The True COVID Death Toll in New York State Long-Term Care Facilities

Executive Summary

For nearly a year, New York Governor Andrew Cuomo touted his administration’s response to the coronavirus, including in nursing homes. In an Aug. 10 call with the press, Cuomo cited New York’s relative success to other states: “You can see what you want to see in the numbers, but the numbers are the numbers.”

He continued that “as a percentage of nursing home deaths” New York was “all the way at the bottom of the list of states.”

As Cuomo’s administration comes under federal investigation for an alleged cover-up of COVID-19’s death toll in nursing homes and assisted living facilities, new federal data enables a more accurate estimate of deaths in these vulnerable populations.

Instead of ranking among the best states in protecting long-term care (LTC) residents, as Cuomo asserted, our analysis over several months shows that New York is among the poorest-performing states, with more than 12% of its long-term care population succumbing to the virus. Using data from the Centers for Medicare and Medicaid Services, as well as New York state data released by court order in fulfillment of numerous Freedom of Information Law (FOIL) requests, including from FREOPP, our estimates now indicate that 1,242 LTC residents per 10,000 have died of coronavirus in New York as of Jan. 31, compared to 740 per 10,000 if one excludes long-term care residents who died in hospitals, as New York officials did.

COVID-19 fatalities both inside nursing homes and outside (in hospitals) have closely mirrored each other throughout the pandemic. Generally, nursing home deaths in hospitals have made up 40–45% of COVID deaths among long-term care residents. (Graphic: G. Girvan / FREOPP)

After including deaths of LTC residents in the hospital setting, New York’s ranking among the 50 states and D.C. changes from No. 20 to No. 5 for the highest per-capita long-term care death rate. Only Rhode Island (1,311 per 10,000 long-term care residents), New Jersey (1,358), Massachusetts (1,512), and Connecticut (1,847) have fared worse.

What we now know about New York nursing home deaths

On Jan. 28, the New York Attorney General, Letitia James, released a report on the impact of COVID in the state’s nursing homes. In a sample of 62 nursing homes, the Attorney General found 1,914 total facility deaths, as compared to 1,229 reported by the state’s Department of Health. In other words, in the sample, COVID-19 fatalities were 56% greater than what the Cuomo administration had reported.

As a result of the AG’s report and continuing pressure from advocates, the Department of Health began releasing additional details, including nursing home fatalities among residents who died in a hospital. In total, New York’s official count of COVID fatalities in nursing homes increased from about 9,000 deaths to over 13,000. Just days later, the state released a full facility-level dataset to the Empire Center for Public Policy, a New York-based think tank that made the initial FOIL request six months prior. By Feb. 7, the state also released data on the number of residents in other LTC facilities, including assisted living facilities, increasing the number of LTC residents dying from COVID-19 by more than 1,500. The state’s Department of Health also acknowledged it was still reviewing more than 600 death records from hospitals that constituted “presumed” COVID deaths (with 60 other presumed deaths ruled not COVID related).

There were early clues of the magnitude of the state’s undercount earlier in the pandemic. Beginning on May 8, 2020, the Centers for Medicare and Medicaid Services required all of the nation’s licensed nursing homes (those that are permitted to accept Medicare and/or Medicaid payments) to report data related to COVID. These data include weekly and cumulative cases and fatalities, as well as each nursing home’s stores of personal protective equipment.

We analyzed the CMS dataset when it was first released on May 26. However, the data contained numerous errors — including nursing homes reporting more COVID deaths than beds— as nursing homes adjusted to the new data requirements imposed by CMS.

(Importantly, the CMS data only examines nursing homes, also known as skilled nursing facilities (SNFs). Nursing homes include 24-hour supervision, nursing care, three meals a day, and assistance with activities of daily living. Nursing home residents are usually people with long-term physical, medical, or mental conditions requiring 24-hour supervision, but can also include patients recently discharged from a hospital who need such care temporarily. Our broader analysis of long-term care facility deaths also includes residents of assisted living facilities, sometimes called residential care homes or personal care homes, which are similar to nursing homes but are tailored to seniors who don’t require full-time medical care on-site.)

In the meantime, we collected data from individual states to assess how each was responding to the crisis. When we released our first report on nursing home and assisted living facility deaths on May 7, 2020, the analysis found 5,003 confirmed and presumed deaths in nursing and assisted living facilities in New York, or 5.2% of the state’s LTC population. Compared to surrounding states, New York’s figures were an anomaly — New Jersey reported 11.4% of LTC residents died, while Massachusetts and Connecticut reported 6.9 % and 6.6% fatality rates, respectively.

Further, the number of LTC residents who died as a percentage of all COVID deaths was abnormally low in New York (20%). Again, for the same northeastern states of New Jersey, Massachusetts, and Connecticut, 53.3%, 60.0%, and 55.1% of state residents who died were LTC residents, respectively. And if New York only included confirmed COVID deaths, as some states had done, New York’s reported COVID deaths among nursing home residents would have been 2.6% — less than one-fourth the rate of New Jersey.

Because New York’s data was of questionable integrity, our analysis included a measure of the overall impact of COVID across the nation’s nursing homes while intentionally excluding New York state’s numbers. For example, in the first version of the report, we calculated that 40% of all COVID deaths in the U.S. occurred in LTC facilities; but the percentage increased to over 49% when excluding New York.

Analysis of the CMS nursing home fatality dataset

To estimate New York’s nursing home residents who died of COVID in hospitals, we merged the state’s facility-level dataset with the CMS dataset. We then calculated the difference in cumulative deaths for each facility. For each nursing home where CMS reported a higher number of deaths, the most likely reason was that the nursing home was reporting hospital deaths as well as on-site deaths to CMS, while New York state was only reporting on-site deaths.

In these analyses, the CMS data had one major weakness: CMS did not require nursing homes to report data prior to May 8, 2020, the implementation date of the federal data gathering requirement. Nor did the dataset indicate which nursing homes chose to report to the beginning of the pandemic. New York State, on the other hand, reported deaths at all nursing homes back to March 1, 2020. Therefore, we could not directly estimate hospital deaths at nursing homes where a lower number of deaths were reported to CMS than to New York state.

In addition, some nursing homes did not report data in certain weeks, while others made corrections to previous weekly data; all of which caused fluctuations in our undercount estimate. Therefore, it was understood that the calculations represented a ceiling on the potential undercount in New York nursing homes.

Our first analysis comparing the New York state data to CMS data was completed in early December, with data through Nov. 29. Sampling only those nursing homes where the number of deaths reported to CMS was greater than the state of New York, and then applying the same undercount percentage to all nursing homes, we estimated that New York was undercounting nursing home fatalities by as much as 66.6% (4,529 hospital deaths in addition to a reported 6,801 deaths). We continued to evaluate the data and refine our methods over the next two months, including analyses run for data through Dec. 20 and Jan. 17, with estimated undercounts of 4,803 deaths (63.5% undercount) and 4,678 deaths (55.5% undercount), respectively.

Because of CMS’s data limitations, we performed one more analysis for data released by both New York and CMS through Jan. 31, with one adjustment; we calculated fatality data from May 8 onward, the earliest date when all nursing homes began reporting data to CMS. After May 8, the CMS data reported 863 more deaths than the 2,480 reported by New York state; an undercount of 35%. This represented the minimum undercount in our estimates. As with prior analyses, we calculated the maximum undercount percentage of 58% based on nursing homes that had higher death counts in the CMS dataset.

Based on this range of undercount percentages applied to all nursing homes, we estimated that the death count occurring among New York nursing home residents in the hospital range from 3,139 to 5,190, as of January 31, 2021. Taking into account new data released from the state of New York, including the number of presumed hospital deaths among nursing home residents under review, New York reported the number of hospital deaths in nursing homes at 4,713, within 10% of the maximum end of our range (5,190).

Since CMS only collects data on nursing homes, a comparison of assisted living and other adult care facilities was not possible. But New York state data released in early February on assisted living and other LTC facilities revealed 1,714 deaths overall, raising New York’s total to 15,448 deaths in all LTC facilities combined as of Jan. 31, 2021. This number is an undercount of 68%, representing 35% of all COVID deaths in the state and 1,242 deaths per 10,000 LTC residents.

While New York was not counting the number of long-term care residents who died of COVID-19 in hospitals, the state appeared to be outperforming other states along the Acela Corridor. But after accounting for such residents who died in hospitals, New York experienced nursing home and assisted living fatalities comparable to states such as Massachusetts, New Jersey, and Rhode Island, who were among the hardest hit. This is in part due to policy decisions by those states that discharged seniors with active COVID-19 infections from hospitals to LTC facilities. (Graphics: G. Girvan / FREOPP)

New York’s statistical maneuverings

Throughout the pandemic, the Cuomo administration has argued that New York has performed better than most any other state in protecting nursing home residents. As justification, Gov. Cuomo has cited the state’s low percentage of COVID-19 deaths in nursing homes relative to all COVID-19 deaths in the state.

Cuomo’s presentation has obscured the state’s true performance for two reasons. The first is that, as we have discussed, the Cuomo administration did not include COVID-19 deaths of long-term care residents who were hospitalized at the time of death.

The second reason is that Gov. Cuomo is using the wrong denominator. Measuring nursing home deaths as a percentage of all state-wide COVID-19 deaths is illusory. For example, nursing home deaths as a percentage of overall deaths can be low if the number of overall deaths is high: that is to say, a reflection of poor performance overall. A more accurate measure is the percentage of statewide nursing home or long-term care residents who died of COVID-19.

To illustrate, consider two states with roughly the same percentage of COVID deaths occurring in LTC facilities, Maine (58.6%) and Massachusetts (56.6%). At first glance, it appears both states are performing similarly. But when measuring each state based on per capita death rates, Maine performs far better: 279 deaths per 10,000 LTC residents, versus 1,512 in Massachusetts.

Georgia and New Hampshire provide another stark comparison. New Hampshire’s percentage of statewide deaths occurring in LTC facilities (73.8%) is almost three times higher than Georgia’s (28.6%). Yet each state’s deaths per capita in LTC facilities is nearly identical (725 deaths per 10,000 LTC residents in Georgia versus 720 in New Hampshire).

After excluding outlier states, there is virtually no relationship between the percentage of COVID-19 deaths occurring in LTC facilities and the per capita death rate of LTC residents.

The impact of COVID in long-term care facilities varies widely by state. Moreover, the relationship between the percentage of deaths occurring in LTC facilities and the fatality rate among LTC residents is murky; for instance, many states with fatality rates around 700 LTC residents per 10,000 nevertheless vary between 16–74% of statewide COVID deaths occurring within such facilities. (Graphic: G. Girvan / FREOPP)

By focusing on the percentage of state residents dying in nursing homes, the Cuomo administration ignored other metrics that demonstrated the state’s high nursing home fatality rate in contrast to other states. New York’s true death rate in nursing homes is 12.4%, #6 for the worst death rate in the country. Prior to New York’s release of data on LTC residents who died in the hospital, the state reported 9,212 deaths in LTC facilities, or 21.1% of the state’s COVID deaths. That percentage ranked near the bottom (#47) of all 50 states plus the District of Columbia.

Thus, the unusual statistical approach employed by New York officials created the false impression that the state was one of the best performing in the country on nursing home COVID-19 deaths, when in fact it was one of the worst.

The importance of accurate data

New York is not alone in reporting incomplete data on COVID in nursing homes and other adult care facilities. For example, Missouri refuses to report any fatality data at all, whether on its own website or to the media. However, New York’s mishandling of COVID data differs from any other state in one key aspect; the Cuomo administration repeatedly overstated its performance in nursing homes based on data and analyses that the state knew were incomplete and misleading.

In a private conference call to state Democratic lawmakers, Cuomo’s top aide, Melissa DeRosa, claimed that the administration delayed release of complete data on COVID in nursing homes out of fear of an investigation from the U.S. Justice Department. However, J. David Goodman and Danny Hakim of the New York Times discovered that senior Cuomo administration officials, including DeRosa, rewrote a New York State Health Department report in June to exclude LTC residents who died in hospitals, months before federal officials requested the state’s data.

More importantly, New York officials missed a crucial opportunity to provide better data that could have saved lives. Accurate data on the toll of COVID in New York nursing homes and other LTC facilities would have enabled policymakers to target resources to facilities struggling to contain infections. In addition, family caregivers needed accurate information to know where to place their loved ones who are in need of nursing home care.

New York’s politicization of nursing home data serves as a cautionary tale. Reliable data is essential for state and local officials to make good decisions on containment measures, vaccine distribution, and business and school reopenings. It is not yet possible to estimate how many people died because the state concealed how serious the pandemic was in long-term care facilities. But the number is well above zero.