)
Census 2020 2020 % Non-
Tract COVID COVID % Non- % Non- % Non- Hispanic % Non- Distance
Characteri Deaths Deaths Hispanic Hispanic Hispanic Other % Hispanic % from
stics (Jan – Jun) (Jun - Dec AIAN Asian Black Race Hispanic White Poverty hospital)
PRF per 0.118 0.057 -0.032 -0.078 0.234 -0.020 -0.066 -0.124 0.191 0.003
Person
p-values <.0001 <.0001 <.0001 <.0001 <.0001 0.015 <.0001 <.0001 <.0001 0.7052
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 29
OEI-05-20-00580
Exhibit 13: Results for correlation analysis for Rural Hospitals allotments in
nonrural tracts and rural tracts
Note: Census tract characteristics were log transformed in all regression models.
Rural Hospitals Allotments
Pearson Correlation Coefficients (p-values)
Nonrural Tracts (n = 55528 tracts)
Census
tract 2020 2020 % Non-
Charact- COVID COVID % Non- % Non- % Non- Hispanic % Non- Distance
eristics Deaths Deaths Hispanic Hispanic Hispanic Other % Hispanic % from
(Jan–Jun) (Jun – Dec) AIAN Asian Black Race Hispanic White Poverty hospital
PRF per -0.348 0.137 0.140 -0.314 -0.198 -0.056 -0.319 0.271 -0.032 0.252
Person
p-values <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001
Rural Tracts (n = 14814 tracts)
Census 2020 2020 % Non-
Tract COVID COVID % Non- % Non- % Non- Hispanic % Non- Distance
Character Deaths Deaths Hispanic Hispanic Hispanic Other % Hispanic % from
istics (Jan – Jun) (Jun – Dec) AIAN Asian Black Race Hispanic White Poverty hospital
PRF per -0.167 0.074 0.015 -0.097 -0.117 -0.065 -0.149 0.111 0.001 -0.310
Person
p-values <.0001 <.0001 0.0763 <.0001 <.0001 <.0001 <.0001 <.0001 0.8842 <.0001
Exhibit 14: Results for correlation analysis for Tribal Hospitals allotments in
nonrural tracts and rural tracts
Note: Census tract characteristics were log transformed in all regression models.
Tribal Hospitals Allotments
Pearson Correlation Coefficients (p-values)
Nonrural Tracts (n = 55528 tracts)
Census
Tract 2020 2020 % Non-
Charact- COVID COVID % Non- % Non- % Non- Hispanic % Non- Distance
eristics Deaths Deaths Hispanic Hispanic Hispanic Other % Hispanic from
(Jan–Jun) (Jun–Dec) AIAN Asian Black Race Hispanic White % Poverty hospital
PRF per -0.092 0.062 0.448 -0.035 -0.055 0.079 0.057 0.037 0.037 0.001
Person
p-values <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 0.8141
Rural Tracts (n = 14814 tracts)
Census 2020 2020 % Non-
Tract COVID COVID % Non- % Non- % Non- Hispanic % Non- Distance
Charact- Deaths Deaths Hispanic Hispanic Hispanic Other % Hispanic from
eristics (Jan–Jun) (Jun–Dec) AIAN Asian Black Race Hispanic White % Poverty hospital
PRF per -0.008 -0.008 0.641 0.015 -0.067 0.262 0.044 -0.283 0.104 0.045
Person
p-values 0.3218 0.3462 <.0001 0.0673 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 30
OEI-05-20-00580
Regression Analysis Results
We used multivariable linear regressions to determine if the relationships we
identified in the correlation analyses remain when all the census tract characteristics
are included in a single model. Our regressions control for COVID-19 deaths in
January 22-June 10, 2020, and June 11-December 31, 2020, distance to the nearest
hospital, and the census tract social factors (proportion of each race/ethnicity group,
proportion of population experiencing poverty) to understand which factors were
related to PRF per person while controlling for the other factors in the model. We log
transformed both the dependent and independent variables in all regressions to
reduce the influence of outliers. We report regression results for the five different
outcomes: all Targeted PRF allotments per person and each of the four individual
allotments per person (High Impact Hospitals, Safety Net Hospitals, Rural Hospitals,
and Tribal Hospitals). In addition to examining all tracts in a single model, we
modeled the relationships by HRSA designation.
It should be noted that the variation in PRF per person is more fully explained by the
included community characteristics for the High Impact Hospitals, Tribal and Rural
Hospitals allotments and less well explained in the models for Safety Net Hospitals
allotments as indicated by the Adjusted R-Squared values. We also tested alternative
models for Rural, Tribal and Safety Net Hospitals that received PRF in those
allocations. This substantially improved the Adjusted R-Squared for the Safety Net
Hospitals models, but the resulting set of independent variables would have been
different from the models for High Impact Hospitals. Results in Exhibits 15-19 should
be interpreted with this limitation of the models in mind.
Exhibit 15: Linear regression model results for all Targeted PRF allotments
Note: Census tract characteristics and PRF per person were log transformed in all regression models.
All Targeted PRF Allotments
Linear Regression Model of Targeted PRF per Person
Nonrural Rural
(n =55528 tracts) (n = 14814 tracts)
Estimate Std. Error p-value Estimate Std. Error p-value
Intercept 3.488 0.040 <.0001 5.928 0.046 <.0001
COVID-19 Deaths Jan-Jun 0.550 0.003 <.0001 0.009 0.004 0.0345
COVID-19 Deaths Jun-Dec -0.061 0.008 <.0001 0.028 0.006 <.0001
% Non-Hispanic AI/AN 0.043 0.010 <.0001 0.140 0.008 <.0001
% Non-Hispanic Asian/PI -0.074 0.004 <.0001 -0.133 0.011 <.0001
% Non-Hispanic Black 0.045 0.003 <.0001 -0.012 0.005 0.0226
% Non-Hispanic Other Race -0.163 0.006 <.0001 -0.111 0.010 <.0001
% Hispanic/Latino -0.070 0.004 <.0001 -0.125 0.005 <.0001
% Poverty Rate -0.057 0.006 <.0001 0.010 0.012 0.3751
Distance -0.053 0.006 <.0001 -0.252 0.007 <.0001
Adjusted R-Squared 0.4273 0.1359
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 31
OEI-05-20-00580
Exhibit 16: Regression model results for High Impact Hospitals allotments
Note: Census tract characteristics and PRF per person were log transformed.
High Impact Hospitals Allotments
Linear Regression Model of Targeted PRF per Person
Nonrural Rural
(n = 55528 tracts) (n = 14814 tracts)
Estimate Std. Error p-value Estimate Std. Error p-value
Intercept 1.636 0.045 <.0001 2.418 0.073 <.0001
COVID-19 Deaths Jan-Jun 0.988 0.003 <.0001 0.420 0.007 <.0001
COVID-19 Deaths Jun-Dec -0.157 0.009 <.0001 0.018 0.010 0.0722
% Non-Hispanic AI/AN -0.077 0.011 <.0001 -0.017 0.013 0.2035
% Non-Hispanic Asian/PI -0.035 0.005 <.0001 -0.038 0.018 0.032
% Non-Hispanic Black 0.124 0.004 <.0001 0.211 0.008 <.0001
% Non-Hispanic Other Race -0.245 0.007 <.0001 -0.117 0.016 <.0001
% Hispanic/Latino 0.047 0.004 <.0001 -0.009 0.009 0.3203
% Poverty Rate -0.272 0.007 <.0001 -0.374 0.019 <.0001
Distance 0.015 0.007 0.0391 0.044 0.010 <.0001
Adjusted R-Squared 0.655 0.3073
Exhibit 17: Linear regression model results for Safety Net Hospitals
allotments
Note: Census tract characteristics and PRF per person were log transformed in all regression models.
Safety Net Hospitals Allotments
Linear Regression Model of Targeted PRF per Person
Nonrural Rural
(n = 55528 tracts) (n = 14814 tracts)
Estimate Std. Error p-value Estimate Std. Error p-value
Intercept 2.331 0.057 <.0001 2.150 0.085 <.0001
COVID-19 Deaths Jan-Jun 0.232 0.004 <.0001 0.052 0.008 <.0001
COVID-19 Deaths Jun-Dec 0.030 0.011 0.0051 -0.008 0.012 0.4744
% Non-Hispanic AI/AN 0.046 0.013 0.0005 -0.031 0.016 0.0451
% Non-Hispanic Asian/PI -0.016 0.006 0.0066 -0.117 0.021 <.0001
% Non-Hispanic Black 0.090 0.005 <.0001 0.180 0.010 <.0001
% Non-Hispanic Other Race -0.118 0.009 <.0001 -0.011 0.019 0.5697
% Hispanic/Latino -0.006 0.005 0.1873 -0.085 0.010 <.0001
% Poverty Rate 0.060 0.008 <.0001 0.326 0.022 <.0001
Distance -0.131 0.009 <.0001 0.042 0.012 0.0006
Adjusted R-Squared 0.0921 0.0789
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 32
OEI-05-20-00580
Exhibit 18: Linear regression model results for Rural Hospitals allotments
Note: Census tract characteristics and PRF per person were log transformed in all regression models.
Rural Hospitals Allotments
Linear Regression Model of Targeted PRF per Person
Nonrural Rural
(n = 55528 tracts) (n = 14814 tracts)
Estimate Std. Error p-value Estimate Std. Error p-value
Intercept 1.561 0.042 <.0001 5.918 0.059 <.0001
COVID-19 Deaths Jan-Jun -0.229 0.003 <.0001 -0.103 0.005 <.0001
COVID-19 Deaths Jun-Dec 0.220 0.008 <.0001 0.065 0.008 <.0001
% Non-Hispanic AI/AN 0.250 0.010 <.0001 0.069 0.011 <.0001
% Non-Hispanic Asian/PI -0.203 0.004 <.0001 -0.190 0.014 <.0001
% Non-Hispanic Black -0.160 0.003 <.0001 -0.110 0.007 <.0001
% Non-Hispanic Other Race 0.016 0.007 0.0135 -0.111 0.013 <.0001
% Hispanic/Latino -0.277 0.004 <.0001 -0.167 0.007 <.0001
% Poverty Rate 0.084 0.006 <.0001 0.042 0.015 0.006
Distance 0.117 0.007 <.0001 -0.421 0.009 <.0001
Adjusted R-Squared 0.3285 0.2021
Exhibit 19: Linear regression model results for Tribal Hospitals allotments
Note: Census tract characteristics and PRF per person were log transformed in all regression models.
Tribal Hospitals Allotments
Linear Regression Model of Targeted PRF per Person
Nonrural Rural
(n = 55528 tracts) (n = 14814 tracts)
Estimate Std. Error p-value Estimate Std. Error p-value
Intercept -0.123 0.010 <.0001 -0.392 0.035 <.0001
COVID-19 Deaths Jan-Jun -0.003 0.001 <.0001 0.026 0.003 <.0001
COVID-19 Deaths Jun-Dec 0.030 0.002 <.0001 0.019 0.005 <.0001
% Non-Hispanic AI/AN 0.262 0.002 <.0001 0.578 0.006 <.0001
% Non-Hispanic Asian/PI -0.008 0.001 <.0001 -0.031 0.008 0.0002
% Non-Hispanic Black -0.008 0.001 <.0001 -0.021 0.004 <.0001
% Non-Hispanic Other Race 0.019 0.002 <.0001 0.103 0.008 <.0001
% Hispanic/Latino 0.009 0.001 <.0001 -0.008 0.004 0.0704
% Poverty Rate -0.005 0.001 0.0003 0.047 0.009 <.0001
Distance -0.005 0.002 0.0016 0.009 0.005 0.0625
Adjusted R-Squared 0.2101 0.4209
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 33
OEI-05-20-00580
Stepwise Regression Analysis for All Targeted PRF Allotments
To better understand the relationship between proportion of Hispanic/Latino
residents in a census tract and PRF and proportion of Non-Hispanic Black residents
and PRF, we performed stepwise regression analysis on all Targeted PRF allotments.
For both nonrural and rural census tracts, we started with a model including variables
for race, ethnicity, and poverty, and then added controls to the model in steps to
observe how estimates in the model changed.
In Exhibit 20, you can see that there is a positive correlation between the proportion
of Non-Hispanic Black residents and PRF per person in nonrural census tracts when
only race/ethnicity and poverty variables are included in the model. However, the
estimate associated with proportion of Non-Hispanic Black residents decreases
dramatically when the COVID-19 death rate variables are introduced to the model.
This indicates that although higher proportions of Non-Hispanic Black residents are
associated with higher PRF per person in nonrural census tracts, this relationship can
be partially explained by the positive association between COVID-19 death rates and
PRF per person in All Targeted PRF Allotments combined.
In both Exhibits 20 and 21 (both nonrural and rural census tracts), you can see that
the estimate associated with proportion of Hispanic/Latino residents does not change
meaningfully when the COVID-19 death rate variables are introduced to the model.
This indicates that the difference in PRF per person between high-Hispanic/Latino and
low-Hispanic/Latino communities cannot be explained by low COVID-19 impact in
areas with greater concentrations of Hispanic/Latino residents.
Exhibit 20: Stepwise regression model results for all Targeted PRF allotments
in nonrural census tracts
Note: Census tract characteristics and PRF per person were log transformed in all regression models.
All Targeted PRF Allotments
Stepwise Regression Model of Targeted PRF per Person
Nonrural Tracts (n = 55528 tracts)
Step 1 (No COVID-19 Step 2 (Include COVID-19 Step 3 (Include COVID-19
Deaths and Distance) Deaths; No Distance) Deaths and Distance)
Estimate Std. p-value Estimate Std. p-value Estimate Std. p-value
Error Error Error
Intercept 4.817 0.020 <.0001 3.333 0.036 <.0001 3.488 0.040 <.0001
% Non-Hispanic AI/AN -0.158 0.012 <.0001 0.043 0.010 <.0001 0.043 0.010 <.0001
% Non-Hispanic Asian/PI 0.027 0.005 <.0001 -0.067 0.004 <.0001 -0.074 0.004 <.0001
% Non-Hispanic Black 0.171 0.004 <.0001 0.047 0.003 <.0001 0.045 0.003 <.0001
% Non-Hispanic Other Race -0.307 0.008 <.0001 -0.161 0.006 <.0001 -0.163 0.006 <.0001
% Hispanic/Latino -0.070 0.004 <.0001 -0.067 0.003 <.0001 -0.070 0.004 <.0001
% Poverty Rate -0.135 0.007 <.0001 -0.046 0.006 <.0001 -0.057 0.006 <.0001
COVID-19 Deaths Jan-Jun 0.555 0.003 <.0001 0.550 0.003 <.0001
COVID-19 Deaths Jun-Dec -0.061 0.008 <.0001 -0.061 0.008 <.0001
Distance -0.053 0.006 <.0001
Adjusted R-Squared 0.064 0.427 0.427
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 34
OEI-05-20-00580
Exhibit 21: Stepwise regression model results for all Targeted PRF
allotments in rural census tracts
Note: Census tract characteristics and PRF per person were log transformed in all regression models.
All Targeted PRF Allotments
Stepwise Regression Model of Targeted PRF per Person
Rural Tracts (n = 14814 tracts)
Step 1 (No COVID-19 Step 2 (Include COVID-19 Step 3 (Include COVID-19
Deaths and Distance) Deaths; No Distance) Deaths and Distance)
Estimate Std. p-value Estimate Std. p-value Estimate Std. p-value
Error Error Error
Intercept 5.330 0.033 <.0001 5.125 0.043 <.0001 5.928 0.046 <.0001
% Non-Hispanic AI/AN 0.112 0.009 <.0001 0.111 0.009 <.0001 0.140 0.008 <.0001
% Non-Hispanic Asian/PI -0.080 0.011 <.0001 -0.069 0.012 <.0001 -0.133 0.011 <.0001
% Non-Hispanic Black 0.012 0.005 0.0214 0.003 0.006 0.5816 -0.012 0.005 0.0226
% Non-Hispanic Other -0.099 0.011 <.0001 -0.088 0.011 <.0001 -0.111 0.010 <.0001
Race
% Hispanic/Latino -0.104 0.006 <.0001 -0.106 0.006 <.0001 -0.125 0.005 <.0001
% Poverty Rate 0.049 0.012 <.0001 0.046 0.012 0.0002 0.010 0.012 0.3751
COVID-19 Deaths Jan-Jun 0.008 0.004 0.0744 0.009 0.004 0.0345
COVID-19 Deaths Jun-Dec 0.047 0.007 <.0001 0.028 0.006 <.0001
Distance -0.252 0.007 <.0001
Adjusted R-Squared 0.042 0.048 0.133
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 35
OEI-05-20-00580
ACKNOWLEDGMENTS AND CONTACT
Acknowledgments
Samantha Handel-Meyer and Hilary Slover served as the team leaders for this study,
and Thaddeus Chatto served as the lead analyst. Others in the Office of Evaluation
and Inspections Regional Offices who conducted the study include Rebecca Gorges
and Brianna Weldon. Office of Evaluation and Inspections headquarters staff who
provided support include Althea Hosein, Robert Gibbons, Michael Novello, and Sarah
Swisher.
We would also like to acknowledge the contributions of other Office of Inspector
General staff, including Elina Breton, Bianca Hernandez, Justin Koppa, and Jason Lee.
This report was prepared under the direction of Laura Kordish, Regional Inspector
General for Evaluation and Inspections in the Chicago Regional Office, Adam
Freeman, Deputy Regional Inspector General, and Hilary Slover, Assistant Regional
Inspector General.
Contact
To obtain additional information concerning this report, contact the Office of Public
Affairs at Public.Affairs@oig.hhs.gov. OIG reports and other information can be found
on the OIG website at oig.hhs.gov.
Office of Inspector General
U.S. Department of Health and Human Services
330 Independence Avenue, SW
Washington, DC 20201
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 36
OEI-05-20-00580
ENDNOTES
1 Commonwealth Fund, Achieving Racial and Ethnic Equity in U.S. Health Care. Accessed at Achieving Racial and Ethnic Equity
in U.S. Health Care: Scorecard | Commonwealth Fund on May 15, 2023.
2 CDC, COVID-19 in Racial and Ethnic Minority Groups. Accessed at COVID-19 racial and ethnic health disparities : 1.
Introduction (cdc.gov) on November 6, 2020.
3 CNN, Black communities account for disproportionate number of COVID-19 deaths in the US, study finds.
Accessed at Black
communities account for disproportionate number of Covid-19 deaths in the US, study finds | CNN on November 6, 2020.
4 Washington Post, The coronavirus is infecting and killing black Americans at an alarmingly high rate.
Accessed at
https://www.washingtonpost.com/nation/2020/04/07/coronavirus-is-infecting-killing-black-americans-an-alarmingly-high-
rate-post-analysis-shows/ on November 6, 2020.
5 Brookings, Racial economic inequality amid the COVID-19 crisis.
Accessed at https://www.brookings.edu/research/racial-
economic-inequality-amid-the-covid-19-crisis/ on November 6, 2020.
6
Pew Research Center, For U.S. Latinos, COVID-19 Has Taken a Personal and Financial Toll. Accessed at For U.S. Latinos,
COVID-19 Has Taken a Personal and Financial Toll | Pew Research Center on August 27 2021.
7 COVID-19 has had a disproportionate impact among high-poverty populations: National Center for Biotechnology
Information, Coronavirus infections and deaths by poverty status: The effects of social distancing. Accessed at
https://pubmed.ncbi.nlm.nih.gov/33362321/ on August 27, 2021.
8 HHS Office of Minority Health, Minority Health Social Vulnerability Index Overview. Accessed at Minority Health Social
Vulnerability Index Overview (hhs.gov) on October 26, 2022.
9 CDC, CDC SVI 2018 Documentation. Accessed at SVI2018Documentation_01192022_1.pdf (cdc.gov) on October 27, 2022.
10 Executive Order 13995, Ensuring an Equitable Pandemic Response and Recovery, Code of Federal Regulations, Vol. 86, No. 15
(enacted January 26, 2021).
11 HHS, HHS Equity Action Plan. Accessed at HHS Equity Action Plan on May 16, 2021.
12 Coronavirus Aid, Relief, and Economic Security (CARES) Act, Div. B, tit. VIII, P.L. 116-136 (enacted March 27, 2020).
13 Ibid.
14 HRSA, Provider Relief Fund: General Information. Accessed at CARES Act Provider Relief Fund: General Information |
HHS.gov on August 19, 2021.
15 HRSA, Provider Relief Fund Past Payments Targeted Distributions. Accessed at Past Targeted Distributions | Official web site
of the U.S. Health Resources & Services Administration (hrsa.gov) on September 8, 2021.
16 HHS, Office of Minority Health, About COVID-19 Health Equity Task Force. Accessed at Health Equity Task Force - The Office
of Minority Health (hhs.gov) on October 1, 2021.
17 HHS, Strategic Plan FY 2018 – 2022, Strategic Objective 2.4: Prepare for and respond to public health emergencies.
Accessed
at Strategic Goal 2: Protect the Health of Americans Where They Live, Learn, Work, and Play | HHS.gov on December 3, 2021.
18 CDC, Health Equity Considerations & Racial & Ethnic Minority Groups. Accessed at Health Equity Considerations and Racial
and Ethnic Minority Groups | CDC on January 25, 2022.
19 HHS, HHS Equity Action Plan. Accessed at HHS Equity Action Plan on May 16, 2021.
20 HHS, HHS Protect FAQ. Accessed at hhs-protect-faqs.pdf on October 25, 2021.
21 Ibid.
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 37
OEI-05-20-00580
22 U.S. Census Bureau, Why We Ask Each Question About / Race, Accessed at Race | American Community Survey | U.S. Census
Bureau on June 11 2023.
23 U.S. Census Bureau, Questionnaire Archive. Accessed at https://www.census.gov/programs-
surveys/acs/methodology/questionnaire-archive.2019.html at on October 7, 2021.
24 HRSA, Defining Rural Population. Accessed at Defining Rural Population | HRSA on April 8, 2022.
25 USA Facts, About USAFacts. Accessed at https://usafacts.org/about-usafacts/ on October 21, 2021.
26 USA Facts, Detailed Methodology and Sources: COVID-19 Data. Accessed at https://usafacts.org/articles/detailed-
methodology-covid-19-data/ on November 12, 2020.
27 HHS, COVID-19 Guidance for Hospital Reporting and FAQs For Hospitals, Hospitals Laboratory, and Acute Care Facility Data
Reporting. Accessed at covid-19-faqs-hospitals-hospital-laboratory-acute-care-facility-data-reporting.pdf< (hhs.gov) on
October 25, 2021.
28 Journal of American Medical Association, Allocation of COVID-19 Relief Funding to Disproportionately Black Counties.
Accessed at Allocation of COVID-19 Relief Funding to Disproportionately Black Counties | Health Disparities | JAMA | JAMA
Network on August 25, 2021.
29 Journal of Health Politics, Policy and Law, Duke University Press, Unsanitized and Unfair: How COVID-19 Bailout Funds Refuel
Inequity in the US Health Care System. Accessed at Unsanitized and Unfair: How COVID-19 Bailout Funds Refuel Inequity in the
US Health Care System | Journal of Health Politics, Policy and Law | Duke University Press (dukeupress.edu) on November 29,
2021.
30 KFF, Changes in Health Coverage by Race and Ethnicity since the ACA, 2010-2018.
Accessed at https://www.kff.org/racial-
equity-and-health-policy/issue-brief/changes-in-health-coverage-by-race-and-ethnicity-since-the-aca-2010-2018/ on March
5, 2020.
31 CARES Act, Div. B, tit. VIII, P.L. 116-136 (enacted March 27, 2020).
32 Paycheck Protection Program and Health Care Enhancement Act, Div. B, tit. I, P.L. 116–139 (enacted April 24, 2020).
33 Coronavirus Response and Relief Supplemental Appropriations Act, 2021, Div. M, tit. III, P.L. 116-260 (enacted December 27,
2020).
34 CARES Act, Div. B, tit. VIII, P.L. 116-136 (enacted March 27, 2020).
35 Ibid.
36 HRSA, COVID-19 Claims Reimbursement to Health Care Providers and Facilities for Testing, Treatment, and Vaccine
Administration for the Uninsured. Accessed at COVID-19 Claims Reimbursement to Health Care Providers and Facilities for
Testing, Treatment, and Vaccine Administration for the Uninsured | Official web site of the U.S. Health Resources & Services
Administration (hrsa.gov) on January 27, 2022.
37 HRSA, COVID-19 Coverage Assistance Fund. Accessed at COVID-19 Coverage Assistance Fund | Official web site of the U.S.
Health Resources & Services Administration (hrsa.gov) on January 27, 2022.
38 HRSA, Provider Relief Fund Past Payments General Distributions. Accessed at Past General Distributions | Official web site of
the U.S. Health Resources & Services Administration (hrsa.gov) on September 8, 2021.
39 Ibid.
40 HRSA, Provider Relief Funds Future Payments.Accessed at Future Payments | Official web site of the U.S. Health Resources
& Services Administration (hrsa.gov) on September 15, 2021.
41 HRSA, Provider Relief Fund Past Payments General Distributions. Accessed at Past General Distributions | Official web site of
the U.S. Health Resources & Services Administration (hrsa.gov) on September 8, 2021.
42 Ibid.
43 Ibid.
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 38
OEI-05-20-00580
44 HRSA, Provider Relief Fund 3: Payment Calculation Methodology. Accessed at Provider Relief Fund Phase 3: Payment
Calculation Methodology (hrsa.gov) on February 7, 2022.
45 HRSA, Provider Relief Funds Future Payments. Accessed at Future Payments | Official web site of the U.S. Health Resources
& Services Administration (hrsa.gov) on October 25, 2022.
46 In March 2020, hospitals reported substantial challenges with mitigation strategies, and needs for assistance related to
personal protective equipment (PPE), testing, staffing, supplies and durable equipment; maintaining or expanding facility
capacity; and financial concerns. OEI-06-20-00300.
47 HHS Press Release, HHS to Begin Immediate Delivery of Initial $30 Billion of CARES Act Provider Relief Funding. Accessed at
2020-12-31 08:51 | Archive of HHS.gov (pagefreezer.com) on January 14, 2022.
48 CARES Act, Div. B, tit. VIII, P.L. 116-136 (enacted March 27, 2020).
49 HRSA, Provider Relief Fund General Information. Accessed at Provider Relief Fund General Information | Official web site of
the U.S. Health Resources & Services Administration (hrsa.gov) on September 17, 2021.
50 HRSA, Provider Relief Fund Past Payments Targeted Distributions. Accessed at Past Targeted Distributions | Official web site
of the U.S. Health Resources & Services Administration (hrsa.gov) on September 8, 2021.
51 Ibid.
52 Ibid.
53 Ibid.
54 Ibid.
55 Ibid.
56 Ibid.
57 HRSA, Current and Future Payments. Accessed at Current and Future Payments | HRSA on December 19, 2022.
58 Ibid.
59 OIG, HHS’s and HRSA’s Controls Related to Selected Provider Relief Fund Program Requirements Could Be Improved. Accessed
at HHS's and HRSA's Controls Related to Selected Provider Relief Fund Program Requirements Could Be Improved A-09-21-
06001 09-26-2022 on December 7, 2022.
60 USDA Economic Research Service, The COVID-19 Pandemic and Rural America. Accessed at USDA ERS - The COVID-19
Pandemic and Rural America on August 17, 2021.
Targeted Provider Relief Funds Allocated to Hospitals Had Some Differences with Respect to the Ethnicity and Race of
Populations Served 39
OEI-05-20-00580
File and source
- File
- targeted-prf-allocated-to-hospitals-had-differences-by-ethnicity-race.pdf
- Size
- 2,069,133 bytes
- SHA-256
- 5940ca2f854fc5c463e178d139671a69157d1e7395cec6fb8c5f85af8374ac54
- Original
- oig.hhs.gov