GAO-23-104291, COVID-19 IN NURSING HOMES: Outbreak Duration Averaged 4 Weeks and Was Strongly Associated with Community Spread
- Issuer
- Government Accountability Office
- Document type
- Report
- Date
- 2020-01-01
Summary
GAO-23-104291 is a Government Accountability Office report to congressional addressees dated December 15, 2022, on COVID-19 outbreaks in nursing homes. It examines how outbreaks affected nursing homes and which nursing home factors were associated with outbreak duration, using CDC and CMS data from June 2020 through December 2021 in a statistical model, a literature review and interviews in California, Florida, Maryland and Michigan. The report finds the average outbreak lasted 4 weeks and that community spread had the strongest association with outbreak duration. It states that, before vaccines, nursing homes in counties with low community spread had outbreaks ending an estimated 7 days earlier, and that 75 percent of outbreaks began with a reported staff case during the first week. Other factors it lists include nursing home size, reported staff shortages and ownership type.
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
Full text
United States Government Accountability Office
Report to Congressional Addressees
COVID-19 IN
December 2022
NURSING HOMES
Outbreak Duration
Averaged 4 Weeks
and Was Strongly
Associated with
Community Spread
GAO-23-104291
December 2022
COVID-19 IN NURSING HOMES
Outbreak Duration Averaged 4 Weeks and Was
Strongly Associated with Community Spread
Highlights of GAO-23-104291, a report to
congressional addressees
Why GAO Did This Study What GAO Found
The COVID-19 pandemic has had a GAO analysis of data from the Centers for Disease Control and Prevention
disproportionate effect on the more (CDC) shows that, from June 2020 through December 2021, nursing homes
than 1 million residents in the nation’s faced many separate COVID-19 outbreaks, with the average outbreak lasting 4
nursing homes. The initial unknown weeks. Officials GAO interviewed at six selected nursing homes in four states
nature of the virus and the scope of the described a range of outbreak experiences, including critical challenges and
pandemic created unprecedented some successes. For example, two critical challenges included:
challenges. The Department of Health
and Human Services, primarily through • Staff shortages. Officials from five nursing homes described experiencing
CMS and CDC, is responsible for staffing shortages during outbreaks. For example, officials from one nursing
taking steps to address COVID-19 in home described a staffing crisis, noting that at one point the home was down
nursing homes, such as issuing about 25 percent of its workforce.
relevant requirements and guidance. • Low staff morale. Officials from three nursing homes discussed the
Many studies have explored the factors psychosocial effect the pandemic had on their staff and the difficulties in
associated with the presence or maintaining staff morale. For example, officials from one nursing home
severity of COVID-19 in nursing described how the outbreaks took away the joy of caregiving from staff, and
homes. However, little was known officials believed that many staff were left traumatized.
about the factors associated with the
GAO’s analysis of CDC and Centers for Medicare & Medicaid Services (CMS)
duration of COVID-19 outbreaks within
nursing homes.
data found that transmission of COVID-19 in the community surrounding a
nursing home, known as community spread, had the strongest association with
The CARES Act directs GAO to the duration of an outbreak. Controlling for other factors, prior to the introduction
monitor the federal pandemic of COVID-19 vaccines, nursing homes in counties experiencing low community
response. GAO was also asked to spread had outbreaks that ended an estimated 7 days earlier than nursing
review CMS oversight of nursing homes in counties experiencing high community spread. GAO also found that
homes in light of the pandemic. This most outbreaks (75 percent) began with a reported staff case during the first
report examines how COVID-19 week. These results could indicate that, during times of higher community
outbreaks affected nursing homes, and spread, staff have a greater likelihood of being exposed to the virus in the
determines how nursing home factors
community and bringing it into the nursing home. Other factors GAO found that
were associated with outbreak
had a strong association with outbreak duration included nursing home size,
duration.
reported staff shortages, and ownership type.
GAO analyzed CDC and CMS nursing
home data from June 2020 through GAO Estimated COVID-19 Outbreak Duration for Nursing Homes with Selected Characteristics
December 2021, in a statistical model,
and conducted a literature review.
GAO also interviewed representatives
from nursing homes and state
agencies from a non-generalizable
sample of four states. The states were
selected for variation in factors such as
number of nursing home beds and
COVID-19 cases.
View GAO-23-104291. For more information, Note: Using multivariate statistical models, GAO estimated the association of nursing home factors,
contact John Dicken at (202) 512-7114 or such as facility characteristics, with the likelihood that an ongoing COVID-19 outbreak would end
dickenj@gao.gov. during a given week.
United States Government Accountability Office
Contents
Letter 1
Background 5
Nursing Home COVID-19 Outbreaks Lasted 4 Weeks on Average
and Selected Nursing Homes Described a Range of Outbreak
Experiences 10
Transmission of COVID-19 in the Surrounding Community Had
the Strongest Association with Outbreak Duration in Nursing
Homes 17
Agency Comments 23
Appendix I Literature Review—Methodology Description and Results 27
Appendix II Detailed Description of Methodology for Multivariate Statistical
Models 44
Appendix III Nursing Home Officials’ Descriptions of COVID-19 Outbreaks,
including Challenges and Successes 62
Appendix IV GAO Contact and Staff Acknowledgments 64
Related GAO Products 65
Tables
Table 1: Nursing Home Officials’ Descriptions of COVID-19
Outbreaks 13
Table 2: Studies Included in GAO’s Review That Focused on
Factors Associated with the Presence, Severity, or
Duration of COVID-19 Outbreaks in Nursing Homes,
January 1, 2020, through October 1, 2021 29
Table 3: Nursing Home Facility Characteristics, Categorical
Variables, June 14, 2020, through January 2, 2022 46
Table 4: Infection Prevention and Control Deficiencies Cited in
Nursing Homes, by Year, 2017 through 2021 49
Page i GAO-23-104291 Nursing Home COVID-19 Outbreaks
Table 5: Nursing Home Average Daily Staffing Levels, June 14,
2020, through January 2, 2022 50
Table 6: Reported Nursing Home Staff Shortages during the
COVID-19 Pandemic, June 14, 2020, through January 2,
2022 51
Table 7: Nursing Home Resident Demographics, June 14, 2020,
through January 2, 2022 52
Table 8: Number of Weeks Counties Were Designated by CDC as
Having High, Substantial, Moderate, or Low Levels of
Community Transmission of COVID-19, June 14, 2020,
through January 2, 2022 53
Table 9: Multivariate Regression Results from Cox Proportional
Hazard Model, COVID-19 Outbreak Duration by
Vaccination Time Period, June 14, 2020, through January
2, 2022 57
Table 10: Nursing Home Officials’ Descriptions of COVID-19
Outbreaks, including Challenges and Successes 62
Figures
Figure 1: New Weekly Confirmed COVID-19 Cases and Deaths
among U.S. Nursing Home Residents and Staff, Weeks
Ending May 31, 2020, through October 2, 2022 9
Figure 2: Duration of COVID-19 Outbreaks in Nursing Homes,
June 14, 2020, through January 2, 2022 11
Figure 3: Average Duration of COVID-19 Outbreaks in Nursing
Homes by Month the Outbreak Began, June 14, 2020,
through January 2, 2022 12
Abbreviations
CDC Centers for Disease Control and Prevention
CMS Centers for Medicare & Medicaid Services
HHS Department of Health and Human Services
This is a work of the U.S. government and is not subject to copyright protection in the
United States. The published product may be reproduced and distributed in its entirety
without further permission from GAO. However, because this work may contain
copyrighted images or other material, permission from the copyright holder may be
necessary if you wish to reproduce this material separately.
Page ii GAO-23-104291 Nursing Home COVID-19 Outbreaks
Letter
441 G St. N.W.
Washington, DC 20548
December 15, 2022
Congressional Addressees
COVID-19 emerged as a new and highly contagious respiratory disease
that has resulted in catastrophic loss of life. These devastating
consequences have been particularly pronounced for the nation’s more
than 1 million nursing home residents. A disproportionate share of all
COVID-19 deaths reported by the Centers for Disease Control and
Prevention (CDC) has been among nursing home residents. 1 The initial
unknown nature of the virus that causes COVID-19, along with the scope
of the pandemic, created unprecedented challenges for the state and
federal agencies working to ensure the quality of care delivered in the
more than 15,000 Medicare- and Medicaid-certified nursing homes. 2
Nursing home residents are at a high risk of infection and death due to
COVID-19, as older adults and those with underlying health conditions
are at higher risk of severe disease. 3 In addition, the congregate nature of
nursing homes, with staff caring for multiple residents and communal
spaces, increases the risk that COVID-19 will easily spread. The
introduction of COVID-19 vaccines in conjunction with improved infection
prevention and control practices coincided with a sharp decline in nursing
1Combined nursing home resident and staff deaths from COVID-19 consistently
represented about 30 percent of all COVID-19 deaths in the U.S. from May 2020 through
February 2021. In May 2021, this percentage declined to about 23 percent and further
decreased to about 16 percent in May 2022. See Centers for Disease Control and
Prevention, COVID Data Tracker, accessed September 7, 2022,
https://covid.cdc.gov/covid-data-tracker.
2Federal statutes and their implementing regulations use the terms “skilled nursing facility”
(Medicare) and “nursing facility” (Medicaid). For the purposes of this report, we use the
term “nursing home” to refer to both skilled nursing facilities and nursing facilities.
Medicare, the federal health insurance program for people age 65 and older, individuals
under age 65 with certain disabilities, and individuals diagnosed with end-stage renal
disease, covers some short-term skilled nursing and rehabilitative care for beneficiaries
following an acute care hospital stay. Medicaid, a joint federal-state health program for
certain low-income and medically needy individuals, is the nation’s primary payer of long-
term services and supports for children and adults with disabilities and aged individuals.
3Centers for Disease Control and Prevention, COVID-19: People Who Live in a Nursing
Home or Long-Term Care Facility, accessed January 5, 2021,
https://www.cdc.gov/coronavirus/2019-ncov/need-extra-precautions/people-in-nursing-ho
mes.html.
Page 1 GAO-23-104291 Nursing Home COVID-19 Outbreaks
home cases and deaths from their initial peaks in December 2020.
However, the emergence of more transmissible virus variants led to the
highest resident and staff case rates in nursing homes thus far in January
2022.
While there have been several studies exploring the factors associated
with the presence and severity of COVID-19 in nursing homes, little is
known about the factors that are associated with the duration of
outbreaks once the virus enters the home. 4
The CARES Act includes a provision directing us to monitor the federal
response to the COVID-19 pandemic. 5 Further, you asked us to examine
the Centers for Medicare & Medicaid Services’ (CMS) oversight of
infection prevention and control protocols and the adequacy of
emergency preparedness standards for emerging infectious diseases in
nursing homes, as well as CMS’s response to the pandemic. Since 2020,
we have examined multiple aspects of the response to COVID-19 in
nursing homes. Some reports are published while work on others is
ongoing. 6
This report (1) examines how COVID-19 outbreaks affected nursing
homes and (2) determines how certain nursing home factors were
associated with the duration of COVID-19 outbreaks.
To understand how COVID-19 outbreaks have affected nursing homes,
we analyzed CDC data on nursing home COVID-19 cases. To help inform
nursing home response efforts, CDC defines the start of a COVID-19
4Studies have found a range of factors, such as COVID-19 transmission in the
surrounding community, racial disparities, and staffing levels, that are associated with the
presence or severity of COVID-19 cases and deaths in nursing homes. See appendix I for
a list of such studies.
5Pub. L. No. 116-136, § 19010(b), 134 Stat. 281, 580 (2020). Throughout the pandemic,
we regularly issued government-wide reports on the federal response to COVID-19. All
government-wide reports are available on GAO’s website at
https://www.gao.gov/coronavirus.
6For example, see GAO, Infection Control Deficiencies Were Widespread and Persistent
in Nursing Homes Prior to COVID-19 Pandemic, GAO-20-576R (Washington, D.C.: May
20, 2020); COVID-19 in Nursing Homes: Most Homes Had Multiple Outbreaks and Weeks
of Sustained Transmission from May 2020 through January 2021, GAO-21-367
(Washington, D.C.: May 19, 2021); and COVID-19 in Nursing Homes: CMS Needs to
Continue to Strengthen Oversight of Infection Prevention and Control, GAO-22-105133
(Washington, D.C.: Sep. 14, 2022).
Page 2 GAO-23-104291 Nursing Home COVID-19 Outbreaks
outbreak as the week a nursing home reports a new COVID-19 case in a
resident or staff member and defines the end as when the nursing home
has 2 consecutive weeks where it reports no such cases. 7 Using this
definition of an outbreak, we first generated descriptive statistics,
including the number and duration of outbreaks each of the over 15,000
Medicare- and Medicaid-certified nursing homes experienced. We did not
include the 2 consecutive weeks without a new COVID-19 case when
calculating an outbreak’s duration.
In addition, we interviewed state officials and nursing home officials in a
non-generalizable sample of four states: California, Florida, Maryland,
and Michigan. These states were selected to provide variation across: (1)
geographic location; (2) number of nursing home beds; and (3) number of
nursing home residents and staff with confirmed positive cases of COVID-
19. 8 In each of these four states, we interviewed state survey agency and
state long-term care ombudsman officials, as well as officials from at least
one nursing home. 9 In total, we interviewed officials from six nursing
homes that we selected to obtain variation in factors such as bed count
and profit or not-for-profit status. 10 We asked nursing home officials to
describe how COVID-19 outbreaks affected their nursing home, as well
as to describe the challenges faced while preventing and managing
COVID-19 outbreaks.
Additionally, we interviewed researchers with work related to nursing
homes and COVID-19, as well as national associations with knowledge of
nursing homes issues. These associations included the American Health
Care Association and National Consumer Voice for Quality Long-Term
7See Centers for Disease Control and Prevention, Interim Infection Prevention and
Control Recommendations to Prevent SARS-CoV-2 Spread in Nursing Homes, accessed
April 18, 2022, https://www.cdc.gov/coronavirus/2019-ncov/hcp/long-term-care.html.
8Nursing home characteristics data were as of June 2020, and COVID-19 case rates were
as of November 1, 2020.
9CMS enters into agreements with survey agencies in each state (known as state survey
agencies) to monitor compliance with federal quality of care standards by conducting both
recurring comprehensive standard surveys (inspections) and as-needed investigations.
Under the federal Older Americans Act, every state is required to have an Ombudsman
Program that addresses complaints and advocates for improvements in the long-term care
system, including nursing homes, as a condition of receiving funds under the act.
10We conducted the original interviews from December 2020 through August 2021 and
followed up with officials at each state survey agency, state long-term care ombudsman,
and nursing home from November 2021 through March 2022.
Page 3 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Care. We discussed with researchers and association representatives the
challenges faced by nursing homes during the pandemic.
To determine how certain nursing home factors were associated with the
duration of COVID-19 outbreaks, we analyzed CMS data on nursing
home staffing levels, resident demographics, and facility characteristics;
CDC data on county-level COVID-19 community transmission levels and
nursing home COVID-19 cases; and Health Resources and Services
Administration data on county-level metropolitan, micropolitan, and rural
designations. We developed multivariate statistical models to determine
the significance of each selected factor in estimating the probability of an
outbreak ending at any given week from June 14, 2020, through January
2, 2022 (the last full week of the 2021 calendar year).
Specifically, our statistical models controlled for the transmission levels of
COVID-19 in the surrounding community, several nursing home
characteristics, several measures of nursing home staffing levels, and
several other factors, including prior infection prevention and control
deficiencies. We could not control for all factors that may affect outbreak
duration, such as state and local COVID-19 policies, local compliance
with those policies, or nursing home adherence to infection prevention
and control practices. Therefore, our model provides information on
possible associations in the data and does not establish a causal
relationship between the factors and outbreak duration.
To provide context to these findings, we conducted a literature review to
determine what is known about the factors associated with COVID-19
outbreaks in nursing homes and identified 40 pertinent studies, which we
reviewed. For a description of our literature review methodology and its
findings, see appendix I.
We assessed the reliability of the various data sets we used by reviewing
relevant CDC, CMS, and Health Resources and Services Administration
data documentation, interviewing knowledgeable agency officials, and
performing tests of the data to identify any outliers or anomalies. We
determined that these datasets were sufficiently reliable for the purposes
of our reporting objectives. For additional information about our
descriptive and statistical analyses and model, see appendix II.
We conducted this performance audit from April 2020 to December 2022
in accordance with generally accepted government auditing standards.
Those standards require that we plan and perform the audit to obtain
sufficient, appropriate evidence to provide a reasonable basis for our
Page 4 GAO-23-104291 Nursing Home COVID-19 Outbreaks
findings and conclusions based on our audit objectives. We believe that
the evidence obtained provides a reasonable basis for our findings and
conclusions based on our audit objectives.
The Department of Health and Human Services (HHS), primarily through
Background CMS and CDC, has led the response to the COVID-19 pandemic in
nursing homes. CMS is the federal oversight agency responsible for
ensuring that nursing homes meet federal quality standards to be eligible
to participate in the Medicare and Medicaid programs. 11 CDC issues
guidance with recommendations for preventing and managing infectious
diseases and operates infectious disease surveillance systems. CDC also
provides technical assistance through programs aimed at supporting and
assessing infection prevention and control in nursing homes, and tracking
infection prevention and control data. 12
The COVID-19 pandemic has had a disproportionate effect on nursing
home residents thus far. Nursing home residents are at a high risk of
infection and death due to COVID-19, as older adults and those with
underlying health conditions are at a higher risk of severe disease. In
addition, the congregate nature of nursing homes increases the risk of
transmission. 13 We have previously reported on a number of COVID-19
issues in nursing homes, including challenges nursing homes have faced
and the number and duration of outbreaks early in the pandemic. For
example, our past work found that most nursing homes had multiple
11To monitor compliance with these standards, CMS enters into agreements with survey
agencies in each state to conduct recurring comprehensive standard surveys and as-
needed investigations. Beginning in March 2020, CMS required state survey agencies to
conduct focused infection control surveys, a new type of survey in response to the
pandemic with a narrower scope than a standard survey, for 20 percent of nursing homes
in their state annually, prioritizing those facilities that report new COVID-19 cases and low
vaccination rates. If a state survey agency determines that a nursing home violated a
federal standard, the nursing home is cited for the deficiency.
12According to CDC officials, this includes developing resources to support
implementation of prevention and surveillance activities, conducting and supporting onsite
and remote infection prevention and control assessments, and supporting outbreak field
investigations.
13According to CDC, COVID-19 is spread in three main ways: (1) breathing in small
droplets or particles exhaled by an infected person; (2) having these small droplets and
particles land on the eyes, nose, or mouth, especially through a cough or a sneeze; and
(3) touching eyes, nose, or mouth with hands that have the virus on them. See Centers for
Disease Control and Prevention, How COVID-19 Spreads, accessed April 16, 2022,
https://www.cdc.gov/coronavirus/2019-ncov/prevent-getting-sick/how-covid-spreads.html.
Page 5 GAO-23-104291 Nursing Home COVID-19 Outbreaks
outbreaks and weeks of sustained COVID-19 transmission from May
2020 through January 2021. 14
Many studies have explored the factors associated with the presence or
severity of COVID-19 in nursing homes, finding a range of associated
factors, such as COVID-19 in the surrounding community, nursing home
size, and racial disparities. For example, several studies all found that
nursing homes with fewer beds were associated with less severe
outbreaks. 15
The duration of an outbreak within a nursing home is also important as it
may indicate how well a nursing home can control the spread of the virus,
as well as having implications related to staffing needs, space constraints,
and resident well-being. For example, CDC guidance recommends a
nursing home maintain dedicated space for residents with confirmed
infections during a COVID-19 outbreak, if possible. 16 This can require
additional nursing staff and strain a home’s available space and
resources. In addition, in March 2020, CMS released guidance restricting
visitation and suspending group dining and activities to prevent the
spread of the virus. Initially these restrictions were at all times; starting in
September 2020, visitation was restricted only during active COVID-19
14For challenges, see, for example, GAO, COVID-19: Urgent Actions Needed to Better
Ensure an Effective Federal Response, GAO-21-191 (Washington, D.C.: Nov. 30, 2020).
For number and duration of outbreaks, see GAO, COVID-19 in Nursing Homes: Most
Homes Had Multiple Outbreaks and Weeks of Sustained Transmission from May 2020
through January 2021, GAO-21-367 (Washington, D.C.: May 19, 2021).
15See appendix I. For example, Elizabeth M. White et al., “Variation in SARS-CoV-2
Prevalence in U.S. Skilled Nursing Facilities,” Journal of the American Geriatrics Society,
vol. 68, no. 10 (2020): 2167-2173, and Sheryl Zimmerman, et al., “Nontraditional Small
House Nursing Homes Have Fewer COVID-19 Cases and Deaths,” JAMDA, vol. 22, no. 3
(2021): 489-493.
16CDC guidance specifies that staff assigned to the COVID-19 unit only work in this unit
when it is in use and that, at a minimum, such staff include the primary nursing assistants
and nurses assigned to care for these residents. See Centers for Disease Control and
Prevention, Interim Infection Prevention and Control Recommendations to Prevent SARS-
CoV-2 Spread in Nursing Homes, accessed July 24, 2022,
https://www.cdc.gov/coronavirus/2019-ncov/hcp/long-term-care.html.
Page 6 GAO-23-104291 Nursing Home COVID-19 Outbreaks
outbreaks. 17 These restrictions had unintended consequences on
residents, such as isolation and declining health. 18 The longer a nursing
home is responding to a COVID-19 outbreak, the longer staff and
residents face these issues. However, we found that little was known
about the factors that contribute to the length of outbreaks once the virus
enters the home. 19
CMS has required that nursing homes report COVID-19 information
weekly to CDC, including resident and staff case and death counts, since
May 2020. 20 CDC data show that, for the week ending May 31, 2020,
through the end of 2020, there were fluctuations in new weekly confirmed
cases among residents and staff. The introduction of vaccines in nursing
homes at the end of December 2020, in conjunction with improved
infection prevention and control practices, was a key turning point when
cases and deaths began to decline. COVID-19 nursing home cases and
deaths remained low through the first part of 2021, but, from the summer
of 2021 through the end of that calendar year (coinciding with the
emergence of the COVID-19 Delta variant), nursing home cases began to
rise. Then, in January 2022 (coinciding with the emergence of the
COVID-19 Omicron variant), cases in residents and staff quickly jumped
to levels far exceeding the peak seen prior to the introduction of COVID-
17CMS initially restricted visitation and suspended group dining and activities in March
2020. After the initial restrictions, CMS changed its guidance multiple times—in
September 2020, March 2021, and April 2021—to allow for more visitation and group
activities, while identifying some situations where limitations would be appropriate. In
November 2021, all visitation limitations were fully lifted. See Centers for Medicare &
Medicaid Services, Guidance for Infection Control and Prevention of COVID-19 in Nursing
Homes, QSO-20-14-NH, (Baltimore, Md.: March 13, 2020) and Centers for Medicare &
Medicaid Services, Nursing Home Visitation – COVID-19, QSO-20-39-NH, (Baltimore,
Md.: Sept. 17, 2020), revised March 10, 2021, April 27, 2021, and November 12, 2021.
18We have reported on the worsening of several key indicators of nursing home resident
mental and physical health during 2020 and 2021. Nursing home officials and national
organizations attributed this worsening in part to the isolation residents experienced. See
GAO, COVID-19 in Nursing Homes: CMS Needs to Continue to Strengthen Oversight of
Infection Prevention and Control, GAO-22-105133 (Washington, D.C: Sept. 14, 2022).
19One study in our literature review examined the number of weeks a nursing home was in
the top decile of the national distribution of weekly confirmed resident COVID-19 cases,
which is described as the persistence of the outbreak. The study found that nursing home
quality ratings were associated with COVID-19 persistence. See Christianna S. Williams,
et al., “The association of nursing home quality ratings and spread of COVID-19,” Journal
of the American Geriatrics Society, vol. 69, no. 8 (2021): 2070-2078.
2042 C.F.R. § 483.80(g) (2021).
Page 7 GAO-23-104291 Nursing Home COVID-19 Outbreaks
19 vaccines. 21 However, COVID-19 deaths have not risen at the same
rate during these more recent peaks in cases, likely due to protection
from severe illness and death provided by vaccines. 22 See figure 1.
21According to CDC, viruses, such as the virus that causes COVID-19, constantly change
through mutation, and new variants are expected to occur. The Delta variant became the
dominant COVID-19 strain circulating in the U.S. during the week ending July 3, 2021.
The Omicron variant superseded the Delta variant to become the dominant COVID-19
strain circulating in the U.S. during the week ending December 25, 2021. As of April 5,
2022, CDC has reported the presence of multiple sublineages of the Omicron variant in
the U.S., including BA.2; BA.2 became the dominant strain circulating in the U.S. during
the week ending March 26, 2022.
22As of May 1, 2022, according to CDC data, nursing homes reported that 88.2 percent of
residents and 89.7 percent of staff had a complete vaccination—meaning the resident had
received all doses required to be fully vaccinated (two doses of a two-dose mRNA series
or one dose of a single-dose vaccine). Additionally, nursing homes reported that 80.6
percent of residents and 53.2 percent of staff had received an additional primary or
booster dose of the vaccine.
Page 8 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Figure 1: New Weekly Confirmed COVID-19 Cases and Deaths among U.S. Nursing Home Residents and Staff, Weeks Ending
May 31, 2020, through October 2, 2022
Page 9 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Notes: Dates refer to the end of a week (e.g., May 31, 2020, refers to the entire week from May 25,
2020, through May 31, 2020).
According to CDC, data used in this figure are part of a live data set, meaning that facilities can
correct the data at any time. This figure reflects data downloaded as of October 24, 2022, and
includes data through the week ending October 2, 2022. We excluded data for the week ending May
24, 2020, because it is the first week for which data are available and includes cases and deaths from
multiple weeks dating back to January 1, 2020.
Weekly case and death counts are likely underreported because they do not include data for the
nursing homes that did not report COVID-19 data to CDC for that week or from nursing homes that
submitted data that failed data quality assurance checks. Additionally, CMS does not require nursing
homes to report data prior to May 2020, although nursing homes may do so voluntarily.
From June 14, 2020, through January 2, 2022, nursing home COVID-19
Nursing Home outbreaks lasted an average of 4 weeks. 23 Officials we interviewed at six
COVID-19 Outbreaks selected nursing homes described a range of experiences during such
outbreaks, including many challenges and some areas of success.
Lasted 4 Weeks on
Average and Selected
Nursing Homes
Described a Range of
Outbreak
Experiences
Nursing Homes Nursing homes experienced many COVID-19 outbreaks from June 14,
Experienced Many 2020, through January 2, 2022, with the average outbreak lasting 4
weeks. Specifically, the 15,281 nursing homes in our analysis
COVID-19 Outbreaks from
experienced from one to 16 outbreaks, with an average of 7.6 outbreaks
June 2020 through and a total of 102,992 outbreaks across all homes during our period of
December 2021 with the review. 24 The average outbreak duration was 4 weeks, ranging from 1
Average Outbreak Lasting week (42 percent of total outbreaks) to 53 weeks (less than 1 percent of
4 Weeks total outbreaks). While shorter outbreaks were more common, nearly all
nursing homes (13,898 of 15,281 nursing homes, or 91 percent)
experienced at least one longer-lasting outbreak of 5 or more weeks. 25
23An outbreak begins when a nursing home reports a new case of COVID-19 in a resident
or staff member, and an outbreak ends when the nursing home has 2 consecutive weeks
without a new COVID-19 case. The outbreak duration does not include the 2 consecutive
weeks without a new COVID-19 case.
24During our period of review, 25 nursing homes reported zero COVID-19 outbreaks.
25We previously reported that, from May 2020 through January 2021—a shorter period of
review than this analysis—most nursing homes had multiple COVID-19 outbreaks (with an
average of about three outbreaks), and 85 percent of homes had outbreaks lasting 5 or
more weeks. See GAO-21-367.
Page 10 GAO-23-104291 Nursing Home COVID-19 Outbreaks
(See fig. 2). This illustrates how nursing homes were heavily affected by
the COVID-19 pandemic—requiring nursing home staff to respond to, and
residents to experience, repeated outbreaks and weeks of continued
spread of the virus.
Figure 2: Duration of COVID-19 Outbreaks in Nursing Homes, June 14, 2020, through January 2, 2022
Note: An outbreak begins when a nursing home reports a new case of COVID-19 in a resident or staff
member, and an outbreak ends when the nursing home has 2 consecutive weeks without a new
COVID-19 case. The outbreak duration does not include the 2 consecutive weeks without a new
COVID-19 case.
We found that the average duration of COVID-19 outbreaks beginning
near the end of both 2020 and 2021 were longer than outbreaks
beginning at other points in either year. Specifically, the longest average
outbreaks began in November 2020 and December 2021, lasting an
average of 7 weeks and ranging from 1 to 32 weeks and from 1 to 17
weeks, respectively. (See fig. 3). These trends follow a similar pattern to
the trends seen in the number of COVID-19 cases reported in nursing
homes over the same periods. There were spikes in cases in late 2020,
prior to the availability of vaccines in nursing homes, and at the beginning
of 2022, corresponding with the emergence of the initial COVID-19
Omicron variant. However, the increases in average outbreak duration
appear to precede the spikes in case numbers reported in nursing homes
by approximately 1 month. This may be because any outbreaks beginning
Page 11 GAO-23-104291 Nursing Home COVID-19 Outbreaks
before the spike in cases continued to spread due to the growing
caseloads.
Figure 3: Average Duration of COVID-19 Outbreaks in Nursing Homes by Month the Outbreak Began, June 14, 2020, through
January 2, 2022
Note: An outbreak begins when a nursing home reports a new case of COVID-19 in a resident or staff
member, and an outbreak ends when the nursing home has 2 consecutive weeks without a new
COVID-19 case. The outbreak duration does not include the 2 consecutive weeks without a new
COVID-19 case.
High rates of COVID-19 in the community surrounding a nursing home
can increase the risk that the virus will enter the home (for example,
through staff or visitors) and could spread. We found that nursing homes
Page 12 GAO-23-104291 Nursing Home COVID-19 Outbreaks
reported that most outbreaks began with a staff case. 26 Specifically, 74.9
percent of all outbreaks began with a reported staff case the first week of
the outbreak, 12.0 percent began with a reported resident case, and 13.1
percent began with both a reported staff and resident case. 27
Nursing Homes Described Officials at the six nursing homes we interviewed described a range of
a Range of Outbreak experiences during their COVID-19 outbreaks. In describing their
experiences, the nursing home officials highlighted a number of
Experiences, Noting
challenges as well as successes when discussing how they prevented
Challenges and and managed outbreaks. See table 1 for a summary of each nursing
Successes home’s experience with COVID-19 outbreaks. For more information on
these nursing homes’ experiences during outbreaks, see Appendix III.
Table 1: Nursing Home Officials’ Descriptions of COVID-19 Outbreaks
Nursing home Nursing home officials’ descriptions of COVID-19 outbreaks
Nursing home A This nursing home’s first case was a staff member in July 2020, and it was caught early. The required
biweekly COVID-19 testing caught additional staff cases throughout the pandemic. The nursing home’s
first resident outbreak was in October 2020 and started with five residents and then spread to an
additional 16 residents. The nursing home officials believed that COVID-19 entered the home from a
resident who left the home regularly for dialysis. A year later, in October 2021, the home had a second
major resident outbreak.
Nursing home B This nursing home experienced its largest outbreak in December 2021 with the Omicron surge. During
this time, 20 percent of its residents and staff tested positive. Officials attributed this outbreak in part to
the re-opening of visitation that occurred in November 2021. Prior to the outbreak, the home had only
seen a few cases at a time, and its COVID-19-positive residents had avoided severe illness.
Nursing home C Officials from this nursing home said that it kept COVID-19 out of the facility until May 2020. The first of
two resident outbreaks started with a few staff contracting COVID-19 from the community. Soon after,
two residents tested positive, then two more, and so on. During the first outbreak, the virus spread slowly,
but the second outbreak in late 2020 spread more rapidly—from two residents to 10 or 12 residents
almost immediately.
Nursing home D Early in 2020, a COVID-19 outbreak affected about a dozen residents. The nursing home’s second
outbreak in the fall of 2020 was very hard on residents and staff, with 40 residents dying. Officials
believed the virus entered the facility through staff as, at the time, the home’s staff and the surrounding
community had low vaccination rates. In addition to the emotional toll on residents and staff, the number
of resident deaths plus staff attrition resulted in the nursing home losing about 50 percent of its capacity
and closing a sister nursing home.
26Our calculation is based on data reported by nursing homes to CDC on a weekly basis.
We do not take into account how many staff or residents were infected during the week,
but just identify which category was reported first. Multiple contagion lines are possible for
any outbreak, including contagion lines that were not detected by the testing regiment in
place at the time.
27This is consistent with our prior reporting where we found that 66 percent of all nursing
homes from May 2020 through January 2021 reported that their longest-lasting COVID-19
outbreak started with a reported staff case the first week of the outbreak. See
GAO-21-367.
Page 13 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Nursing home Nursing home officials’ descriptions of COVID-19 outbreaks
Nursing home E Officials from this nursing home said that it kept COVID-19 out of the facility until July 2020. The outbreak
lasted about 5 weeks and culminated with 32 of about 70 residents and 12 staff infected. Eight of those
residents died. After that outbreak, the nursing home did not experience any additional large outbreaks,
only a few isolated cases. Officials could not trace the origin of the outbreak, but believed that COVID-19
was brought into the facility by a staff member.
Nursing home F Early in the pandemic, this nursing home had some staff members test positive, but COVID-19 was not
transferred to the residents. The first resident did not test positive until November 2020. Officials believed
that it was brought into the facility by staff. The home had another outbreak in January 2021 and was
able to contain COVID-19 to just one floor of their three-floor facility for a time, but eventually the virus
spread.
Source: GAO interviews with officials from selected nursing homes in four states. │GAO-23-104291
While describing their experiences with COVID-19 outbreaks, officials
from all six selected nursing homes highlighted a number of critical
challenges their homes faced. Specifically, the challenges they described
included staffing issues, physical and psychosocial effects on residents,
physical space constraints, and supply issues.
• Staffing issues. Officials from five of the six nursing homes described
challenges related to staffing, including staffing shortages and issues
with maintaining staff morale. In addition, the long-term care
ombudsman and state survey agency officials in all four selected
states also highlighted issues around insufficient staffing, particularly
how it can negatively affect resident care. 28
• Staff shortages. Officials from five of the six nursing homes
described experiencing staff shortages during COVID-19
outbreaks. 29 For example, Nursing Home A officials described a
28For example, one long-term care state ombudsman described that, because of staffing
shortages, staff were only able to address the basic needs of residents, which has
resulted in reports of an increase in pressure sores and other issues of resident neglect.
Another state long-term care ombudsman described resident concerns of having
insufficient staff to meet their needs, where residents of one nursing home described
being left on the toilet or bedpan for an hour waiting for staff assistance.
29In September 2020 we reported that, as of July 26, 2020, CDC data showed that 18
percent of nursing homes were reporting a shortage of aides, about 16 percent were
reporting a shortage of nursing staff, about 9 percent were reporting a shortage of other
staff, and 2 percent were reporting a shortage of clinical staff. These shortages persisted
over time. See GAO, COVID-19: Federal Efforts Could Be Strengthened by Timely and
Concerted Actions, GAO-20-701 (Washington, D.C.: Sept. 21, 2020), 131; COVID-19:
Urgent Actions Needed to Better Ensure an Effective Federal Response, GAO-21-191
(Washington, D.C.: Nov. 30, 2020), 68; COVID-19: Critical Vaccine Distribution, Supply
Chain, Program Integrity, and Other Challenges Require Focused Federal Attention,
GAO-21-265 (Washington, D.C.: Jan. 28, 2021), 64; and COVID-19: Sustained Federal
Action Is Crucial as Pandemic Enters Its Second Year, GAO-21-387 (Washington, D.C.:
Mar. 31, 2021), 77.
Page 14 GAO-23-104291 Nursing Home COVID-19 Outbreaks
staffing “crisis,” noting that at one point the home lost about 25
full-time employees, an approximately 25 percent reduction in
workforce. Nursing Home E officials described losing about half of
its staff, which resulted in the closing of a section of the home and
a reliance on traveling nurses and certified nurse assistants to
replenish its workforce. Nursing Home F officials noted that, with
such limited staff, they had to make difficult choices, including
moving the remaining staff between COVID-19 and non-COVID-
19 units, which increased the risk of spreading the virus to
uninfected residents. Further, Nursing Home D officials explained
that being located in a rural area meant that they had fewer
available staff, which sometimes resulted in “scary” staffing
shortages.
• Staff morale. Officials from three of the six nursing homes
discussed the psychosocial effect the pandemic had on their staff
and the difficulties in maintaining staff morale. For example,
Nursing Home D officials described how the outbreaks took away
from staff the joy of caregiving and felt that many staff were left
traumatized. In addition, Nursing Home B officials found it difficult
to find staff willing to work in the COVID-19 units due to the risk of
exposure to the virus.
• Physical and psychosocial effects on residents. Officials from five
of the six nursing homes described challenges with balancing the
costs and benefits of nursing home visitation restrictions and other
changes to resident rights. 30 While visitation restrictions may have
helped prevent or mitigate the spread of COVID-19 in nursing homes,
interviewees highlighted the physical and psychosocial harm that
these restrictions caused for residents and families. For example,
Nursing Home C officials noted that the home relies heavily on social
interaction, so, even more than visitation, the suspension of dining
facility activities resulted in declines in resident physical and mental
health. Specifically, the staff began to see unplanned resident weight
loss and an increased need for anti-depressant drugs. Officials told us
many of the residents just “gave up.” All four selected states’ long-
30These challenges are consistent with our prior reporting, which found worsening mental
and physical health indicators among nursing home residents during the first year of the
pandemic. See GAO-22-105133.
Page 15 GAO-23-104291 Nursing Home COVID-19 Outbreaks
term care ombudsmen and two of the four states’ survey agencies
also highlighted this challenge. 31
• Physical space constraints. Officials from five of the six nursing
homes described how the nursing homes’ physical space limited their
ability to comply with COVID-19 infection control guidelines. For
example, Nursing Home E officials explained that, in order to isolate
COVID-19-positive residents, the home had to shuffle healthy
residents from one room to another, which may have unintentionally
spread the virus. In addition, Nursing Home C officials described how
they had to keep expanding the boundaries of the COVID-19 unit as
more and more residents were diagnosed with the virus. Among the
four selected states, one long-term care ombudsmen and officials
from two state survey agencies also noted this challenge. 32
• Supply issues. Officials from five of the six nursing homes discussed
difficulty acquiring sufficient personal protective equipment early in the
pandemic. 33 Officials at Nursing Home E explained that the limited
supply of N95 respirators exacerbated an outbreak in July 2020, as
many residents and some staff had to use regular procedure masks
instead, which are not as protective against the virus. Three selected
states’ long-term care ombudsmen and officials at all four states’
survey agencies also described this as a challenge early in the
pandemic. 34
31One state long-term care ombudsman said that her staff had witnessed loneliness,
weight loss, and significant cognitive declines among residents they had known prior to
the pandemic; the official said that residents who had previously recognized ombudsmen
staff now looked at them “shell-shocked and blank.” Another state’s long-term care
ombudsman said that some residents appeared to have aged 10 years over a 2- or 3-
month time period.
32Officials at the two state survey agencies said that the physical layout and aging
infrastructure of nursing homes made it difficult to establish separate COVID-19 zones
and isolation rooms.
33This is consistent with our prior reporting on nursing home supply shortages during the
COVID-19 pandemic. For example, in September 2020 we reported that, as of July 26,
2020, CDC data showed that 22 percent of nursing homes were reporting that they did not
have a 1-week supply of one or more types of personal protective equipment (N95
respirators, surgical masks, gloves, eye protection, or gowns). However, as we reported in
January 2021, by December 6, 2020, this number had decreased to 10 percent of nursing
homes. See GAO-20-701, 129; GAO-21-191, 67; and GAO-21-265, 63.
34One state long-term care ombudsman described how residents could not get the
therapy they needed as providers would not come in to a home that lacked adequate
personal protective equipment.
Page 16 GAO-23-104291 Nursing Home COVID-19 Outbreaks
While describing their experiences with COVID-19 outbreaks, officials
from five of the six nursing homes we interviewed also noted some areas
of success. For example, officials from three nursing homes described
having success early in the pandemic preventing COVID-19 from entering
the home. Nursing Home B officials attributed their early success to
county actions that resulted in low levels of COVID-19 in the community
surrounding the nursing home. Specifically, Nursing Home B is located in
an isolated area and the county decided to close roads into the area to
keep travelers out. Nursing Homes A and C officials discussed proactive
measures that they took to prevent COVID-19, such as suspending new
admissions, restricting residents to their rooms, and adopting face mask
requirements. However, ultimately, all six selected nursing homes had
COVID-19 outbreaks.
Based on our analyses, we determined that transmission of COVID-19 in
Transmission of the community surrounding a nursing home—known as community
COVID-19 in the spread—had the strongest association with the duration of an outbreak in
the home, followed by the size of the nursing home, whether the nursing
Surrounding home reported staff shortages, and the type of nursing home ownership. 35
Community Had the Specifically, prior to the introduction of COVID-19 vaccines, being a small,
for-profit nursing home that did not report staffing shortages and was
Strongest Association located in a county experiencing low community spread was associated
with Outbreak with a higher probability of a COVID-19 outbreak ending roughly 12 days
earlier when compared to a large, government-owned nursing home that
Duration in Nursing reported staff shortages and was located in a county experiencing high
Homes community spread.
35These factors were statistically significant at a p-value less than 0.05 and had a
significant effect on the probability of an outbreak ending in a given week, which we
defined as at least plus or minus 1 day (parameter estimate of at least plus or minus 0.14)
in either one of the two models. We created two models—one for prior to the introduction
of COVID-19 vaccines in nursing homes and one for after the introduction of vaccines in
nursing homes. Additional factors were also statistically significant at a p-value less than
0.05, but their impact on outbreak duration was smaller. See table 9 in appendix II for a list
of all factors analyzed.
CDC COVID-19 community transmission levels—high, substantial, moderate, or low—
describe the amount of COVID-19 spread within each county based on the total new
cases per 100,000 persons within the last 7 days and the percentage of positive
diagnostic and screening nucleic acid amplification tests during the last 7 days. Health
care facilities use these transmission levels to determine infection control interventions.
This is unique from CDC COVID-19 community levels, which the agency recommends
that individuals and communities use to decide which prevention actions to take based on
the latest information.
Page 17 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Using multivariate statistical models, we estimated the association of
nursing home factors with the likelihood that an ongoing COVID-19
outbreak would end during a given week over two different time periods—
prior to the introduction of COVID-19 vaccines in nursing homes and after
the introduction of vaccines in nursing homes. 36 Generally, the magnitude
of the association of a given factor on outbreak duration was lower in the
period after the introduction of vaccines compared to the period prior to
the introduction of vaccines. 37 This was the case for community spread,
nursing home size, and nursing home-reported staff shortages, but to
varying degrees.
Community Spread Community spread of COVID-19 had the strongest association with the
duration of an outbreak in a nursing home throughout our period of
review. Specifically, we found that nursing homes located in counties
experiencing low, moderate, or substantial levels of community spread
had a higher probability of their outbreaks ending during a given week—
meaning their outbreaks were shorter—when compared to nursing homes
in counties with high community spread.
The magnitude of the association of community spread on outbreak
duration was particularly pronounced prior to the introduction of vaccines.
Specifically, in the period of time prior to the introduction of vaccines, our
model estimated that the duration of a COVID-19 outbreak in a nursing
home located in a county experiencing low community spread was
approximately 7 days shorter than homes located in counties
experiencing high community spread. Even after the introduction of
COVID-19 vaccines, community spread was still associated with outbreak
duration—the outbreak duration was an estimated 5 days shorter for
36We ran two models—one for the period of time prior to the introduction of COVID-19
vaccines (June 14, 2020, through January 3, 2021) and one for the period of time after
introduction of COVID-19 vaccines in nursing homes (January 4, 2021, through January 2,
2022). The first nursing home vaccine clinics were held on December 21, 2020. According
to CDC, it takes 2 weeks or more for a vaccine to reach full effectiveness, so we set the
start date for the period after the introduction of vaccines as 2 weeks after the first vaccine
clinics. See appendix II for a detailed description of our methodology.
37Throughout the pandemic there have been changes to federal, state, and local COVID-
19 policies, improvements in infection prevention and control practices, increased
availability of personal protective equipment, and improvements to and increased
availability for COVID-19 testing. We have previously reported on the importance of
infection prevention and control practices and CDC continues to emphasize the
importance of rigorous infection prevention and control practices to prevent the spread of
COVID-19 in nursing homes. See GAO-22-105133.
Page 18 GAO-23-104291 Nursing Home COVID-19 Outbreaks
homes located in counties experiencing low community spread compared
to homes located in counties experiencing high community spread.
The number of ongoing outbreaks in a given week throughout the period
of review was typically much lower in nursing homes located in counties
experiencing low community spread compared to nursing homes located
in counties experiencing high community spread. These results could
indicate that, during times of higher levels of community spread, staff
have a greater likelihood of being exposed to the virus in the community
and then bringing it into the nursing home. 38 As we previously noted, the
first week of most outbreaks (about 75 percent) coincided with a reported
staff case rather than a reported resident case. The importance of
community spread in determining the probability of an outbreak ending is
consistent with what we found in our literature review, with many studies
finding that community spread was associated with the presence and
severity of COVID-19 within a nursing home. 39
Nursing Home Size The size of a nursing home, as measured by the number of Medicare- or
Medicaid-certified beds, was also strongly associated with the duration of
a COVID-19 outbreak in a nursing home throughout our period of review.
Specifically, we found that small (fewer than 50 beds) and medium (50 to
99 beds) nursing homes had a higher probability of their outbreaks ending
during a given week—meaning their outbreaks were shorter—when
compared to large (100 to 199 beds) and very large (200 or more beds)
nursing homes.
38We recognize that staff may live or visit counties outside of the one where the nursing
home is located, but there are no data available to determine this. In addition, visitors,
when allowed, could also bring COVID-19 into the nursing home, but there are no data
available on nursing home visitors.
39For example, one study found that the prevalence of COVID-19 in the community was
the strongest predictor of COVID-19 cases and deaths in nursing homes. See Rebecca J.
Gorges and R. Tamara Konetzka, “Staffing Levels and COVID-19 Cases and Outbreaks in
U.S. Nursing Homes,” Journal of the American Geriatrics Society, vol. 68, no. 11 (2020):
2462-2466. Another study found that county-level COVID-19 rates and per-capita income
were the most significant predictors of COVID-19 outbreaks within nursing homes. See
Margaret M. Sugg et al., “Mapping community-level determinants of COVID-19
transmission in nursing homes: A multi-scale approach,” Science of the Total
Environment, vol. 752 (2021). And a third study identified a nursing home’s county
infection rate as one of the strongest predictors of COVID-19 infection. See Christopher
L.F. Sun et al., “Predicting Coronavirus Disease 2019 Infection Risk and Related Risk
Drivers in Nursing Homes: A Machine Learning Approach,” JAMDA, vol. 21, no. 11 (2020):
1533-1538.
Page 19 GAO-23-104291 Nursing Home COVID-19 Outbreaks
The association of nursing home size with outbreak duration remained
fairly consistent both before and after the introduction of vaccines.
Specifically, in the period of time prior to the introduction of vaccines, our
model estimated that the duration of a COVID-19 outbreak in a small
nursing home was just over 2.5 days shorter than large and very large
homes and, after the introduction of vaccines, the association decreased
slightly to just under an estimated 2.5 days shorter.
These findings may be a result of smaller homes simply having fewer
residents and staff who can get infected by the virus, so the outbreaks do
not continue for as long as in larger nursing homes. However, it could
also be affected by other differences between smaller and larger nursing
homes, such as fewer staff and visitors entering smaller nursing homes,
which could decrease the likelihood of continued transmission of the
virus. The importance of nursing home size being associated with the
probability of an outbreak ending is consistent with what we found in our
literature review, with multiple studies finding that larger homes were
associated with the presence of COVID-19 or higher numbers of COVID-
19 cases. 40
Staff Shortages Nursing home-reported staff shortages were also strongly associated with
the duration of COVID-19 outbreaks prior to the introduction of vaccines.
We found that nursing homes that reported staff shortages had a lower
probability of their outbreaks ending during a given week—meaning their
40For example, one study found that, as of May 2020, larger nursing home size was
significantly associated with both an increased probability of having a COVID-19 case and
having a larger outbreak. See Hannah R. Abrams et al., “Characteristics of U.S. Nursing
Homes with COVID-19 Cases,” Journal of the American Geriatrics Society, vol. 68, no. 8
(2020): 1653-1656. Another study found that, from March to August 2020, large nursing
homes (150 or more beds) reported 2.6 fewer staff COVID-19 cases per 100 beds
compared with small nursing homes (less than 50 beds). See Kira L. Ryskina et al.,
“Characteristics of Nursing Homes by COVID-19 Cases among Staff: March to August
2020,” JAMDA, vol. 22, no. 5 (2021): 960-965.
Page 20 GAO-23-104291 Nursing Home COVID-19 Outbreaks
outbreaks were longer—when compared to homes that did not report staff
shortages. 41
Prior to the introduction of vaccines, the magnitude of the association of
reported staff shortages on outbreak duration was substantial, but the
association was minimal after the introduction of vaccines. Specifically,
our model estimated that, prior to the introduction of vaccines, the
duration of an outbreak at a nursing home reporting staff shortages was
over a day longer than homes that did not report shortages. After the
introduction of vaccines, the magnitude of the association of nursing
home-reported staff shortages on outbreak duration decreased to less
than half a day.
These findings may be due to the associated effects of staff shortages.
For instance, nurse staff may need to cover more residents and rooms or
temporary staff from other homes or health care settings may need to fill
the vacancies, both of which could inadvertently cause the virus to be
brought into the home or spread throughout more of the home. As
previously noted, nursing homes, state survey agencies, and state long-
term care ombudsmen we interviewed described staff shortages and
associated challenges, such as adhering to CDC-recommended infection
prevention and control practices, as concerns during the pandemic.
We also analyzed variables that directly measure actual nursing home
staff levels, rather than perceived shortages reported by nursing home
officials, and found less clear results. For instance, prior to the
introduction of vaccines, the greater the percentage of new employees,
the longer an outbreak lasted, but, while statistically significant, the
magnitude of the association was small. After the introduction of
vaccines, we found the inverse association—the greater the percentage
of new employees, the shorter an outbreak lasted—and, again, the
magnitude of the association was statistically significant but small. In
addition, we found that the associations of other staffing variables, such
41A few studies analyzed the association of various factors with the likelihood of nursing
homes reporting staff shortages. For example, two studies found that in mid-2020 reported
shortages were greater in nursing homes with COVID-19 cases among residents and
staff, as well as in homes with lower quality scores and those serving more Medicaid
beneficiaries. See Diane M. Gibson and Jessica Greene, “State Actions and Shortages of
Personal Protective Equipment and Staff in U.S. Nursing Homes,” Journal of the American
Geriatrics Society, vol. 68, no. 12 (2020): 2721-2726; and Brian E. McGarry et al., “Severe
Staffing and Personal Protective Equipment Shortages Faced By Nursing Homes During
the COVID-19 Pandemic,” Health Affairs, vol. 39, no. 10 (2020): 1812-1821.
Page 21 GAO-23-104291 Nursing Home COVID-19 Outbreaks
as average hours per employee per week and number of employees per
resident, on outbreak duration were not always statistically significant.
Our literature review identified one study that analyzed a measure of
outbreaks over time and it found an inverse association with nurse
staffing ratings. 42 However, our nuanced results are consistent with other
studies in our literature review. For example, one study found that,
whereas higher staffing levels increase the probability of having at least
one COVID-19 case, higher nurse aide and total nursing hours were
associated with a lower probability of a larger outbreak and fewer
deaths. 43 Another study found that the average number of unique staff
members working in a facility on a given day was strongly associated with
COVID-19 outcomes in nursing homes, but that, after staff size was
controlled for, there was no clear relationship between COVID-19
outcomes and more traditional measures of staffing. 44
Ownership The type of nursing home ownership—whether the home is for-profit,
non-profit, or government-owned—was also strongly associated with
COVID-19 outbreak duration at times during our period of review. We
found that nursing homes designated as non-profit or government-owned
had a lower probability of their outbreaks ending during a given week—
meaning their outbreaks were longer—when compared to homes
designated as for-profit. For instance, our model estimated that the
duration of an outbreak in a government-owned nursing home was about
1.5 days longer than in for-profit homes after the introduction of vaccines.
The studies in our literature review that found an association between
COVID-19 outcomes in nursing homes and nursing home ownership
generally found that non-profit homes did better than for-profit homes on
the presence or severity of cases and the number of deaths. For
42The study examined the total number of weeks a nursing home was in the top decile of
the national distribution of weekly confirmed resident COVID-19 incidence and found that
nursing homes with lower nurse staffing ratings had 18 to 22 percent more weeks with
high COVID-19 incidence than homes with the highest staffing rating. See Christianna S.
Williams et al., “The Association of Nursing Home Quality Ratings and Spread of COVID-
19,” Journal of the American Geriatrics Society, vol. 69, no. 8 (2021): 2070-2078.
43See Rebecca J. Gorges and R. Tamara Konetzka, “Staffing Levels and COVID-19
Cases and Outbreaks in U.S. Nursing Homes,” Journal of the American Geriatrics Society,
vol. 68, no. 11 (2020): 2462-2466.
44See Brian E. McGarry et al., “Larger Nursing Home Staff Size Linked To Higher Number
Of COVID-19 Case In 2020,” Health Affairs, vol. 40, no. 8 (2021): 1261-1269.
Page 22 GAO-23-104291 Nursing Home COVID-19 Outbreaks
example, one study of California nursing homes found that as of May 1,
2020, the size of COVID-19 outbreaks was 12.7 times larger in for-profit
nursing homes than in their non-profit counterparts. 45 One study,
however, found an inverse association—that, between March and August
2020, for-profit nursing homes were associated with fewer staff COVID-19
cases when compared to non-profit homes. 46 While our results were
unexpected based on the general findings of the studies we reviewed, we
analyzed a longer period of time and measured a different outcome—
outbreak duration rather than the presence or severity of COVID-19
cases or deaths.
We provided a draft of this report to HHS for review and comment. HHS
Agency Comments provided technical comments, which we addressed as appropriate. HHS
comments noted some considerations for potential further analysis and
potential methodological limitations, and our report describes limitations.
HHS comments also highlighted how critical infection prevention and
control practices within nursing homes are to reducing the risk of COVID-
19 transmission, a point that we and others have also made.
We are sending copies of this report to the appropriate congressional
committees, the Secretary of HHS, and other interested parties. In
addition, the report is available at no charge on the GAO website at
http://www.gao.gov.
If you or your staff have any questions about this report, please contact
me at (202) 512-7114 or at dickenj@gao.gov. Contact points for our
Offices of Congressional Relations and Public Affairs may be found on
45See Ram Gopal, Xu Han, and Niam Yaraghi, “Compress the Curve: A Cross-Sectional
Study of Variations in COVID-19 Infections across California Nursing Homes,” BMJ Open,
vol. 11 (2021): e042804.
46See Kira L. Ryskina et al., “Characteristics of Nursing Homes by COVID-19 Cases
among Staff: March to August 2020,” JAMDA, vol. 22, no. 5 (2021): 960-965.
Page 23 GAO-23-104291 Nursing Home COVID-19 Outbreaks
the last page of this report. GAO staff who made key contributions to this
report are listed in Appendix IV.
John E. Dicken
Director, Health Care
Page 24 GAO-23-104291 Nursing Home COVID-19 Outbreaks
List of Addressees
The Honorable Patrick Leahy
Chairman
The Honorable Richard Shelby
Vice Chairman
Committee on Appropriations
United States Senate
The Honorable Ron Wyden
Chairman
The Honorable Mike Crapo
Ranking Member
Committee on Finance
United States Senate
The Honorable Patty Murray
Chairman
The Honorable Richard Burr
Ranking Member
Committee on Health, Education, Labor, and Pensions
United States Senate
The Honorable Gary C. Peters
Chairman
The Honorable Rob Portman
Ranking Member
Committee on Homeland Security and Governmental Affairs
United States Senate
The Honorable Rosa L. DeLauro
Chair
The Honorable Kay Granger
Ranking Member
Committee on Appropriations
House of Representatives
The Honorable Frank Pallone, Jr.
Chairman
The Honorable Cathy McMorris Rodgers
Republican Leader
Committee on Energy and Commerce
House of Representatives
Page 25 GAO-23-104291 Nursing Home COVID-19 Outbreaks
The Honorable Bennie G. Thompson
Chairman
The Honorable John Katko
Ranking Member
Committee on Homeland Security
House of Representatives
The Honorable Carolyn B. Maloney
Chairwoman
The Honorable James Comer
Ranking Member
Committee on Oversight and Reform
House of Representatives
The Honorable Richard E. Neal
Chair
The Honorable Kevin Brady
Republican Leader
Committee on Ways and Means
House of Representatives
The Honorable Michael F. Bennet
United States Senate
Page 26 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Appendix I: Literature Review—Methodology
Appendix I: Literature Review—Methodology
Description and Results
Description and Results
To understand what is known about the factors associated with COVID-
19 outbreaks in nursing homes, we conducted a systematic literature
review from January 1, 2020, through October 1, 2021.
Methodology We conducted a systematic literature review of studies using data to
examine the association of any factor with COVID-19 outbreaks in U.S.
nursing homes. We searched across five research platforms (EBSCO,
Scopus, SSRN, ProQuest, and Dialog’s 96 databases), focusing on those
databases that covered health and medical issues. 1 We conducted
additional targeted internet searches using the Harvard Kennedy School
Think Tank Search to capture additional studies. After reviewing the initial
returns from our search, we expanded and updated our search strategy to
increase the relevance and breadth of our returns. We ultimately
searched the original five research platforms plus PubMed and Google
Scholar, covering the time period of January 1, 2020, through October 1,
2021.
We identified a total of 49 studies that met our selection criteria, and we
conducted a high-level quality review to assess the reliability and
methodological soundness of those studies. We considered studies to be
sufficiently reliable if they had been peer-reviewed and published in an
academic journal, if they were included in a federal or state government
publication with similarly rigorous review processes, or if they were
prepared by or on behalf of a federal or state government entity. 2 As a
result of this process, we determined that 43 studies were sufficiently
reliable for our use. We excluded three of those studies because, upon
further review, they did not meet our original criteria—to select studies
1For instance, Embase, EMCare, HSELINE: Health and Safety, Global Health, Medline,
and the New England Journal of Medicine.
2If studies did not meet this criterion, we considered them to be sufficiently reliable if they
met all of the following three criteria: (1) the study was marked as provisional (e.g., pre-
publication status) for a peer-reviewed journal or government publication with a similarly
rigorous review process; (2) the study was transparent about its data sources and
methods, those sources and methods appeared appropriate for the researchable
questions, and the study discussed possible limitations; and (3) the study reported all
relevant results, and conclusions and recommendations were adequately supported by
those results.
Due to the rapidly evolving nature of the COVID-19 pandemic, we recognized that recent
studies may not have had time to complete the full peer review process. We chose to
include studies still pending peer review because they could cover emerging issues or use
more timely and complete data than studies from earlier in the pandemic. Four of the final
40 studies were marked as provisional at the time of selection. All four of those studies
were published by the time of our reporting—three in peer reviewed journals.
Page 27 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Appendix I: Literature Review—Methodology
Description and Results
that examined the association of one or more factors with COVID-19
outbreaks in U.S. nursing homes. This resulted in a total of 40 selected
studies.
To review these selected studies, we applied a framework of three
analytical categories describing the type of analysis conducted for each
study: presence, severity, and duration of COVID-19 outbreaks in nursing
homes. Some studies conducted more than one of these types of
analyses.
• Presence of COVID-19 outbreaks. Studies investigating the
presence of COVID-19 outbreaks assessed the association of state,
county, nursing home, or resident factors with a categorical or binary
outcome relating to whether or not COVID-19 affected a nursing
home. For example, some studies compared nursing homes that had
at least one COVID-19 case or death with homes that did not have
any COVID-19 cases or deaths. Eighteen studies were categorized as
assessing presence.
• Severity of COVID-19 outbreaks. Studies investigating the severity
of COVID-19 outbreaks assessed the association of state, county,
nursing home, or resident factors with a continuous outcome, such as
by comparing the cumulative number of COVID-19 cases or deaths in
each nursing home. Thirty-three studies were categorized as
assessing severity.
• Duration of COVID-19 outbreaks. Studies investigating the duration
of COVID-19 outbreaks assessed the association of state, county,
nursing home, or resident factors with the amount of time that each
nursing home was experiencing COVID-19 outbreaks. One study was
categorized as assessing duration.
Results Table 2 lists the 40 studies we reviewed.
Page 28 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Appendix I: Literature Review—Methodology
Description and Results
Table 2: Studies Included in GAO’s Review That Focused on Factors Associated with the Presence, Severity, or Duration of
COVID-19 Outbreaks in Nursing Homes, January 1, 2020, through October 1, 2021
Methodological focus of study
Factors Factors
associated Factors associated
with the associated with the
presence with the duration
of COVID- severity of of COVID-
Authors 19 COVID-19 19
outbreaks outbreaks outbreaks
Journal in nursing in nursing in nursing
Title Publication date homes homes homes Summary of key findings
1 “Is There a Link He, Mengying, – In California from April 23, 2020, through
between Nursing Yumeng Li, and June 2, 2020, nursing homes with a five-star
Home Reported Fang Fang rating and a higher percentage of White
Quality and COVID- JAMDA residents were less likely to have COVID-19
19 Cases? Evidence cases and deaths compared to other nursing
from California July 2020 homes in the state, after adjusting for nursing
Skilled Nursing home size, ownership, and years of
Facilities” operation.
2 “Nursing Home Unruh, Mark – – As of mid-April 2020, nursing homes in
Characteristics Aaron et al. Connecticut, New Jersey, and New York with
Associated with JAMDA the highest percentages of Medicaid patients
COVID-19 Deaths in had an 8.6 percentage point greater
Connecticut, New July 2020 probability of six or more COVID-19 deaths
Jersey, and New than nursing homes with the lowest
York” percentages of these patients in the same
three states. Other characteristics associated
with COVID-19 deaths included having
patients with higher activities of daily living
scores, more total beds, higher occupancy
rates, and being for-profit.
3 “Nurse Staffing and Harrington, – – Between March and May 2020, California
Coronavirus Charlene et al. nursing homes with total registered nurse
Infections in Policy, Politics, & staffing levels under 0.75 hours per resident
California Nursing Nursing Practice day had a two times greater probability of
Homes” having COVID-19 resident infections.
July 7, 2020 Nursing homes with lower Medicare five-star
ratings on total nurse and registered nurse
staffing levels (adjusted for acuity), higher
total health deficiencies, and more beds had
a higher probability of having COVID-19
residents.
Page 29 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Appendix I: Literature Review—Methodology
Description and Results
Methodological focus of study
Factors Factors
associated Factors associated
with the associated with the
presence with the duration
of COVID- severity of of COVID-
Authors 19 COVID-19 19
outbreaks outbreaks outbreaks
Journal in nursing in nursing in nursing
Title Publication date homes homes homes Summary of key findings
4 “Characteristics and Chatterjee, Paula – – In nursing homes from the District of
Quality of US et al. Columbia and 23 states reporting COVID-19
Nursing Homes JAMA Network cases from April 22, 2020, through April, 29,
Reporting Cases of Open 2020, homes that reported COVID-19 cases
Coronavirus Disease had similar mean scores as those that did
2019 (COVID-19)” July 29, 2020 not on overall five-star ratings, as well as star
ratings on deficiencies and staffing.
Compared to homes that did not report
cases, those that did had higher rates of
health deficiencies, emergency
preparedness deficiencies, reported
incidents, and substantiated complaints.
Rates of COVID-19 were nearly twice as
high in counties where homes reported
COVID-19 cases than in those without
reported cases.
5 “Characteristics of Abrams, Hannah – In all nursing homes from 30 states with
U.S. Nursing Homes R. et al. facility-level COVID-19 data as of May 11,
with COVID-19 Journal of the 2020, larger facility size, urban location,
Cases” American greater percentage of African American
Geriatrics Society residents, non-chain status, and state were
significantly related to the increased
August 2020 probability of having a COVID-19 case. Five-
star rating, prior infection violation, Medicaid
dependency, and ownership were not
significantly related. The size of the outbreak
was significantly associated with facility size,
for-profit status, and state but not with other
studied characteristics.
6 “Mapping Sugg, Margaret – As of June 30, 2020, county-level COVID-19
Community-Level M. et al. rates, per-capita income, average household
Determinants of Science of the size, population density, unemployment, and
COVID-19 Total a higher percentage of African Americans, in
Transmission in Environment combination with nursing home staffing
Nursing homes: A levels (licensed practical nurses and total
Multi-Scale August 25, 2020 staff) and number of fines issued in 2020
Approach” were associated with an increased risk of
COVID-19 cases in nursing homes. Lower
overall quality rating and staffing rating
increased the risk of COVID-19 cases, but
the results were not statistically significant.
Page 30 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Appendix I: Literature Review—Methodology
Description and Results
Methodological focus of study
Factors Factors
associated Factors associated
with the associated with the
presence with the duration
of COVID- severity of of COVID-
Authors 19 COVID-19 19
outbreaks outbreaks outbreaks
Journal in nursing in nursing in nursing
Title Publication date homes homes homes Summary of key findings
7 “Staffing Levels and Gorges, Rebecca – As of June 14, 2020, higher registered
COVID-19 Cases J. and R. Tamara nurse-hours were associated with a higher
and Outbreaks in Konetzka probability of a nursing home experiencing
U.S. Nursing Homes” Journal of the any COVID-19 cases. However, among
American nursing homes with at least one case, higher
Geriatrics Society nurse aide hours and total nursing hours
were associated with a lower probability of
August 28, 2020 experiencing an outbreak and with fewer
deaths. The strongest predictor of cases and
outbreaks in nursing homes was per capita
cases in the county.
8 “COVID-19 Infections Li, Yue et al. – Among Connecticut nursing homes with at
and Deaths among Journal of the least one confirmed COVID-19 case, as of
Connecticut Nursing American April 16, 2020, every 20-minute increase in
Home Residents: Geriatrics Society registered nurse staffing (per resident day)
Facility Correlates” was associated with 22 percent fewer
September 2020 confirmed cases. Compared with one- to
three-star facilities, four- or five-star facilities
had 13 percent fewer confirmed cases, and
facilities with a high concentration of
Medicaid residents or racial/ethnic minority
residents had 16 percent and 15 percent
more confirmed cases, respectively, than
their counterparts. Among nursing homes
with at least one COVID-19 death, every 20-
minute increase in registered nurse staffing
significantly predicted 26 percent fewer
COVID-19 deaths. Other characteristics did
not show statistically significant associations
with COVID-19 deaths.
9 “Association of Figueroa, Jose F. – – For 4,254 nursing homes across eight states
Nursing Home et al. from January 1, 2020, through June 30,
Ratings on Health JAMA 2020, nursing homes with high ratings on
Inspections, Quality nurse staffing had fewer COVID-19 cases
of Care, and Nurse September 15, than nursing homes with low ratings on
Staffing with COVID- 2020 nurse staffing. There was no significant
19 Cases” association between high- and low-
performing nursing homes in the health
inspections or quality measures domains
with COVID-19 cases.
Page 31 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Appendix I: Literature Review—Methodology
Description and Results
Methodological focus of study
Factors Factors
associated Factors associated
with the associated with the
presence with the duration
of COVID- severity of of COVID-
Authors 19 COVID-19 19
outbreaks outbreaks outbreaks
Journal in nursing in nursing in nursing
Title Publication date homes homes homes Summary of key findings
10 “Association Bui, David P. et – – During March through June 2020, 14 (11
Between CMS al. percent) of 123 West Virginia nursing homes
Quality Ratings and Morbidity and experienced COVID-19 outbreaks.
COVID-19 Outbreaks Mortality Weekly Compared with one-star–rated (lowest
in Nursing Homes— Report rating) nursing homes, the odds of a COVID-
West Virginia, March 19 outbreak were 87 percent lower among
17-June 11, 2020” September 18, two- to three-star–rated facilities and 94
2020 percent lower among four- to five-star–rated
facilities.
11 “A Study of the Rowan, Patricia – – Through July 22, 2020, nursing homes in
COVID-19 Outbreak et al. Connecticut with a high staffing rating had
and Response in Connecticut significantly fewer COVID-19 cases and
Connecticut Long- Department of deaths per licensed bed than nursing homes
Term Care Facilities” Public Health with a lower staffing rating. Nursing homes in
Connecticut located in communities with a
September 30, greater incidence of COVID-19 were more
2020 likely to experience higher numbers of cases
and deaths. Nursing homes in Connecticut
with a larger share of residents that received
dialysis and cancer treatments, which tend to
be delivered off site, had more cases per
licensed bed. The total number of residents
and the share of the licensed beds that were
filled also significantly predicted greater
COVID-19 spread.
12 “Variation in SARS- White, Elizabeth – As of late April and early May 2020, a
CoV-2 Prevalence in M. et al. difference in county prevalence of 1,000
U.S. Skilled Nursing Journal of the COVID-19 cases per 100,000 (1 percent)
Facilities” American was associated with a 33.6 percentage point
Geriatrics Society difference in the probability of an outbreak
for Genesis HealthCare-owned nursing
October 2020 homes (sample of 341 nursing homes in 25
states) and non-Genesis nursing homes
(sample of 3,016 nursing homes in 12 states)
combined, and a difference of 12.5 cases per
facility for Genesis nursing homes. A 10-bed
difference in facility size was associated with
a 0.9 percentage point difference in the
probability of outbreak. No consistent
relationship was found between nursing
home Five-Star ratings or past infection
control deficiency citations and probability or
severity of outbreak.
Page 32 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Appendix I: Literature Review—Methodology
Description and Results
Methodological focus of study
Factors Factors
associated Factors associated
with the associated with the
presence with the duration
of COVID- severity of of COVID-
Authors 19 COVID-19 19
outbreaks outbreaks outbreaks
Journal in nursing in nursing in nursing
Title Publication date homes homes homes Summary of key findings
13 “Prevalence of Bowblis, John – Through mid-June 2020, just under one-third
COVID-19 in Ohio and Robert of nursing homes in Ohio had at least one
Nursing Homes: Applebaum resident with COVID-19, with over 82
What’s Quality Got to Journal of Aging percent of all cases in the state coming from
Do with It?” & Social Policy 37 percent of nursing homes. Overall
findings on the association between facility
October 11, 2020 quality and the prevalence of COVID-19
showed that having any resident case of the
virus or even having a high caseload of
residents with the virus is not more likely in
nursing homes with lower quality ratings.
14 “Comparative Braun, Robert – – From May 17, 2020, through July 2, 2020,
Performance of Tyler et al. there were no statistically significant
Private Equity- JAMA Network differences in staffing levels, COVID-19
Owned US Nursing Open cases or deaths, or deaths by any cause
Homes During the between private equity-owned nursing
COVID-19 October 28, 2020 homes and for-profit, nonprofit, and
Pandemic” government-owned facilities. For-profit,
nonprofit, and government-owned nursing
homes were more likely to have at least a 1-
week supply of N95 masks and a 1-week
supply of medical gowns than private equity-
owned nursing homes.
15 “Mortality Rates from Dean, Adam, – – In 355 New York nursing homes with
COVID-19 Are Lower Atheendar available data from March 1, 2020, through
in Unionized Nursing Venkataramani, May 31, 2020, health care worker unions
Homes” and Simeon were associated with a 1.29 percentage
Kimmel point decrease in the COVID-19 mortality
Health Affairs when compared with nursing homes without
these unions. Unions were also associated
November 2020 with greater access to personal protective
equipment.
Page 33 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Appendix I: Literature Review—Methodology
Description and Results
Methodological focus of study
Factors Factors
associated Factors associated
with the associated with the
presence with the duration
of COVID- severity of of COVID-
Authors 19 COVID-19 19
outbreaks outbreaks outbreaks
Journal in nursing in nursing in nursing
Title Publication date homes homes homes Summary of key findings
16 “Racial and Ethnic Li, Yue et al. – For COVID-19 data reported the week of
Disparities in COVID- Journal of the May 25, 2020, nursing homes with a high-
19 Infections and American proportion of racial and ethnic minorities
Deaths Across U.S. Geriatrics Society were 76 percent more likely to have at least
Nursing Homes” one new resident case compared to nursing
November 2020 homes with a low-proportion of racial and
ethnic minorities. In addition, similar across-
facility disparities were found for the weekly
count of new COVID-19 deaths among
residents and in the weekly count of new
COVID-19 confirmed cases among staff. No
substantial disparities in self-reported
shortages of staff or personal protective
equipment were found.
17 “Stemming the Tide Lipsitz, Lewis A. – – In response to the COVID-19 pandemic, a
of COVID-19 et al. statewide effort in Massachusetts offered all
Infections in Journal of the 360 nursing homes in the state weekly
Massachusetts American webinars and answers to questions
Nursing Homes” Geriatrics Society regarding infection control procedures. The
effort also targeted a subset of 123 nursing
November 2020 homes with previous infection control
deficiencies for intervention, which included
on-site and virtual consultations. During the
9-week intervention period from May 10,
2020, through July 5, 2020, both resident
and staff COVID-19 infection rates started
higher in the targeted subset of nursing
homes, then rapidly declined to the same low
level as other nursing homes. Adherence to
infection control processes, especially proper
wearing of personal protective equipment
and cohorting, was significantly associated
with declines in weekly infection and
mortality rates.
Page 34 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Appendix I: Literature Review—Methodology
Description and Results
Methodological focus of study
Factors Factors
associated Factors associated
with the associated with the
presence with the duration
of COVID- severity of of COVID-
Authors 19 COVID-19 19
outbreaks outbreaks outbreaks
Journal in nursing in nursing in nursing
Title Publication date homes homes homes Summary of key findings
18 “Predicting Sun, Christopher – – As of April 20, 2020, the strongest predictors
Coronavirus Disease L.F. et al. of COVID-19 cases in 1,146 nursing homes
2019 Infection Risk JAMDA in Massachusetts, Georgia, and New Jersey
and Related Risk were the nursing home’s county’s infection
Drivers in Nursing November 2020 rate and the number of separate units in the
Homes: A Machine nursing home; other predictors included the
Learning Approach” county’s population density, the nursing
home’s historical health deficiency citations,
and the nursing home’s resident density (in
persons per 1,000 square feet). In addition,
the nursing home’s historical percentage of
non-Hispanic White residents was identified
as a protective factor.
19 “The Role of the LeRose, Jennifer – – In May 2020, Michigan nursing homes
Social Vulnerability J. et al. located in the most vulnerable areas of the
Index in Personal Infection Control state, which was determined using the
Protective Equipment & Hospital Centers for Disease Control and
Shortages, Number Epidemiology Prevention’s social vulnerability index, were
of Cases, and 2.3 times more likely to experience personal
Associated Mortality November 13, protective equipment shortages than homes
During the 2020 located in the least vulnerable areas. In
Coronavirus Disease addition, nursing homes located in the most
2019 (COVID-19) vulnerable areas had 1.6 times the number
Pandemic in of COVID-19 cases and 1.9 times the
Michigan Skilled mortality rate compared to nursing homes
Nursing Facilities” located in the least vulnerable areas.
20 “Nursing Home Staff Chen, M. Keith, – – During the 11-week period from March 13,
Networks and Judith A. 2020, to May 31, 2020, 5.1 percent of 50
COVID-19” Chevalier, and million smartphone users who visited a
Elisa F. Long nursing home for at least one hour also
Proceedings of visited another nursing home and nursing
the National homes, on average, shared connections with
Academy of an estimated 7.1 other homes. Comparing
Sciences demographically and geographically situated
nursing homes of similar quality suggested
December 28, that 49 percent of nursing home COVID-19
2020 cases were attributable to shared staff
transmitting the virus across multiple nursing
homes.
Page 35 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Appendix I: Literature Review—Methodology
Description and Results
Methodological focus of study
Factors Factors
associated Factors associated
with the associated with the
presence with the duration
of COVID- severity of of COVID-
Authors 19 COVID-19 19
outbreaks outbreaks outbreaks
Journal in nursing in nursing in nursing
Title Publication date homes homes homes Summary of key findings
21 “Compress the Gopal, Ram, Xu – For California nursing homes as of May 1,
Curve: A Cross- Han, and Niam 2020, the size of COVID-19 outbreaks
Sectional Study of Yaraghi among residents in for-profit nursing homes
Variations in COVID- BMJ Open was 12.7 times larger than their non-profit
19 Infections across counterparts. Higher star ratings for health
California Nursing January 6, 2021 inspections were associated with a lower
Homes” number of infections among both staff and
residents and higher star ratings for staffing
were associated with a lower number of
infections among residents. Conversely,
higher star ratings for quality measures were
associated with a higher number of infections
among staff and residents.
22 “Factors Associated Gorges, Rebecca – – As of September 13, 2020, nursing homes in
with Racial J. and R. Tamara which more than 40 percent of residents
Differences in Konetzka were non-White experienced case and death
Deaths Among JAMA Network counts that were 3.3-fold higher than those
Nursing Home Open of nursing homes with low proportions of
Residents with non-White residents. These differences in
COVID-19 Infection February 10, resident deaths by race were associated with
in the U.S.” 2021 nursing home size and community-level
outbreak severity, but not with aggregate
health status of residents or other nursing
home characteristics.
23 “Nontraditional Small Zimmerman, – – Green House and other small nursing home
House Nursing Sheryl et al. models are considered nontraditional due to
Homes Have Fewer JAMDA their size (10-12 beds), universal caregivers,
COVID-19 Cases and other home-like features. From January
and Deaths” March 2021 20, 2020, through July 31, 2020, 219 Green
House and other similar small nursing homes
located in 20 states had lower COVID-19
incidence and mortality rates compared to
392 geographically proximate traditional
nursing homes, especially among the higher
and extreme values.
24 “Predicting COVID- Poltavskiy, – – As of May 31, 2020, publicly reported
19 at Skilled Nursing Eduard et al. nursing home metrics were associated with
Facilities in BMJ Open COVID-19 presence in California nursing
California: Do the Quality homes. Specifically, nursing home health
Stars Align?” inspection star ratings, emergency room
March 3, 2021 visits, and short-stay hospitalizations were
significantly associated with nursing homes
reporting COVID-19 cases.
Page 36 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Appendix I: Literature Review—Methodology
Description and Results
Methodological focus of study
Factors Factors
associated Factors associated
with the associated with the
presence with the duration
of COVID- severity of of COVID-
Authors 19 COVID-19 19
outbreaks outbreaks outbreaks
Journal in nursing in nursing in nursing
Title Publication date homes homes homes Summary of key findings
25 “High-Minority Weech- – – As of October 25, 2020, nursing homes with
Nursing Homes Maldonado, a high percentage of minority residents
Disproportionately Robert et al. reported 6.5 COVID-19 deaths as compared
Affected by COVID- Frontiers in to 2.6 deaths for nursing homes that had no
19 Deaths” Public Health racial or ethnic minorities. After controlling for
interstate differences, facility-level resident
March 22, 2021 characteristics, resource availability, and
organizational characteristics, nursing homes
with a high percentage of minority residents
had 61 percent more COVID-19 deaths as
compared to nursing homes with no
minorities.
26 “Assessment of Travers, Jasmine – – From January 20, 2020, through July 19,
Coronavirus Disease L. et al. 2020, nursing homes with any Black
2019 Infection and JAMDA residents showed significantly more COVID-
Mortality Rates 19 infections and deaths than nursing homes
among Nursing April 2021 with no Black residents. There were 13.6
Homes with Different percentage points more infections and 3.5
Proportions of Black percentage points more deaths in nursing
Residents” homes with 50 percent or greater Black
residents than in nursing homes with no
Black residents. Although facility
characteristics explained some of the
differences found in multivariable analyses,
county-level factors and rurality explained
more of the differences.
27 “Short-Term Impact Mor, Vincent et – – Two hundred and eighty nursing homes in 21
of Nursing Home al. states owned and operated by Genesis
SARS-CoV-2 Journal of the Healthcare were divided into two COVID-19
Vaccinations on New American vaccination groups based on the date the
Infections, Geriatrics Society nursing home had its initial vaccine clinic—
Hospitalizations, and early vaccination group (December 18, 2020,
Deaths” April 16, 2021 through January 2, 2021) and late
vaccination group (January 3, 2021, through
January 18, 2021). After 1 week, early
vaccinated nursing homes had a predicted
2.5 fewer incident COVID-19 infections per
100 at-risk residents per week compared
with what would have been expected based
on the experience of the late vaccinated
nursing homes. The rates remained
significantly lower for several weeks.
Page 37 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Appendix I: Literature Review—Methodology
Description and Results
Methodological focus of study
Factors Factors
associated Factors associated
with the associated with the
presence with the duration
of COVID- severity of of COVID-
Authors 19 COVID-19 19
outbreaks outbreaks outbreaks
Journal in nursing in nursing in nursing
Title Publication date homes homes homes Summary of key findings
28 “SARS-CoV-2 in Ehrlich, Hanna Y. – – The Connecticut Department of Public
Nursing Homes after et al. Health began point prevalence survey testing
3 Months of Serial, Emerging in nursing homes in early May 2020, and this
Faciltywide Point Infectious type of testing was formally recommended
Prevalence Testing, Diseases on May 11, 2020, and mandated weekly in
Connecticut, USA” staff effective June 14, 2020. The state
May 2021 initially prioritized a subset of 34 Connecticut
nursing homes to receive test kits and these
nursing homes performed their first round of
testing on or before May 20, 2020. After
adjusting for community incidence of COVID-
19, the implementation of serial point
prevalence survey testing in the 34 nursing
homes was associated with a significant
decrease in nursing home COVID-19
incidence rates compared to the pre-
implementation period.
29 “Shifting US Patterns Kumar, Amit et – – During the first 6 weeks of a 30-week study
of COVID-19 al. period, which started on June 1, 2020,
Mortality by Race JAMDA nursing homes with a higher proportion of
and Ethnicity from Black residents reported more COVID-19
June-December May 2021 deaths per 1,000 followed by nursing homes
2020” with a higher proportion of Hispanic
residents. Between 7 and 12 weeks, nursing
homes with a higher proportion of Hispanic
residents reported more deaths per 1,000,
followed by nursing homes with a higher
proportion of Black residents. However, after
23 weeks (mid-November 2020), nursing
homes serving a higher proportion of White
residents reported more deaths per 1,000
than nursing homes serving a high
proportion of Black and Hispanic residents.
Page 38 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Appendix I: Literature Review—Methodology
Description and Results
Methodological focus of study
Factors Factors
associated Factors associated
with the associated with the
presence with the duration
of COVID- severity of of COVID-
Authors 19 COVID-19 19
outbreaks outbreaks outbreaks
Journal in nursing in nursing in nursing
Title Publication date homes homes homes Summary of key findings
30 “Characteristics of Ryskina, Kira L. – – From March to August 2020, after
Nursing Homes by et al. accounting for local COVID-19 prevalence,
COVID-19 Cases JAMDA nursing homes in the highest quartile of
Among Staff: March confirmed resident cases reported 18.9 more
to August 2020” May 2021 staff cases per 100 beds compared with
nursing homes that had no resident cases.
Large nursing homes (150 or more beds)
reported 2.6 fewer staff cases per 100 beds
compared with small nursing homes (<50
beds) and for-profit nursing homes reported
0.8 fewer staff cases per 100 beds compared
with nonprofit nursing homes. Higher
occupancy and more direct-care hours per
day were associated with more staff cases.
Estimates associated with resident
demographics, payer mix, or regional
socioeconomic characteristics were not
statistically significant.
31 “The BNT162b2 Domi, Marsida et – – Dividing the 2,501 nursing homes from the
Vaccine Is al. first 17 states to hold vaccine clinics into
Associated with Journal of the three groups—vaccine clinic ending the
Lower New COVID- American week of December 27, 2020 (cohort 1),
19 Cases in Nursing Geriatrics Society January 3, 2021 (cohort 2), or January 10,
Home Residents and 2021 (cohort 3)—found that resident and
Staff” May 6, 2021 staff cases trended downward in all three
cohorts following the vaccine clinic. Time
following the first clinic at 5 and 6 weeks was
consistently associated with fewer resident
cases, resident deaths, and staff cases.
Other factors associated with fewer resident
and staff cases included facilities with less
than 50 beds and high nurse staffing per
resident day. Higher Hispanic non-White
resident census was associated with fewer
resident cases and deaths.
Page 39 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Appendix I: Literature Review—Methodology
Description and Results
Methodological focus of study
Factors Factors
associated Factors associated
with the associated with the
presence with the duration
of COVID- severity of of COVID-
Authors 19 COVID-19 19
outbreaks outbreaks outbreaks
Journal in nursing in nursing in nursing
Title Publication date homes homes homes Summary of key findings
32 “Trends in COVID-19 Gilman, Matlin – – From May 25, 2020 to April 18, 2021, death
Death Rates by and Mary T. rates in nursing homes categorized into the
Racial Composition Bassett highest and lowest quintiles of White
of Nursing Homes” Journal of the residents varied over time. Specifically, high-
American White quintile nursing homes initially had a
Geriatrics Society lower death rate than low-White quintile
nursing homes. However, by late December
May 15, 2021 2020, high-White quintile nursing homes had
experienced 3 months of higher community
spread and its death rate had increased to
nearly three times that of low-White quintile
nursing homes. Through the first few months
of 2021, death rates declined substantially
for both groups of nursing homes, but,
overall, high-White quintile nursing homes
had more total deaths than low-White quintile
nursing homes despite having fewer beds
and higher star ratings.
33 “The Association of Williams, – As of January 10, 2021, higher nursing home
Nursing Home Christianna S. et quality ratings were associated with COVID-
Quality Ratings and al. 19 incidence and mortality, as well as with
Spread of COVID- Journal of the fewer high-incidence weeks, after controlling
19” American for local community COVID-19 spread and
Geriatrics Society relevant facility-level factors. For example,
nursing homes with a lower overall rating
May 31, 2021 (one- to three-stars) had about 10 percent
more weeks with high COVID-19 incidence
than the highest (five-star) rated homes, and
nursing homes with an overall star rating of
one- to three-stars also had about 13 to 16
percent higher incidence and 8 to 10 percent
higher cumulative COVID-19 mortality than
five-star homes. Higher county COVID-19
incidence was also strongly and consistently
associated with higher nursing home COVID-
19 incidence and mortality, as well as
number of high-incidence weeks.
Page 40 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Appendix I: Literature Review—Methodology
Description and Results
Methodological focus of study
Factors Factors
associated Factors associated
with the associated with the
presence with the duration
of COVID- severity of of COVID-
Authors 19 COVID-19 19
outbreaks outbreaks outbreaks
Journal in nursing in nursing in nursing
Title Publication date homes homes homes Summary of key findings
34 “Trends in Racial and Li, Yue et al. – – From April 13, 2020, through June 19, 2020,
Ethnic Disparities in Infection Control Connecticut nursing homes caring for
Coronavirus Disease & Hospital predominately racial and ethnic minority
2019 (COVID-19) Epidemiology residents tended to have higher COVID-19
Outcomes among incidence and fatality rates. Specifically,
Nursing Home June 16, 2021 compared to nursing homes with a low
Residents” proportion of racial and ethnic minority
residents, the adjusted COVID-19 incidence
rate ratios for the nursing homes with a high
proportion of racial and ethnic minority
residents were 1.18 in week 1 of the study
period and 1.54 in week 10, showing a 30
percent relative increase. Adjusted
disparities in COVID-19 fatalities similarly
increased over time.
35 “COVID-19 Cases Cai, Shubing, Di – – From June 7, 2020, through August 23,
and Death in Nursing Yan, and Orna 2020, the racial and ethnic composition of
Homes: The Role of Intrator nursing homes and their communities were
Racial and Ethnic JAMDA both associated with the likelihood of having
Composition of COVID-19 cases and death in nursing
Facilities and Their July 2021 homes. The racial and ethnic composition of
Communities” the community played an independent role in
the likelihood of COVID-19 cases and death
in nursing homes, even after accounting for
the COVID-19 infection rate in the
community. In addition, the relationship
between nursing home characteristics and
the probability of COVID-19 cases or death
varied with the racial and ethnic composition
of the community. For example, although an
hour increase in registered nurse hours per
resident per day was associated with 2.8
percentage point reduction in the probability
of COVID-19 cases in nursing homes located
in high-minority communities, such
relationship was smaller in low-minority
communities.
Page 41 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Appendix I: Literature Review—Methodology
Description and Results
Methodological focus of study
Factors Factors
associated Factors associated
with the associated with the
presence with the duration
of COVID- severity of of COVID-
Authors 19 COVID-19 19
outbreaks outbreaks outbreaks
Journal in nursing in nursing in nursing
Title Publication date homes homes homes Summary of key findings
36 “Larger Nursing McGarry, Brian – – From June 7, 2020, through September 27,
Home Staff Size E. et al. 2020, sample nursing homes (7,154 nursing
Linked to Higher Health Affairs homes without resident COVID-19 cases by
Number of COVID- the start of the study period) in the lowest
19 Cases in 2020” August 2021 quartile of staff size had 6.2 resident cases
and 0.9 deaths per 100 beds, compared with
11.9 resident cases and 2.1 deaths per 100
beds among nursing homes in the highest
quartile. Staff size, including staff members
not involved in resident care, was strongly
associated with COVID-19 outcomes, even
after nursing home size was accounted for.
Staffing quality measures, including direct
care staff-to-resident ratios and skill mix,
were not significant predictors of COVID-19
cases or deaths.
37 “Interpreting COVID- Das Gupta, – From June 1, 2020, through January 31,
19 Deaths among Debasree et al. 2021, nursing homes with a higher overall
Nursing Home PLOS ONE quality rating were associated with lower
Residents in the US: rates of COVID-19 deaths. However, this
The Changing Role September 1, association diminished over the study period.
of Facility Quality 2021 In addition, the duration of self-reported staff
Over Time” shortages by nursing homes were
associated with COVID-19 mortality rates
and the adverse role of staff shortages did
not change over the study period. Nursing
home ownership and shortage of personal
protective equipment were not associated
with COVID-19 mortality at any point during
the study period, but community level
variables (such as local COVID-19 death
rate) were significant determinants of nursing
home mortality outcomes.
38 “State Social Li, Yue et al. – From June to August 2020, stronger state
Distancing Scientific Reports social distancing measures were associated
Restrictions and with lower weekly rates of new COVID-19
Nursing Home January 20, 2022 confirmed cases and related deaths among
Outcomes” nursing home residents, as well as lower
weekly COVID-19 new confirmed case rate
among nursing home staff. The magnitude of
these associations was larger for nursing
homes serving disproportionately more racial
and ethnic minority residents.
Page 42 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Appendix I: Literature Review—Methodology
Description and Results
Methodological focus of study
Factors Factors
associated Factors associated
with the associated with the
presence with the duration
of COVID- severity of of COVID-
Authors 19 COVID-19 19
outbreaks outbreaks outbreaks
Journal in nursing in nursing in nursing
Title Publication date homes homes homes Summary of key findings
39 “Nursing Home Cronin, – – From the start of the pandemic through
Quality, COVID-19 Christopher J. September 13, 2020, a nursing home’s
Deaths, and Excess and William N. overall five-star rating was highly predictive
Mortality” Evans of COVID-19 mortality, with five-star nursing
Journal of Health homes having 15 percent fewer resident
Economics COVID-19 deaths than one-star nursing
homes. This relationship between quality and
January 21, 2022 COVID-19 deaths did not continue after
September 13, 2020, through April 25, 2021.
Conversely, higher-quality nursing homes
had much higher non-COVID-19 mortality,
with five-star nursing homes experiencing
11.4 percent more non-COVID-19 deaths
than one-star homes as of September 13,
2020. This relationship strengthened over
time through April 2021.
40 “Firm Finances and Begley, Taylor A., – – For a sample of 7,045 nursing homes, as of
the Spread of and Daniel October 18, 2020, those nursing homes with
COVID-19: Evidence Weagley less liquidity and those experiencing more
from Nursing Homes” SSRN severe cash flow shocks had more cases of
COVID-19. For example, increased Medicaid
July 28, 2022 nursing home reimbursement rates was
associated with much lower rates of COVID-
19 and the difference was largest for nursing
homes with the highest share of Medicaid
residents.
Source: Studies GAO reviewed. │GAO-23-104291
Notes: We selected studies for inclusion in our review if they used data to examine the association of
any factor with COVID-19 outbreaks in U.S. nursing homes.
Studies listed in the table used nationwide data except where specific states are indicated in the
summary.
Page 43 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Appendix II: Detailed Description of
Appendix II: Detailed Description of
Methodology for Multivariate Statistical Models
Methodology for Multivariate Statistical
Models
This appendix describes our methods for analyzing how certain nursing
home factors were associated with the duration of COVID-19 outbreaks.
To do this, we used Centers for Disease Control and Prevention (CDC),
Centers for Medicare & Medicaid Services (CMS), and Health Resources
and Services Administration data to develop Cox proportional hazard
statistical models, which simultaneously evaluate the association between
many factors on the probability of a specified outcome and the
significance of the relationship. Specifically, our models estimated the
probability of an outbreak ending at any given week from June 14, 2020,
through January 2, 2022, controlling for a host of factors, such as nursing
home characteristics and levels of COVID-19 in the surrounding
community, among others. The following describes our definitions, data
sources, and methodology.
CDC COVID-19 Nursing We used CDC’s COVID-19 nursing home data as the foundation for our
Home Data statistical model. CDC collects weekly nursing home-level COVID-19
information, including cases, deaths, and shortages of staff and supplies,
from all Medicare- or Medicaid-certified nursing homes. 1 We analyzed the
data for the week ending June 14, 2020, through the week ending
January 2, 2022, the last full week of the 2021 calendar year. CMS began
requiring nursing homes to report COVID-19 data to CDC in May 2020.
Reporting information retroactively to the beginning of the pandemic was
voluntary and the data do not clearly distinguish between nursing homes
that opted to report information prior to this time and those that did not.
Therefore, we started our analysis with data for the week ending June 14,
2020, and excluded data from the first 3 weeks of reporting because
these data contained information from multiple weeks dating back to
February 2020.
COVID-19 Outbreaks in According to CDC, a COVID-19 outbreak starts the week a nursing home
Nursing Homes reports a new COVID-19 case in a resident or staff member and ends
when the nursing home has 2 consecutive weeks (not included in the
outbreak duration) where they report no new staff or resident cases. 2
Using this definition of an outbreak, we analyzed CDC’s COVID-19
1The CDC data on COVID-19 in nursing homes were accessed on May 12, 2022, for the
week ending May 1, 2022, from https://data.cms.gov/covid-19/covid-19-nursing-home-
data. Nursing homes report aggregate data to CDC on a weekly basis. According to CDC,
data used in this analysis are part of a live data set, meaning that facilities can correct the
data at any time.
2See Centers for Disease Control and Prevention, Interim Infection Prevention and
Control Recommendations to Prevent SARS-CoV-2 Spread in Nursing Homes, accessed
September 7, 2022, https://www.cdc.gov/coronavirus/2019-ncov/hcp/long-term-care.html.
Page 44 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Appendix II: Detailed Description of
Methodology for Multivariate Statistical Models
nursing home data for descriptive statistics, including the number and
duration of nursing home outbreaks each nursing home experienced.
From June 14, 2020, through January 2, 2022, 15,281 nursing homes
reported data to CDC and we found that those homes had a total of
102,992 outbreaks. Nursing homes had an average of 7.6 outbreaks and
a median of eight outbreaks (2,435 nursing homes) over the review
period, ranging from zero (25 nursing homes) to 16 (one nursing home)
outbreaks. Forty-two percent of the outbreaks (43,753 of 102,992
outbreaks) lasted 1 week, with outbreak duration averaging 4 weeks
(6,316 outbreaks) and ranging to 53 weeks (one outbreak).
Other Data Sources We used additional data sources to create measures of independent
variables for the regression analysis. In order to understand our universe
of nursing homes, we examined the descriptive information for each
variable.
Nursing Home Characteristics CMS’s publicly available nursing home provider information file contains
information on a number of characteristics, such as number of beds and
non-profit or for-profit status, of nursing homes certified to participate in
Medicare or Medicaid. 3 This file contained information on 15,622 nursing
homes during our period of review. To determine whether a nursing home
was located in a metropolitan or micropolitan core based statistical area,
or neither, we merged the provider information file with Health Resources
and Services Administration’s publicly available county-level Area Health
Resources Files. 4 Using the merged file, we examined the breakdown of
all Medicare- and Medicaid-certified nursing homes by facility
characteristic for all categorical variables. (See table 3). In addition, for
two continuous variables, we calculated the average across all nursing
3The CMS publicly available nursing home provider information files were accessed on
July 22, 2022, from https://data.cms.gov/provider-data/dataset/4pq5-n9py.
4The Health Resources and Services Administration publicly available 2020 and 2021
Area Health Resources Files (county-level data) were accessed on July 22, 2022, from
https://data.hrsa.gov/data/download.
The Health Resources and Services Administration defines core based statistical areas as
the following: (a) Metropolitan—at least one urbanized area of 50,000 or more population
plus adjacent territory that has a high degree of social and economic integration with the
core as measured by commuting ties; and (b) Micropolitan—at least one urban cluster of
at least 10,000 but less than 50,000 population, plus adjacent territory that has a high
degree of social and economic integration with the core as measured by commuting ties.
We categorized all counties that did not meet the prior two criteria as rural.
Page 45 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Appendix II: Detailed Description of
Methodology for Multivariate Statistical Models
homes, finding that nursing homes had an average of 106 beds and an
average of 78 residents during our period of review.
Table 3: Nursing Home Facility Characteristics, Categorical Variables, June 14,
2020, through January 2, 2022
Number of nursing Percentage of
Nursing home facility characteristics homes (n = 15,622) nursing homes
Ownership type
For-profit 10,651 68.2
Nonprofit 3,384 21.7
Government-owned 834 5.3
Mixed ownershipa 753 4.8
Certification type
Medicare-only 667 4.3
Medicaid-only 348 2.2
Medicare and Medicaid 14,606 93.5
Facility type
Hospital-based 699 4.5
Freestanding 14,923 95.6
Location typeb
Metropolitan 11,297 72.3
Micropolitan 2,144 13.7
Rural 2,173 13.9
Transitioning area 5 0
Change of ownership in last 12 months
Home had a change of ownership 1,128 7.2
Home did not have a change of 14,494 92.8
ownership
Average number of Medicare- and Medicaid-certified beds
Less than 50 beds 1,946 12.5
50 to 99 beds 5,948 38.1
100 to 199 beds 6,834 43.8
200 or more beds 894 5.7
Special Focus Facility statusc
Special Focus Facility 68 0.4
Special Focus Facility candidate 975 6.2
Both Special Focus Facility home and 101 0.6
candidate
Page 46 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Appendix II: Detailed Description of
Methodology for Multivariate Statistical Models
Number of nursing Percentage of
Nursing home facility characteristics homes (n = 15,622) nursing homes
Not a Special Focus Facility or 14,478 92.7
candidate
Continuing care residential communityd
Continuing care residential 1,872 12.0
community
Not a continuing care residential 13,750 88.0
community
Resident and family councilse
Resident council only 11,601 74.3
Family council only 28 0.2
Both resident and family councils 3,481 22.3
Separate resident council and family 18 0.1
council
No resident or family councils 494 3.2
Overall Five-Star ratingf
Average rating of four or five stars 5,420 34.7
Average rating of three stars 3,308 21.2
Average rating of one or two stars 6,894 44.1
Health inspections Five-Star ratingf
Average rating of four or five stars 3,739 23.9
Average rating of three stars 3,645 23.3
Average rating of one or two stars 8,238 52.7
Staffing Five-Star ratingf
Average rating of four or five stars 3,571 22.9
Average rating of three stars 4,326 27.7
Average rating of one or two stars 7,725 49.5
Quality measure Five-Star ratingf
Average rating of four or five stars 7,371 47.2
Average rating of three stars 4,096 26.2
Average rating of one or two stars 4,155 26.6
Source: GAO analysis of Centers for Medicare & Medicaid Services data and the Health Resources and Services Administrations’ Area
Health Resources Files. │GAO-23-104291
Notes: Percentages do not always add to 100 due to missing data and rounding. The percentage of
nursing homes with missing data, if any, was less than 1 percent for each category.
a
”Mixed ownership” refers to nursing homes that changed their ownership type at any point over the
period of review.
b
The Health Resources and Services Administration defines core based statistical areas as the
following: (a) Metropolitan—at least one urbanized area of 50,000 or more population plus adjacent
territory that has a high degree of social and economic integration with the core as measured by
commuting ties; and (b) Micropolitan—at least one urban cluster of at least 10,000 but less than
50,000 population, plus adjacent territory that has a high degree of social and economic integration
Page 47 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Appendix II: Detailed Description of
Methodology for Multivariate Statistical Models
with the core as measured by commuting ties. We categorized all counties that did not meet the prior
two criteria as rural. In addition, we considered a “transitioning area” to be a county where the
statistical area designation changed at some point during the period of review.
c
Nursing homes with chronic noncompliance with federal standards can be selected for the Special
Focus Facility program, which requires state survey agencies to conduct more frequent oversight and
the nursing homes to improve performance or risk termination from the Medicare and Medicaid
programs.
d
Continuing care resident communities offer multiple housing options and levels of care. A nursing
home is typically the most service-intensive housing option. Residents may move from one level to
another based on their particular needs, while typically still remaining in the community.
e
Resident and family councils are usually organized and managed by nursing home residents or the
residents’ families to address concerns and improve the quality of care and life for the resident.
The Five-Star Quality Rating System assigns each nursing home participating in the Medicare or
f
Medicaid programs an overall “star” rating, ranging from one to five. Nursing homes with five stars are
considered to have much above average quality, while nursing homes receiving one star are
considered to have much below average quality. Calculation of the overall star rating is based on
separate ratings that nursing homes receive for each of three components: health inspections,
staffing, and quality measures. We calculated the average overall and component ratings for each
nursing home during our period of review.
Nursing Home History of CMS maintains data on nursing home surveys and deficiencies, including
Infection Prevention and the dates of surveys and the number and type of deficiencies cited. 5 To
Control Deficiencies determine the number of infection prevention and control deficiencies
each nursing home received and the average number of these
deficiencies cited per nursing home, we merged three different CMS
datasets—the Certification and Survey Provider Enhanced Reports
system data, the publicly available nursing home inspection dates file,
and the publicly available nursing home health citations file. 6 (See table
5In general, CMS requires that state survey agencies conduct standard surveys, or
evaluations, approximately once each year of the state’s nursing homes and investigate
both complaints from the public and facility-reported incidents regarding resident care or
safety. Beginning in March 2020, CMS required state survey agencies to conduct focused
infection control surveys, a new type of survey in response to the pandemic with a
narrower scope than a standard survey. State survey agencies are required to perform
focused infection control surveys for 20 percent of nursing homes in their state annually,
prioritizing those facilities that report new COVID-19 cases and low vaccination rates. If a
surveyor from a state survey agency determines that a nursing home violated a federal
standard during a survey or investigation, a nursing home receives a deficiency code
specific to that standard, known as a deficiency.
6For this analysis, we analyzed the deficiency code F-880 for nursing homes that were
cited for not meeting federal standards for establishing and maintaining an infection
prevention and control program. This code went into effect as part of CMS’s restructuring
of its deficiency codes on November 28, 2017, replacing a prior deficiency code that had
been in effect for several years. For infection prevention and control deficiencies cited in
2017 prior to this restructuring, we used the prior code.
The CMS publicly available nursing home survey dates files were accessed on March 28,
2022, from https://data.cms.gov/provider-data/dataset/svdt-c123. The publicly available
nursing home health citations files were accessed on April 12, 2022, from
https://data.cms.gov/provider-data/dataset/r5ix-sfxw.
Page 48 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Appendix II: Detailed Description of
Methodology for Multivariate Statistical Models
4.) These files contained information on 13,614 nursing homes with
deficiencies in 2017 and 12,343 nursing homes with deficiencies in 2021.
Table 4: Infection Prevention and Control Deficiencies Cited in Nursing Homes, by
Year, 2017 through 2021
Total number of Average infection
Number of infection prevention prevention and control
surveyed and control deficiencies per surveyed
Year nursing homes deficiencies nursing home
2017 14,184 6,527 0.5
2018 14,594 6,874 0.5
2019 14,776 6,913 0.5
2020 15,406 9,775 0.6
2021 14,128 6,738 0.5
Source: GAO analysis of Centers for Medicare & Medicaid Services data. │GAO-23-104291
Nursing Home Staffing CMS’s payroll-based journal collects employee-level payroll information
for various types of nursing home staff, including registered nurses,
licensed practical nurses, and nurse aides, from all Medicare- or
Medicaid-certified nursing homes. This dataset included information from
15,419 nursing homes during our period of review. 7 Using the payroll-
based journal data, we determined the daily staffing levels during our
period of review for the 32 types of staff CMS requires nursing homes to
report. 8 (See table 5.)
7According to CMS officials, 878 nursing homes had data issues outside of their control
from December 11, 2021, through December 31, 2021, possibly resulting in missing
payroll data for that time period. CMS provided us with a list of the affected nursing
homes, and 874 were in our database. We checked to see if those homes had staffing
levels consistent with projections based on prior month and prior year data and we
excluded any outbreaks that had incomplete data.
8Nursing homes must submit accurate information on direct-care staffing to CMS’s payroll-
based journal each quarter. This information must be based on payroll and other verifiable
and auditable data. CMS allows nursing homes to submit staffing information for 40
different types of staff, including both nurse staff and non-nurse staff. However, the
agency only requires reporting for the 32 types of direct-care staff. According to CMS, the
remaining eight types of staff (dentist, podiatrist, vocational service worker, clinical
laboratory service worker, diagnostic x-ray service worker, blood service worker,
housekeeping service worker, other service worker) are voluntary for nursing homes to
report because they either do not meet the definition of direct-care or are not paid by
nursing homes, so CMS could not audit their data. Due to concerns with the reliability of
the data for these eight types of staff, we did not include them in our analysis.
Page 49 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Appendix II: Detailed Description of
Methodology for Multivariate Statistical Models
Table 5: Nursing Home Average Daily Staffing Levels, June 14, 2020, through
January 2, 2022
Minimum hours Average hours Maximum hours
Type of staff worked per day worked per day worked per day
Administrator 7.6 7.7 7.8
Medical director 2.6 2.9 3.3
Other physician 4.4 4.9 5.1
Physician assistant 5.2 5.9 6.3
Registered nurse director of 7.8 7.9 8.0
nursing
Registered nurse with 7.7 7.7 7.8
administrative duties
Registered nurse 8.3 8.4 8.5
Licensed practical nurse with 7.9 8.0 8.1
administrative duties
Licensed practical nurse 8.5 8.6 8.7
Certified nurse aide 7.8 7.9 8.0
Nurse aide in training 7.3 7.6 7.8
Medication aide/technician 8.3 8.5 8.6
Nurse practitioner 5.7 6.0 6.1
Clinical nurse specialist 7.1 7.5 7.8
Pharmacist 3.4 3.8 4.2
Dietitian 6.7 6.8 7.0
Paid feeding assistant 6.9 7.1 7.3
Occupational therapist 6.0 6.2 6.3
Occupational therapy 6.2 6.4 6.5
assistant
Occupational therapy aide 6.7 6.9 7.1
Physical therapist 5.9 6.2 6.4
Physical therapy assistant 6.2 6.4 6.5
Physical therapy aide 6.6 6.7 6.8
Respiratory therapist 8.1 8.4 8.5
Respiratory therapy 8.1 8.5 8.8
technician
Speech/language pathologist 5.1 5.5 5.7
Therapeutic recreation 7.2 7.4 7.5
specialist
Qualified activities 7.4 7.6 7.7
professional
Other activities staff 7.0 7.1 7.2
Page 50 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Appendix II: Detailed Description of
Methodology for Multivariate Statistical Models
Minimum hours Average hours Maximum hours
Type of staff worked per day worked per day worked per day
Qualified social worker 7.6 7.7 7.8
Other social worker 7.5 7.6 7.7
Mental health service worker 7.0 7.2 7.4
Source: GAO analysis of Centers for Medicare & Medicaid Services data. │GAO-23-104291
Note: The 32 staff types listed in the table are those CMS requires nursing homes to report to its
payroll-based journal.
Reported Nursing Home Staff Using CDC’s COVID-19 nursing home data, discussed above, we
Shortages during the COVID- determined the average number of weeks during our period of review that
19 Pandemic nursing homes reported facing staff shortages. (See table 6.) The dataset
included information reported from 15,466 nursing homes during our
period of review.
Table 6: Reported Nursing Home Staff Shortages during the COVID-19 Pandemic,
June 14, 2020, through January 2, 2022
Types of reported staffing shortages Average number of weeks nursing
homes reported shortage
Shortage of nursing staff 14.1
Shortage of clinical staff 1.8
Shortage of aides 15.8
Shortage of other staff 8.0
Source: GAO analysis of Centers for Disease Control and Prevention data. │GAO-23-104291
Resident Demographics CMS’s Minimum Data Set collects information on each resident’s
strengths and needs, as well as resident demographic information,
through regular comprehensive assessments conducted by all nursing
homes certified to participate in Medicare or Medicaid. 9 The dataset
included information for 15,416 nursing homes during our period of
review. Using admission, quarterly, and annual assessments in the
Minimum Data Set, we determined the average demographic makeup of
all Medicare- and Medicaid-certified nursing homes during our period of
review. (See table 7.) We also determined each nursing home’s case-
9The CMS Minimum Data Set is reported by nursing homes, which are required to
complete resident assessments at regular intervals as part of federal requirements to
participate in the Medicare and Medicaid programs. Nursing homes are required to
conduct resident assessments at entry, quarterly, at discharge, and if there are any
significant changes or corrections. During standard surveys, surveyors can evaluate
whether a nursing home’s assessments meet federal standards for accuracy.
Page 51 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Appendix II: Detailed Description of
Methodology for Multivariate Statistical Models
mix—that is, a numeric representation of the acuity of residents within a
nursing home—using CMS’s methodology. 10 We found that the total
average case-mix score in nursing homes was 43.8, ranging from 0.5 to
568.9.
Table 7: Nursing Home Resident Demographics, June 14, 2020, through January 2,
2022
Percentage of residents within a
Nursing home resident demographics nursing home
Resident age
Less than 50 3.1
50 to less than 65 13.7
65 to less than 75 22.4
75 to less than 85 30.0
Greater than 85 30.8
Resident gender
Male 40.9
Female 59.1
Resident race and ethnicity
White 75.1
Black or African American 12.7
Hispanic or Latino 5.1
Asian 1.8
Native Hawaiian or Other Pacific 0.2
Islander
American Indian or Alaska Native 0.4
Mixed ethnicity 0.4
Unknown 4.7
Source: GAO analysis of Centers for Medicare & Medicaid Services data. │GAO-23-104291
Note: Percentages do not always add to 100 due to rounding.
Community Conditions during CDC county-level COVID-19 community transmission data contain the
the COVID-19 Pandemic daily levels of community transmission by county, known as community
spread. The dataset included information on 3,222 counties. Using this
data, we analyzed the number of weeks during our period of review that
10We used CMS’s methodology to calculate the numeric case-mix index value associated
with the resource utilization group obtained from non-therapy index maximizing
classification.
Page 52 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Appendix II: Detailed Description of
Methodology for Multivariate Statistical Models
the level of COVID-19 community spread in each county was categorized
by CDC as high, substantial, moderate, or low. 11 (See table 8.)
Table 8: Number of Weeks Counties Were Designated by CDC as Having High,
Substantial, Moderate, or Low Levels of Community Transmission of COVID-19,
June 14, 2020, through January 2, 2022
Minimum Average Maximum
CDC levels of community number of number of number of
spread weeks weeks weeks
High 0 50.0 80.4
Substantial 0 11.9 37.0
Moderate 0 13.4 41.7
Low 0 6.7 82.0
Source: GAO analysis of Centers for Disease Control and Prevention (CDC) data. │GAO-23-104291
Note: CDC calculated the level of community transmission using two indicators: (1) the total new
cases per 100,000 persons within the last 7 days and (2) the percentage of positive diagnostic and
screening nucleic acid amplification tests during the last 7 days. If the two indicators suggest different
transmission levels, the higher level is selected. Possible transmission categories are high (counties
with 100 or more cumulative cases per 100,000 population or a cumulative test positivity result of
10.0 percent or higher in the past 7 days), substantial (counties with 50-99 cumulative cases per
100,000 population or a cumulative test positivity result between 8.0-9.9 percent in the past 7 days),
moderate (counties with 10-49 cumulative cases per 100,000 population or a cumulative test positivity
result between 5.0-7.9 percent in the past 7 days), or low (counties with fewer than 10 cumulative
cases per 100,000 population in the past 7 days and a cumulative percent test positivity result below
5 percent in the past 7 days).
Analyzing the Data We merged the CDC COVID-19 nursing home data with the data sets
described in the sections above, creating a single analytic data file for our
study period of June 14, 2020, through January 2, 2022.
Exclusions We merged the data using weekly COVID-19 outbreaks as our unit of
analysis. First, we excluded 117 nursing homes that had 50 weeks or
more of CDC COVID-19 nursing home data that was missing or did not
pass quality control. Second, we excluded 24 nursing homes that did not
11CDC calculated the level of community transmission using two indicators: (1) the total
new cases per 100,000 persons within the last 7 days and (2) the percentage of positive
diagnostic and screening nucleic acid amplification tests during the last 7 days. If the two
indicators suggest different transmission levels, the higher level is selected. Possible
transmission categories are high (counties with 100 or more cumulative cases per
100,000 population or a cumulative test positivity result of 10.0 percent or higher in the
past 7 days), substantial (counties with 50-99 cumulative cases per 100,000 population or
a cumulative test positivity result between 8.0-9.9 percent in the past 7 days), moderate
(counties with 10-49 cumulative cases per 100,000 population or a cumulative test
positivity result between 5.0-7.9 percent in the past 7 days), or low (counties with fewer
than 10 cumulative cases per 100,000 population in the past 7 days and a cumulative
percent test positivity result below 5 percent in the past 7 days).
Page 53 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Appendix II: Detailed Description of
Methodology for Multivariate Statistical Models
have data in CMS’s publicly available data files. This resulted in a total of
15,322 nursing homes with 100,780 outbreaks. Third, we excluded 4,632
outbreaks due to missing data in CMS’s Minimum Data Set, CMS’s
payroll-based journal, or both. All other data files were complete and did
not result in any exclusions. After these exclusions, our analytic file
included 15,091 nursing homes that had a total of 96,148 COVID-19
outbreaks.
Factors Not Included in We could not control for all factors that may be associated with COVID-19
Analysis outbreak duration in nursing homes, often due to inconsistent data,
incomplete data, or a lack of data entirely. Some examples include the
following.
• Equipment shortages. We have reported on the lack of COVID-19
tests and personal protective equipment in nursing homes, particularly
during the first year of the pandemic. 12 However, while CDC collected
information from nursing homes on these types of shortages in its
COVID-19 nursing home data, these variables were inconsistent
across our period of review. For example, after the week ending on
March 7, 2021, CDC removed questions related to whether nursing
homes had any supply or a one week supply of personal protective
equipment and added in similar, but slightly different questions,
making it difficult to track consistently over time.
• Vaccination rates. The development and subsequent availability of
COVID-19 vaccines for nursing home residents and staff at the end of
2020 was an important milestone in the pandemic, which
corresponded with dramatic decreases in nursing home cases and
deaths through the first part of 2021. However, nationwide data on the
rate of vaccinations within each nursing home is incomplete—
reporting was not required until May 2021—so we could not include
these data as a variable in our regression. Instead, we ran two
separate models to account for the time period before the introduction
of vaccines and the time period after the introduction of vaccines. This
is explained further in the following section.
• State and local policies. States and localities varied throughout the
pandemic on what, if any, COVID-19 policies, such as masking
requirements or the closure of non-essential businesses, they had in
place. However, even when policies were in place, we could not
quantify how well communities surrounding a nursing home enforced
or complied with the policies. Therefore, rather than assess the
12See GAO-20-701, 129; GAO-21-191, 67; and GAO-21-265, 63.
Page 54 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Appendix II: Detailed Description of
Methodology for Multivariate Statistical Models
reliability of data on state and local COVID-19 policies, we assumed
that analyzing the effect of transmission of COVID-19 in the county
where each nursing home is located would reflect both the local
policies and the community’s compliance with the policies.
• Adherence to infection prevention and control practices. We
have previously reported on the importance of infection prevention
and control practices and CDC continues to emphasize the
importance of rigorous infection prevention and control to prevent the
spread of COVID-19 in nursing homes. 13 Because CMS collects data
on nursing home deficiencies, we were able to control for a nursing
home’s history of infection prevention and control deficiencies;
however, we could not account for the day-to-day adherence by
nursing home staff and residents to infection prevention and control
practices.
Multivariate Regression Model We conducted a series of multivariate Cox proportional hazard regression
models. The purpose of the model is to evaluate simultaneously the effect
of several factors on outbreak duration. A hazard regression model allows
us to examine how specified factors may be associated with the rate of a
particular event happening (e.g., outbreak ending) at a particular point in
time and allows time-varying covariates to capture the effects of changes
in independent variables over time. 14 The rate is commonly referred to as
the hazard rate. Predictor variables (or independent factors) are usually
termed covariates in the survival-analysis literature. The Cox model is
expressed by the hazard function denoted by λ(t). Briefly, the hazard
function can be interpreted as the probability of the outbreak ending at
time t. It can be estimated as follow:
λk (t; Zki) = λk0 exp{β'kZki(t)}
where
• k is the event, which in our case is the outbreak (k = 1, …, K).
• i is the subject, which in our case is the nursing home (i = 1, …, n).
13See GAO-22-105133.
14We did not use a hierarchical model on top of an already complex model. A hierarchical
model would help to account for potential correlation between outbreak durations within
the same nursing home. When we checked for this potential correlation in an hierarchical
model with no other factors included, we found there was no significant correlation
between outbreak durations within the same nursing home.
Page 55 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Appendix II: Detailed Description of
Methodology for Multivariate Statistical Models
• Z represents the set of covariates. Zki represents the specific covariate
associated with ki.
• λk0(t) is the baseline hazard function (hazard rate at time t).
• exp{β'kZki(t)} is the expected hazard ratio of covariates at time (t).
The value of exp(β) is the hazard ratio. A value of β greater than zero, or
equivalently a hazard ratio greater than one, indicates that as the value of
the covariate increases, the hazard increases and thus the duration of the
outbreak decreases.
Multivariate regression modeling is a statistical method that examines
multiple variables simultaneously to estimate whether each of these
variables are more likely or less likely to be associated with a certain
outcome, controlling for the other variables. A multivariate regression
analyzes the statistical influence of each individual factor with the
outcome. This type of modeling allowed us to test the association
between nursing home factors, such as nursing home size or nursing
home ownership, and the probability (hazard rate) of the COVID-19
outbreak ending, while holding other nursing home information constant
(such as staffing levels and resident demographics as independent
variables).
A Cox proportional hazard model provides an estimated hazard ratio for
each independent variable, where a value greater than one indicates a
higher likelihood of the COVID-19 outbreak ending (the dependent, or
outcome, variable) and an estimated hazard ratio less than one indicates
lower probability of the outbreak ending, controlling for all the other
independent variables. For example, an overall staffing level with an
estimated hazard ratio of 2.0 would indicate a higher probability of an
outbreak ending. In contrast, an overall staffing level with an estimated
staffing ratio of 0.5 would indicate a higher probability that the outbreak
continues.
The statistical significance of the results for each variable is determined
by a p-value of less than 0.05. As a result, in our report, we state that
hazard ratios that are statistically significant and greater than 1.00 or
lower than 1.00 indicate that nursing homes with that characteristic (e.g.,
rural location) are more likely or less likely, respectively, to have their
COVID-19 outbreak end at that point in time, relative to their
corresponding reference category. In cases where the p-value was
greater than 0.05, we report that we could not identify any statistically
significant differences, which means that we could not conclude that there
Page 56 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Appendix II: Detailed Description of
Methodology for Multivariate Statistical Models
was an association between that attribute and the likelihood of the
outbreak ending.
We developed multivariate Cox proportional hazard models to test the
extent of association with outcome and statistical significance of all
independent factors presented in table 9. 15 We ran two models—one for
the period of time prior to the introduction of COVID-19 vaccines (June
14, 2020, through January 2, 2021) and one for the period of time after
introduction of COVID-19 vaccines in nursing homes (January 3, 2021,
through January 2, 2022). 16 Hazard ratios from the multivariate model
represent simultaneous relationships of all independent factors specified
in the model with the outcome. Therefore, there is a reference category
for each categorical factor specified in the multivariate model. As an
example, for nursing home size, small and medium nursing homes are
compared against large and very large nursing homes as the reference
category in the model. Based on the results in Table 9, in the pre-vaccine
time period, small nursing homes had a statistically significant hazard
ratio of 1.444, indicating that small homes were distinct from large and
very large homes and that the small homes had shorter outbreaks than
larger homes. Further, the parameter estimate of 0.367 indicates that an
outbreak was likely to end approximately 2.5 days earlier in small homes
than in large and very large homes.
Table 9: Multivariate Regression Results from Cox Proportional Hazard Model, COVID-19 Outbreak Duration by Vaccination
Time Period, June 14, 2020, through January 2, 2022
Hazard ratio (parameter estimate)
Pre-vaccine time period (June 14, Post-vaccine time period (January
Independent factors 2020, through January 2, 2021) 3, 2021, through January 2, 2022)
Categorical variables
Reference category: High community spread
Low community spread 2.784** (1.024) 1.999** (0.692)
Moderate community spread 2.382** (0.868) 1.920** (0.652)
Substantial community spread 2.014** (0.700) 1.703** (0.533)
Reference category: Metropolitan
Rural 1.023 (0.023) 1.042* (0.041)
15The model results presented are right-censored.
16We initially ran a single model with a dummy variable for the time period that COVID-19
vaccinations were available but found that this dummy variable was overly dominate.
Therefore, by running two separate models we were able to tease out the effects of other
independent variables.
Page 57 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Appendix II: Detailed Description of
Methodology for Multivariate Statistical Models
Hazard ratio (parameter estimate)
Pre-vaccine time period (June 14, Post-vaccine time period (January
Independent factors 2020, through January 2, 2021) 3, 2021, through January 2, 2022)
Micropolitan 0.978 (-0.022) 0.990 (-0.010)
Reference category: Large (100 – 199 beds) or very
large (200 or more beds)
Medium (50 to <100 beds) 1.177** (0.163) 1.167** (0.155)
Small (<50 beds) 1.444** (0.367) 1.414** (0.346)
Reference category: For-profit
Non-profit 0.882** (-0.125) 0.884** (-0.124)
Government-owned 0.903** (-0.102) 0.799** (-0.225)
Reference category: No reported staff shortages
Reported staff shortages 0.843** (-0.170) 0.944** (-0.058)
Reference category: Lower overall star rating (one- to
three-stars)
Higher overall star rating (four- and five-stars) 0.933** (-0.070) 0.956** (-0.045)
Reference category: Lower health inspection star
rating (one- to three-stars)
Higher health inspection star rating (four- and five- 1.004 (0.004) 1.006 (0.006)
stars)
Continuous variables
Number of prior outbreaks 0.959** (-0.042) 1.074** (0.072)
Percentage of new employeesa 0.987** (-0.013) 1.003** (0.003)
Average hours per employee per week 0.996 (-0.004) 1.000 (0.000)
Number of employees per resident 0.984 (-0.016) 0.969** (-0.031)
Nursing home case-mix 0.995** (-0.005) 0.995** (-0.005)
Percentage of occupied beds 1.001 (0.001) 0.999** (-0.001)
Number of infection prevention and control deficiencies 0.955** (-0.046) 1.006 (0.006)
cited in the last year
Percentage of residents over 75 1.001 (0.001) 1.001 (0.001)
Percentage of residents less than 65 1.005* (0.005) 1.006** (0.006)
Percentage of female residents 1.001 (0.001) 1.002** (0.002)
Percentage of minority residents 1.001** (0.001) 1.001* (0.001)
Legend:
“*” indicates that this variable is statistically significant at p-value <0.05.
“**” indicates that this variable is statistically significant at p-value <0.001.
Source: GAO analysis of Centers for Disease Control and Prevention, Centers for Medicare & Medicaid Services (CMS), and Health Resources and Services Administration data. │GAO-23-104291
Notes: Multivariate Cox Proportional Hazard models are used to test the extent of association with
outcome and statistical significance of all independent factors.
The hazard ratio is the probability of an outbreak ending for the given factor relative to its reference
category. In our report, we state that hazard ratios that are statistically significant and greater than
Page 58 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Appendix II: Detailed Description of
Methodology for Multivariate Statistical Models
1.00 or lower than 1.00 indicate that nursing homes with that characteristic are more likely or less
likely, respectively, to have their COVID-19 outbreak end at that point in time.
The parameter estimate is the measure of the direction and size of the effect that a single unit change
in the independent variable has on the dependent variable. The parameter estimate is based on
weekly data (parameter estimate of 1.00 equals 1 week); therefore, a parameter estimate of
approximately +/- 0.143 is equivalent to one day.
a
New employees were defined as any employee at a specific nursing home without a record in CMS’s
payroll-based journal in the prior 6 weeks.
Data and Model Limitations Although the data quality is sufficiently reliable for the purposes of our
objective, there are some limitations to note. The CDC COVID-19 nursing
home data, which forms the backbone of the COVID-19 outbreak variable
and other variables, is self-reported weekly by nursing homes. The
number of COVID-19 cases is based on the testing technology and
infrastructure available to homes at the time the data were reported; the
availability and accuracy of the tests may have changed over time, but
there is no reason to believe that any bias in testing occurred between
homes at the time. In addition, according to CDC officials, the agency’s
definition of a COVID-19 outbreak is an epidemiologic one and has been
used during the pandemic to trigger the implementation of additional
infection prevention and control practices, such as screening testing for
residents and staff to determine if unrecognized cases were in the nursing
home. Therefore, an individual outbreak in our analysis does not
necessarily imply that all of the COVID-19 cases reported during that
outbreak were related to a single introduction of COVID-19 into the
facility.
CDC officials also noted that the inclusion of staff COVID-19 cases in our
analysis of outbreaks may inflate the total number and duration of
outbreaks because routine screening testing of asymptomatic nursing
home staff could have identified infected staff who may have never
entered the facility. This situation would reflect COVID-19 transmission in
the community rather than transmission in the nursing home.
Other variables in the COVID-19 nursing home data weekly reports, such
as the availability of staffing and personal protective equipment, are also
self-reported. While noting the staffing shortages are self-reported, we
also included CMS payroll-based journal data that provide an objective
measure of staffing. In the case of personal protective equipment and
infection containment supplies (such as hand sanitizer), CDC altered the
questions at different periods during the pandemic. This resulted in data
that could not be tracked over time and thus we did not include variables
on personal protective equipment shortages in the final models. The
Page 59 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Appendix II: Detailed Description of
Methodology for Multivariate Statistical Models
vaccination of staff and residents posed a slightly different problem, in
that CMS did not require nursing homes to report data to CDC on the
number and percent of staff and residents vaccinated until about 5
months into the vaccination program. To account for the possible role that
the presence of the vaccine played in COVID-19 outbreaks, we first
created a dummy variable to acknowledge the availability of the vaccine.
When this proved to be an overwhelmingly significant variable, we split
the model into two time periods, as described above.
Other datasets also had limitations. In its payroll-based journal CMS
allows nursing homes to submit staffing information for 40 different types
of staff, including both nurse and non-nurse staff. While CMS has
instituted checks and audits to ensure that nurse staffing data reported in
the payroll-based journal are accurate, CMS does not have a process to
ensure the accuracy and completeness of staffing information for non-
nurse staff. 17 Based on interviews with CMS officials, we determined that,
despite this limitation, we could still use most of the non-nurse payroll-
based journal data for the purposes of our analysis. However, CMS only
requires reporting for 32 of the 40 types of staff. Data for the remaining
eight types of non-nurse staff—which include housekeepers and others
who may not care for residents in a therapeutic manner but still may
come into close contact with residents—are neither required to be
reported nor audited for accuracy. 18 As a result, we decided to exclude
these eight types of staff from our analysis. Therefore, we do not know if
these types of staff were associated with the duration of outbreaks.
When using staffing variables created based on data in the payroll-based
journal, we were generally not able to separate out the influence of
certain staff roles (e.g., registered nurses and nurse aides) to our
satisfaction. While many of the homes we interviewed pointed to staffing
17The Department of Health and Human Services’ Office of Inspector General has
recommended that CMS take steps to ensure the accuracy of payroll-based journal data
for non-nurse staff that it chooses to include on its Care Compare website. For example,
CMS may consider adding physical therapists—the only type of non-nurse staff with data
currently found on Care Compare—to its audits of payroll-based journal data. See
Department of Health and Human Services Office of Inspector General, CMS Use of Data
on Nursing Home Staffing: Progress and Opportunities To Do More, OEI-04-18-00451
(Washington, D.C.: March 2021).
18According to CMS, these eight types of staff (dentist, podiatrist, vocational service
worker, clinical laboratory service worker, diagnostic x-ray service worker, blood service
worker, housekeeping service worker, other service worker) are voluntary for nursing
homes to report because they either do not meet the definition of direct-care or are not
paid by nursing homes, so their data would not be able to be audited by CMS.
Page 60 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Appendix II: Detailed Description of
Methodology for Multivariate Statistical Models
as an issue, the actual employment data, despite several different
iterations using different breakouts of staffing roles, did not indicate a
significant role by particular kinds of staff.
CDC data on community spread is based on test results. While there is
no imputation that accounts for cases that exist but for which no test was
done or tests were done at home and not reported, we used CDC’s
categorization of community spread, which is based, in part, on the
percentage of positive diagnostic and screening nucleic acid amplification
tests, which are required to be reported. CDC data is assumed to be
reliable for our purposes because we can assume that there was no
systematic bias with respect to the duration of outbreaks in nursing
homes.
In general, CMS data on nursing home surveys and investigations and
their resulting deficiencies is considered reliable. We should note that for
a number of months during the first year of the pandemic, regular
standard surveys of nursing homes and low priority investigations were
temporarily suspended, replaced by focused infection control surveys,
which evaluated compliance with CMS infection prevention and control
policies. Based on this and findings in prior reports we focused on a
home’s history of infection prevention and control deficiencies rather than
on their overall number and severity of deficiencies.
The CMS Minimum Data Set was used to determine the case-mix of the
residents in a home as well as the demographics of each home’s
population. Many variables within the Minimum Data Set are self-reported
by the resident, the family, or a staff member on behalf of a resident, and
we acknowledge this can raise some questions of data reliability, some of
which have been reported in the literature. 19 We determined that the
demographic data are sufficiently reliable. Further, we determined that
case-mix data are sufficiently reliable given that any bias in case-mix is
likely to exist in all nursing homes, so comparison of homes is not
particularly problematic.
19Some studies have found that the Minimum Data Set data reported by nursing homes
underreports anti-psychotic use and falls. For examples, see HHS Office of Inspector
General, CMS Could Improve the Data It Uses to Monitor Antipsychotic Drugs in Nursing
Homes, OEI-07-19-00490 (Washington, D.C.: May 3, 2021), and J. Mintz et al., “Validation
of the Minimum Data Set Items on Falls and Injury in Two Long-Stay Facilities,” Journal of
the American Geriatrics Society, vol. 69, no. 4 (April 2021).
Page 61 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Appendix III: Nursing Home Officials’ Appendix III: Nursing Home Officials’
Descriptions of COVID-19 Outbreaks, including
Descriptions of COVID-19 Outbreaks,
Challenges and Successes
including Challenges and Successes
Table 10: Nursing Home Officials’ Descriptions of COVID-19 Outbreaks, including Challenges and Successes
Nursing home Description of COVID-19 outbreaks, including challenges and successes
Nursing home A This nursing home kept COVID-19 out of the facility until July 2020, which was likely due to the
Non-profit proactive steps it took to restrict residents to their rooms and adopt face mask requirements. Their
first case was a staff member, and it was caught early. The required biweekly COVID-19 testing
Large (100-199 beds) caught additional staff cases throughout the pandemic. The challenges posed by the reduction in
staff due to quarantine or symptoms was “manageable although not optimal.” However, the home
reached a staffing “crisis” as the pandemic continued, noting that they were down 25 full-time
employees. The nursing home’s first resident outbreak was in October 2020 and started with five
residents and then spread to an additional 16 residents. The nursing home officials believed that
COVID-19 entered the home from a dialysis patient. The COVID-positive residents were able to be
isolated in one ward of the nursing home and staff who already had tested positive and had returned
from their quarantine were willing to work in the COVID-19 unit. A year later, in October 2021, the
home had a second “major” resident outbreak.
Nursing home B This nursing home experienced its largest outbreak in December 2021 with the Omicron surge.
For-profit During this time, 20 percent of its residents and staff tested positive. Officials attributed this outbreak
in part to the re-opening of visitation that occurred in November 2021. Prior to the outbreak, the home
Large (100-199 beds) had only seen a few cases at a time, and its COVID-19-positive residents had avoided severe illness.
A lot of the home’s prior success in keeping COVID-19 out of the facility was due to community
decisions that resulted in low levels of COVID-19. Located in an isolated area, the county was able to
close a road into the community during the worst of the pandemic in 2020 to keep travelers out.
However, these community policies were discontinued by winter of 2021, and the nursing home could
no longer restrict visitation. Officials also discussed staffing shortages and staff fatigue as challenges
experienced during outbreaks. For example, at one point the home was down five working staff
members as COVID-19-positive staff had to quarantine. In addition, the home had to staff a separate
COVID-19 unit for just a few COVID-19-positive residents during a smaller outbreak, requiring more
staff. It was also difficult to find staff willing to work the COVID-19 unit due to the risk of exposure.
Nursing home C This nursing home kept COVID-19 out of the facility until May 2020, which the home attributed to
For-profit having enough personal protective equipment, suspending new admissions, and adopting face mask
requirements. But mitigation measures only worked for a while. The first of two resident outbreaks
Very large (200+ beds) started with a few staff contracting COVID-19 from the community. Soon after, two residents tested
positive, then two more, and so on. During the first outbreak, the virus spread slowly, but the second
outbreak in late 2020 spread more rapidly—from two residents to 10 or 12 residents almost
immediately. During the first outbreak, the boundaries of the COVID-19 unit never changed, but
during the second outbreak the home had to keep expanding the boundaries. Throughout the
outbreaks, the nursing home was able to maintain its staffing levels—even when staff were out sick
with COVID-19. However, the psychosocial toll had significant negative effects on residents. The
suspension not only of visitation but also dining facility activities resulted in declines in resident
physical and mental health. The staff began to see residents with weight loss and an increased need
for anti-depressant drugs. Many of the residents just “gave up.”
Page 62 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Appendix III: Nursing Home Officials’
Descriptions of COVID-19 Outbreaks, including
Challenges and Successes
Nursing home Description of COVID-19 outbreaks, including challenges and successes
Nursing home D This nursing home was hit “like a brick” in the fall of 2020. Earlier in 2020, a COVID-19 outbreak
Non-profit affected about a dozen residents, but cases in its outbreak in the fall of 2020 “wreaked havoc” on
residents and staff, with 40 residents dying. The loss of so many residents took a heavy emotional
Very large (200+ beds) toll on staff and fellow residents. Officials believed the virus entered the facility through staff as, at the
time, the home’s staff and the surrounding community had low vaccination rates. In addition to the
emotional toll on residents and staff, the number of resident deaths plus staff attrition resulted in the
nursing home losing about 50 percent of its capacity and closing a sister nursing home. The home
learned a lot about the importance of communication during the second outbreak. For example, to
facilitate communication within the home, task force teams met twice a week, and to facilitate
communication to families, a hotline was created and a weekly newsletter was developed to provide
relevant COVID-19 information. However, the psychosocial toll during that time took the joy away
from caregiving and left many staff with trauma. Further, the fact that the home is located in a rural
area with limited available staff, made maintaining staffing levels very difficult, with officials noting
“scary” staffing shortages.
Nursing home E This nursing home kept COVID-19 out of the facility until July 2020 and used its outbreak-free time
For-profit learning to adapt to infection control procedures that very few staff were doing correctly before the
pandemic. The July 2020 outbreak lasted about 5 weeks and culminated with 32 of about 70
Medium (50-99 beds) residents and 12 staff infected. Eight of those residents died. After that outbreak, the nursing home
did not experience any additional large outbreaks, only a few isolated cases. Officials could not trace
the origin of the outbreak, but believed that COVID-19 was brought into the facility by a staff member.
Further exacerbating the outbreak was a limited supply of N95 masks, so many residents and some
staff were using regular procedure masks instead, as well as physical space constraints. Officials
said that to isolate COVID-19-positive residents, the home had to “shuffle” healthy residents from one
room to another, which may have accidentally spread the virus. In addition, the nursing home’s
resident population and nurse staff decreased significantly. Specifically, the nursing home lost 23
residents and about half of its staff left. This resulted in closing a section of the home in December
2020 and having to rely on traveling nurses and aides for the home’s staffing needs.
Nursing home F Early in the pandemic, this nursing home had some staff members test positive, but COVID-19 was
For-profit not transferred to the residents. The first resident did not test positive until November 2020. Officials
believed that the virus was brought into the facility by staff. The home had another outbreak in
Large (100-199 beds) January 2021 and was able to contain COVID to just one floor of their three-floor facility for a time,
but eventually the virus spread. Staff were out sick with COVID-19 during these outbreaks. With
limited staff and the building’s physical space constraints, officials had to make difficult choices. For
example, there were times where nursing home officials felt they had no choice but to move staff
between COVID-19 and non-COVID-19 units. As the pandemic wore on, staff had a hard time
adjusting to the re-opening of visitation in November 2021. Officials said that they believed their staff
were exposed to trauma as a result of their early experiences with COVID-19.
Source: GAO interviews with officials from selected nursing homes in four states. │GAO-23-104291
Page 63 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Appendix IV: GAO Contact and Staff
Appendix IV: GAO Contact and Staff
Acknowledgments
Acknowledgments
John E. Dicken at (202) 512-7114 or dickenj@gao.gov
GAO Contact
In addition to the contact named above, key contributors to this report
Staff were Karin Wallestad (Assistant Director), Kathryn Richter (Analyst-in-
Acknowledgments Charge), Julianne Flowers, Isabella Guyott, Sarah-Lynn McGrath, Elise
Pressma, and Elaina Stephenson. Also contributing to the report were
Laurie Pachter, Vikki Porter, Patricia Powell, Bryan Ricciardi, Ravi
Sharma, Anna Beth Smith, Roxanna Sun, Jeff Tamburello, Jennifer
Whitworth, and Sirin Yaemsiri.
Page 64 GAO-23-104291 Nursing Home COVID-19 Outbreaks
Related GAO Products
Related GAO Products
COVID-19 in Nursing Homes: CMS Needs to Continue to Strengthen
Oversight of Infection Prevention and Control. GAO-22-105133.
Washington, D.C.: September 14, 2022.
Health Care Capsule: Improving Nursing Home Quality and Information.
GAO-22-105422. Washington, D.C.: January 14, 2022.
COVID-19: Continued Attention Needed to Enhance Federal
Preparedness, Response, Service Delivery, and Program Integrity.
(Nursing Homes Enclosure). GAO-21-551. Washington, D.C.: July 19,
2021.
COVID-19 in Nursing Homes: Most Homes Had Multiple Outbreaks and
Weeks of Sustained Transmission from May 2020 through January 2021.
GAO-21-367. Washington, D.C.: May 19, 2021.
COVID-19: Sustained Federal Action Is Crucial as Pandemic Enters its
Second Year. (Nursing Homes Enclosure). GAO-21-387. Washington,
D.C.: March 31, 2021.
COVID-19 in Nursing Homes: HHS Has Taken Steps in Response to
Pandemic, but Several GAO Recommendations Have Not Been
Implemented. GAO-21-402T. Washington, D.C.: March 17, 2021.
COVID-19: Critical Vaccine Distribution, Supply Chain, Program Integrity,
and Other Challenges Require Focused Federal Attention. (Nursing
Homes Enclosure). GAO-21-265. Washington, D.C.: January 28, 2021.
COVID-19: Urgent Actions Needed to Better Ensure an Effective Federal
Response. (Nursing Homes Enclosure). GAO-21-191. Washington, D.C.:
November 30, 2020.
COVID-19: Federal Efforts Could Be Strengthened by Timely and
Concerted Actions. (Nursing Homes Enclosure). GAO-20-701.
Washington, D.C.: September 21, 2020.
COVID-19: Opportunities to Improve Federal Response and Recovery
Efforts. (Nursing Homes Enclosure). GAO-20-625. Washington, D.C.:
June 25, 2020.
Infection Control Deficiencies Were Widespread and Persistent in Nursing
Homes Prior to COVID-19 Pandemic. GAO-20-576R. Washington, D.C.:
May 20, 2020.
(104291)
Page 65 GAO-23-104291 Nursing Home COVID-19 Outbreaks
The Government Accountability Office, the audit, evaluation, and investigative
GAO’s Mission arm of Congress, exists to support Congress in meeting its constitutional
responsibilities and to help improve the performance and accountability of the
federal government for the American people. GAO examines the use of public
funds; evaluates federal programs and policies; and provides analyses,
recommendations, and other assistance to help Congress make informed
oversight, policy, and funding decisions. GAO’s commitment to good government
is reflected in its core values of accountability, integrity, and reliability.
The fastest and easiest way to obtain copies of GAO documents at no cost is
Obtaining Copies of through our website. Each weekday afternoon, GAO posts on its website newly
GAO Reports and released reports, testimony, and correspondence. You can also subscribe to
GAO’s email updates to receive notification of newly posted products.
Testimony
Order by Phone The price of each GAO publication reflects GAO’s actual cost of production and
distribution and depends on the number of pages in the publication and whether
the publication is printed in color or black and white. Pricing and ordering
information is posted on GAO’s website, https://www.gao.gov/ordering.htm.
Place orders by calling (202) 512-6000, toll free (866) 801-7077, or
TDD (202) 512-2537.
Orders may be paid for using American Express, Discover Card, MasterCard,
Visa, check, or money order. Call for additional information.
Connect with GAO on Facebook, Flickr, Twitter, and YouTube.
Connect with GAO Subscribe to our RSS Feeds or Email Updates. Listen to our Podcasts.
Visit GAO on the web at https://www.gao.gov.
Contact FraudNet:
To Report Fraud,
Website: https://www.gao.gov/about/what-gao-does/fraudnet
Waste, and Abuse in
Automated answering system: (800) 424-5454 or (202) 512-7700
Federal Programs
A. Nicole Clowers, Managing Director, ClowersA@gao.gov, (202) 512-4400, U.S.
Congressional Government Accountability Office, 441 G Street NW, Room 7125, Washington,
Relations DC 20548
Chuck Young, Managing Director, youngc1@gao.gov, (202) 512-4800
Public Affairs U.S. Government Accountability Office, 441 G Street NW, Room 7149
Washington, DC 20548
Stephen J. Sanford, Managing Director, spel@gao.gov, (202) 512-4707
Strategic Planning and U.S. Government Accountability Office, 441 G Street NW, Room 7814,
External Liaison Washington, DC 20548
Please Print on Recycled Paper.
File and source
- File
- 23-104291-covid-19-in-nursing-homes-outbreak-duration.pdf
- Size
- 727,310 bytes
- SHA-256
- 1698b26115088038a3739642b3d08d3ed6ce2eef0dd335eeaa5d10cdd95984b0
- Original
- www.gao.gov