The story of COVID-19 deaths in the United States is one of long-term care facilities. Data compiled by The COVID Tracking Project over the last two months shows that at least 418,463 cases and 73,478 deaths have occured in nursing homes, assisted living facilities, and other long-term care settings nationwide. This represents 7 percent of all US cases and an astounding 43 percent of deaths. Yet we are still learning how states count and report these figures. Since some states don’t report data on long-term care facilities (LTCs), or do so in an opaque way, our accounting for the toll of COVID-19 on LTC residents and staff is almost certainly an undercount.
To put the above numbers in context, in 2016 the US Department of Health and Human Services estimated the number of residents of nursing homes and residential care facilities to be about 2.1 million, less than 1 percent of the U.S. population. Using this estimate together with our tallies of deaths in LTCs, we can estimate that more than 1 in 28 LTC residents in the US have died of COVID-19 since March, with the rate much higher in certain states.
The COVID Tracking Project has been capturing daily data published by states on new COVID-19 tests, cases, deaths, and hospitalizations since March. But as reports of the virus's devastating effects in nursing homes and similar congregate-care settings came to light, we realized that we needed to capture the patchwork data that states were beginning to report. Since mid-May, we have been tracking and standardizing long-term care COVID-19 data on deaths, cases, and impacted facilities, creating what we believe is the most comprehensive data set on these facilities to date. This data, though inconsistent, messy, and almost certainly incomplete, allows us to track the trajectory of the virus in long-term care facilities over time and to assess the outsized role these facilities have played in COVID-19 deaths.
The difficulty of mapping inconsistent data
A nationwide map of the share of deaths occurring in long-term care facilities illustrates both how devastating COVID-19 has been for residents and staff of these facilities and how difficult it is to obtain accurate and consistent data. In some states, like New Hampshire and Minnesota, the vast majority of deaths from COVID-19 are connected to LTCs. In states like New York and Nevada, the share is much lower, though still painfully high, at more than 20 percent. That’s likely the result of how different states define deaths in LTCs: New York, for example, reports LTC deaths only when they occur in a facility, not when an LTC resident is transported to the hospital and dies there. If states all used consistent reporting rules, the official death count in LTCs would likely be substantially higher.
All COVID-19 data is inevitably difficult to collect because of the different rules that govern its collection and reporting across the country. For long-term care data, the rules are even more opaque and the reporting even less consistent. Nine states don’t publicly report cases or deaths. Florida reports current cases and facility outbreaks, but cumulative deaths only at the facility level. Some states report staff and resident data combined. States use different metrics for what constitutes an outbreak at a facility. They differ in when they began counting cumulative statistics. Some states only include deaths that occurred within an LTC facility, not those of people who may have contracted the virus in such a facility but died in a hospital.
What’s a long-term care facility?
The COVID Tracking Project’s data on LTCs is divided into four categories: nursing homes/skilled-nursing facilities, assisted-living facilities, other residential facilities, and uncategorized long-term care facilities. Nursing homes and skilled nursing facilities are, as their names suggest, intended for residents with significant health issues. “Nursing home” generally describes a facility where residents are expected to remain permanently, while “skilled nursing” often describes a facility where patients who need care after a hospital stay are discharged until they are ready to return home. In reality, there’s overlap between the two. Importantly, both types of facilities are subject to federal regulations and accept patients covered by Medicare and Medicaid, the federal healthcare programs that provide the majority of nursing-home payments.
The federal government grades these facilities on their quality of care; researchers are currently exploring whether there’s a correlation between these grades and COVID-19 outcomes. In May Tamara Konetzka, who teaches health-services research at the University of Chicago, told a US Senate subcommittee that an analysis she conducted of 12 geographically diverse states did not find a consistent meaningful relationship between COVID-19 deaths and ratings. However, her analysis, along with one conducted by the Chicago Sun-Times, did find a correlation between COVID-19 outcomes and facilities, with the lowest grades in Cook County, Illinois. A recent study in the Journal of the American Medical Association, using data from eight states, found a correlation between high ratings for nurse staffing and low levels of COVID-19 cases at facilities, but no significant correlation between health inspections or general quality ratings and the number of infections. (According to research from Harvard and the University of Rochester, if current trends continue, COVID-19 will make caregiving the deadliest occupation in the United States in 2020, with death rates more than double that of last year’s deadliest job.)
Assisted-living facilities generally serve residents with less-severe medical needs than nursing homes, and just 15 percent of ALF payments are made through Medicaid. As a January 2020 study in Nursing Outlook notes, 97 percent of nursing home residents require help with bathing, and 89 percent need assistance with toileting. By comparison, about two-thirds of ALF residents need bathing help, and 40 percent need help with toileting. But the piece also shows that in recent years, ALFs have taken on more residents with greater medical needs who often require skilled nursing, concluding, “Despite this increase in care demands, little is known about ALF regulations that may affect the quality and safety of care in ALFs.” ALFs are heavily staffed by direct-care workers, who are unlicensed and whose required training varies by state (training is not required at all in Mississippi). ALFs are generally inspected less frequently than nursing homes, and the frequency and manner in which states collect and report complaint data varies.
It does not necessarily follow, however, that assisted-living facilities have worse COVID-19 outcomes than nursing homes. We don’t know, in part because of inconsistencies in how data is collected. The majority of states don’t distinguish between the two in their COVID-19 data reporting. Seventeen states separate out deaths in nursing homes, totaling nearly 26,000 deaths, and of those 17, just 11 separate out deaths in assisted living facilities, which report just over 1,600 deaths. And still, nine states (AK, AL, AZ, MO, MT, NM, NE, SD, and WY) do not publicly report cases or deaths. The “other” residential facilities category includes group homes, personal care homes, residential care homes, and board and care homes, which are generally regulated at the state level. The “uncategorized long-term care facilities” category represents data from states that combine both state and federally regulated facilities into one lump sum for cases, deaths, or facilities. Due to some states’ aggregated reporting, this category may contain nursing home and assisted living facilities along with chemical dependency residential treatment centers, behavioral health residential facilities, and intermediate care facilities for individuals with intellectual disabilities.
What we know about LTC cases and deaths right now
States generally do not report this data with a time-series. That’s where the COVID Tracking Project comes in. Beginning May 21, the project began creating weekly time-series reports by gathering cumulative data from states on an ongoing basis to observe trends of cases and deaths in long-term care facilities.
Following general trends in COVID-19 cases, the vast majority of cases in long-term care facilities in the Northeast occurred in the first three months of the pandemic. The reverse is true in the South, where there have been nearly three times as many cases in the summer as in the spring. The South has reported almost as many cases in June, July, and August as the Northeast has over the past six months, but this likely reflects the greater availability of testing during the later Southern outbreaks.
Reported deaths remain highest in the Northeast, and this is almost certainly a substantial undercount, because New York only reports COVID-19 deaths in facilities, not facility-linked deaths in hospitals. Why has the Northeast had fewer reported cases than the South but so many more deaths? Two factors are likely responsible: a lack of available testing during the early outbreak, leading to an undercount of cases, and a better understanding of the disease and how to prevent infections within LTCs once the coronavirus began spreading in the South. But it’s also possible that deaths in the South will continue to rise relative to the Northeast, even as case counts begin to fall, because of the death lag.
Per capita deaths in the over-65 population remain highest in the states hit hardest at the beginning of the pandemic, beginning with four Northeast Corridor states. Illinois, the first Midwestern state with a major outbreak, is fifth. Following that are four Southern states, followed by Pennsylvania. Colorado is the only Western state in the top 20.
Drawing lessons from the differences in how states responded is difficult at this point, in part because of how data is reported. New York is a good example: the state was the epicenter of the pandemic in its early stages, and its 25,000-plus deaths make up about one-sixth of those nationwide, with about 10,000 more than neighboring New Jersey, another early hotspot. But even with its larger population and higher number of COVID-19 cases, New York reports fewer LTC deaths than New Jersey: about 6,300 versus 6,800, and a much lower per-capita count. This is likely because, as mentioned earlier, New York only counts LTC deaths when a patient dies in a facility; New Jersey, like most states, counts deaths connected to LTCs.
The New York Times explored a common criticism of New York Governor Andrew Cuomo: discharging COVID-19 patients to LTCs in order to alleviate hospital overcrowding spread the disease and may be responsible for New York’s tremendously high death count. But, as the piece notes, “roughly” 12 states did the same, including New Jersey. Connecticut and Massachusetts created COVID-19-only facilities, which was considered an appropriate response, but their per-capita over-65 deaths in LTCs are also quite high. (Seventy percent of deaths in Connecticut are connected to LTCs.) Determining which policies worked—and which didn't work or exacerbated the problem—will be critical for future pandemics, if not for the ongoing one.
Since the first states to suffer from large COVID-19 outbreaks also have the highest per-capita over-65 deaths in LTCs, experts are watching states where cases rose during the summer months. Many of these states currently have a low median age for COVID-19 cases, raising hopes that, while the disease can be very harmful or deadly for younger patients, lessons learned in the Northeast corridor might be keeping the virus out of LTCs. LTC cases and deaths in these states are elevated compared to the spring, but may be heading in the right direction since the last August reporting period.
Florida has seen one of the largest recent outbreaks, and deaths have followed during the last three reporting periods.
Texas, another large Sunbelt state with a substantial summer outbreak, saw a huge spike in LTC infections during July, with more than half its cumulative cases coming over a two-week period. About a third of its cumulative LTC deaths have occurred since mid-July; fortunately, cases and outbreaks have declined in August, and deaths seem to be on the same trajectory.
Arizona is another point of focus for its recent increases in cases, hospitalizations, and deaths. Unfortunately, the state does not separately report cases or deaths in LTC facilities. It does report facilities with outbreaks cumulatively. That number increased during June and July, while numbers in August are comparable to those in May.
In Georgia, LTC cases spiked in July and early August. Recently the state has reported about 1,000 cases a week, about twice as many weekly cases as were reported in late May and the month of June. Cases and deaths fell in the most recent reporting period, closer to late-May levels.
Why did LTC cases and deaths spike in the summer?
COVID-19’s toll on long-term care facilities was evident early in the pandemic, both abroad and in the Northeast corridor, where the early US outbreaks were most concentrated. But despite clear evidence of the virus’ deadly impact on elderly populations, COVID-19 cases and deaths in long-term care facilities grew considerably as the virus spread south and west over the summer. Why weren’t the spring's lessons better learned?
The combination of an uneven regulatory landscape—across states and types of facility—and a lack of resources, such as inadequate testing capacity, remains a problem. According to Pew’s Stateline, “In at least 32 states, at least one round of testing of all long-term care residents and staff has been mandated,” while four states only mandate testing of staff. But ongoing, reliable testing is required to detect and isolate potential outbreaks, and Pew’s Stateline reports significant testing failures: 60 percent of respondents to one survey conducted at the end of June said test results took up to four days. The CEO of one Texas LTC chain told Pew that test results sometimes took up to three weeks, and it cost his company $300,000 a week to test the 4,200 staff and residents in the company’s 31 facilities.
In late April, Senators Elizabeth Warren (D-MA) and Edward Markey (D-MA) began contacting the 11 largest LTC operators in the country to learn about how they are dealing with the pandemic. Last month they released their findings. Operators told the senators that they had sufficient access to testing for symptomatic patients and staff, but not for asymptomatic testing. According to the report, “most operators reiterated that CDC guidance does not recommend routine testing,” though the report notes that a May 3 CDC document calls the testing of asymptomatic individuals “appropriate,” including the testing of those without known exposure “for early identification in special settings.”
On July 22, the US Centers for Medicare & Medicaid Services (CMS) began requiring that nursing homes in states with a positivity rate above 5 percent test all staff, promising 15,000 testing devices “over the next few months,” according to Bloomberg Law. About a week prior, CMS announced it was sending 2,000 point-of-care tests to nursing homes, with a goal to supply all 15,400 facilities in the US. Recently, however, the president of the California Association for Long Term Care Medicine told Talking Points Memo that machines sent to the facilities his organization oversees came without instructions or clear reporting guidelines.
All the facility operators surveyed by Sens. Warren and Markey said they were able to get personal protective equipment, but that the process was expensive and difficult; several reported costs of $1 million to $2 million a month. Two operators set up regional supply hubs with other companies in order to ease the process; one chartered aircraft to deliver PPE. These are, of course, the very largest operators; smaller facilities may have more difficulty obtaining equipment. According to an analysis of CMS data by the Kaiser Family Foundation, at the end of June, 2,700 nursing homes in the US reported PPE shortages. The New York Times reports that a Federal Emergency Management Agency program aiming to supply 15,000 nonprofit nursing-care facilities—chosen for their lack of resources—with two weeks’ worth of PPE, has resulted in complaints about inadequate and poorly manufactured supplies, and a complete absence of N95 masks. Tamara Konetzka, the University of Chicago health economist who recently presented her research on nursing-home grades to Congress, told the Times that “the federal response to protect one of the most vulnerable populations in the country has been a dismal failure.”
The federal government has also struggled to collect data from long-term care facilities. A report critical of the Trump administration from the Senate Special Committee on Aging, which is led by three Democratic senators, notes that the first recorded COVID-19 outbreak in the US occurred at a nursing home in Washington state in late February, but the CMS did not “formally issue requirements for data collection until May 8” and did not report complete data until June 4. The federal government doesn’t regulate assisted living facilities and isn’t collecting data on them. According to ABC News, such facilities have received no direct federal aid or PPE.
In a July 6 piece for The Atlantic, Olga Khazan explored why states downstream of the early COVID-19 outbreaks were struggling with LTC cases, and found a lack of PPE, especially masks; “poor and delayed data collection”; late and inconsistent testing, especially of asymptomatic patients and staff; and uneven responses and capabilities from states and localities to fill in the considerable gaps left by the federal government’s response.
Khazan contrasted the US’s response with that of Hong Kong, which learned hard lessons from SARS and came into the COVID-19 outbreak with a full month’s supply of face masks and PPE. Unlike in the US, LTC staff in Hong Kong used masks from the outset. Rather than scrambling to decide what should happen to COVID-positive LTC residents, Hong Kong required that all such patients would be put in hospital wards until testing negative twice.
The United States has started to learn some of these lessons as well, and there’s some evidence that recent changes have had a real effect on LTC cases and deaths. But as long as data collection and regulatory oversight remain inconsistent, and as long as personal protective equipment and testing remain in short supply, cases in LTCs may remain difficult to manage.
Many people at the The COVID Tracking Project contribute to the research and data compilation efforts that make the Long-Term Care COVID Tracker possible. Special thanks to the long-term care data entry team that compiles and cleans this data every Thursday. The team is led by Artis Curiskis and Kara Oehler. Contributors include Aarushi Sahejpal, Aliya Uteuova, Annie Rydland, Caitlin Glennon, Danielle Bloch, Denise Heitzenroder, Emma Rubin, Emily Temple-Wood, Eva Sher, Evan Sorenson, Glen Johnson, Hannah Cummins, Hannah Hoffman, Isha Pasumarthi, J. Albert Bowden II, Jake Burns, Javier E Thomas, Jeffrey Ndubisi, Jesse Anderson, Matthias Shapiro, Meghan Burks-Lenz, Michael Spicer, Nathanael Roy, Pat Kelly, Rick Palmer, Tara Devlin, Terra July Riley, Sarah Anderson, Stacey Rupolo, Steve Thomas, and William John Condon. Data visualizations by Conor Kelly.
Thanks also to the reporting contributors who gather data directly from states: Caroline Dulaney, Ciara McCarthy, Deirdre Kennedy, Elizabeth Eads, Jessamine Fitzpatrick, Laura Bult, Rowan Moore Gerety, Ryan Kailath, and Sarah Hoffmann. Special thanks to Carol Cronin.
Though cases are rising in parts of the Midwest, hospitalizations in the West and South continued trending downward. The Labor Day holiday impacted data reporting lag times both this week and last, obscuring what had been positive trends in September.
We track and analyze data from cities and counties, including breakdowns by race, ethnicity, and ZIP code when possible. The differences we find between cities, counties, and states give useful insights into response efforts.
Our volunteer-powered work depends on the contributions of hundreds of individuals, but it also wouldn’t be possible without the support of the tech companies that have donated their services to help us connect with each other, run a fast and stable website and API, and tell the stories emerging from the data we compile.