top of page
Writer's picturePaul Francis

The Dog That Didn’t Bark: the House Select Subcommittee on the Coronavirus Pandemic

Commentary # 16 by Paul Francis

September 17, 2024


 
 

This Commentary builds on two previous Commentaries – “Re-Examining the Cuomo Administration’s Nursing Home Policies During Covid-19” and “Debunking the Empire Center Analysis Attributing Nursing Home Covid-19 Deaths to DOH Policy” – regarding the controversy about the Cuomo administration’s nursing home policies during Covid-19.


Introduction


“The dog that didn’t bark,” of course, refers to a situation where something becomes notable by its absence or silence rather than its presence. In the Staff Memorandum from the Select Subcommittee on the Coronavirus Pandemic Majority Staff, of September 9, 2024 (the “Staff Memo”), regarding “Findings from the Select Subcommittee’s Investigation into the Cuomo Administration’s March 25 Directive admitting COVID-positive patients into Nursing Homes,” (the “Staff Memo”), the dog that didn’t bark was all the things the Subcommittee Majority Staff did not say or find in its 48-page Staff Memo.  


The Staff Memo was released in conjunction with the testimony, on September 10, 2024, of former Gov. Andrew Cuomo before the U.S. House Select Subcommittee on the coronavirus pandemic (the “Subcommittee”). The hearing itself was pure theater and provided little insight into the events surrounding the nursing home controversy.


However, to the House Select Subcommittee’s considerable credit, the Staff Memo was accompanied by the release of transcribed interviews of Gov. Cuomo and the nine former New York State employees who were closely involved in the administration’s Covid-19 response (the “Transcribed Interviews”). The accompanying press release also noted that the Subcommittee Majority Staff reviewed nearly 550,000 pages of documents in connection with its work, although only one document was released in connection with the Staff Memo.  


The questioning in the Transcribed Interviews was almost entirely conducted by Committee staff, and – in stark contrast to the approach of Subcommittee members during the hearing – the staff allowed the interviewees to answer the questions. The Transcribed Interviews provide a rare opportunity to develop a first-hand understanding of the facts in this controversy. Reading the ten Transcribed Interviews, in combination with the other parts of the public record I reviewed in my whitepaper titled Re-Examining the Cuomo Administration’s Nursing Home Policies During Covid-19 (“Re-examining Policies”), provides a fairly clear picture of what actually happened as the events related to the nursing home controversy unfolded in the year following the pandemic’s arrival in New York in March 2020.


Such an understanding of the facts did not, however, deter the Subcommittee Staff Majority, which was responsible for the preparation of the Staff Memo, nor did it deter him or (especially) the Subcommittee members from grossly misconstruing the record and mischaracterizing the facts. This was true for all of the Republican Subcommittee members and a few of the Democratic members as well.


The Staff Memo stated on its second page that the DOH March 25 Advisory, which established criteria for admissions by nursing homes of Covid-19 patients treated in a hospital, “caus[ed] predictable but disastrous consequences.”[1] Every Republican Subcommittee member took this statement as a given. Several Republican members even went so far as to say that the March 25 Advisory was the direct cause of 15,000 nursing home deaths in New York. Yet the Staff Memo made no Finding about the extent of additional deaths of nursing home residents that could empirically be associated with the March 25 Advisory. Presumably, this was because, as I explain later in this Commentary, the estimated number would undermine the narrative that the March 25 Advisory was significantly responsible for deaths of nursing home residents in New York.


Instead of focusing on the substance of the issues, the Staff Memo, as well as much of the Subcommittee members’ questioning, focused on the process of decision-making surrounding the two main issues of controversy – i.e., the March 25 Advisory and the decision not to include out-of-facility deaths of nursing home residents under “nursing home deaths” until February 2021. As for the substance, a review of ten Transcribed Interviews and the one new document released by the Subcommittee not only corroborates but strengthens the observations and conclusions of my Re-examining Policies paper.


In Part I of this Commentary, I address what the Staff Memo did say in its “Findings” – and what it did not say – with respect to these issues. In doing so, I’m going to follow the format of my Re-examining Policies paper. In Part II of this Commentary, I’m going to respond to a rebuttal from the Empire Center regarding my paper Debunking the Empire Center Analysis Attributing Nursing Home Covid-19 Deaths to DOH Policy. It is worth noting upfront that the Empire Center Rebuttal significantly reduced the number of nursing home deaths it argues were associated with the March 25 Advisory.


Part I – The House Select Subcommittee on the Coronavirus Pandemic


What was the Genesis of the March 25 Advisory?

A central piece of the narrative against the Cuomo administration is that the March 25 Advisory originated as a directive from the Executive Chamber to the Department of Health (DOH). The implication of the narrative is that the March 25 Advisory was a politically motivated decision not supported by the public health professionals at DOH. The weight of the evidence suggests that the March 25 Advisory was developed organically within DOH and there is no evidence to suggest that the DOH public health professionals did not support the policy.


The Staff Memo made Findings that the Executive Chamber was “involved in the decision that led to the March 25 Advisory” and “approved” the March 25 advisory. The first Finding is based entirely on the recollection of the former DOH Commissioner, Dr. Howard Zucker, of a phone call from “the Greater New York Hospital Association (GNYHA) [to] the governor and the team.”[2] According to Dr. Zucker, GNYHA). Dr. Zucker testified that in the call, GNYHA said “we have individuals who are better, they have recovered, and they are just sitting in a hospital bed but need to go ‘home’,” whether that was a long-term care facility or a private residence.[3] It was Dr. Zucker’s understanding that this phone call led to the March 25 Advisory.


However, no one else in the Transcribed Interviews shares Dr. Zucker’s recollection of such a phone call with GNYHA. Moreover, despite the Transcribed Interviews and the Subcommittee Majority Staff reviewing 550,000 pages of documents, the Staff Memo did not identify how or through whom this directive from the Executive Chamber was communicated to DOH.


Brad Hutton, the Deputy Commissioner of Public Health at DOH at the beginning of the Covid-19 pandemic and the only person interviewed by the Subcommittee who expressed that he or she had any role in the development of the March 25 Advisory, testified that he recalled the direct catalyst for the March 25 Advisory being a conversation he had with the DOH employee who was primarily responsible for nursing homes at this stage of the pandemic. That employee recounted receiving a call from the administrator of a hospital in the Hudson Valley about the need for guidance regarding nursing homes admitting Covid-19 patients in transfers from hospitals. (I generally use “transfers” in this paper to refer to admissions or readmissions of Covid-19 patients from hospitals.)


With respect to both the Finding that the Executive Chamber was involved in the development of the March 25 Advisory and the Finding that the Executive Chamber approved the March 25 Advisory, the Transcribed Interviews make clear that Gov. Cuomo and Melissa DeRosa were surprised by the existence of the March 25 Advisory when it was raised by a reporter in a press conference on April 20, 2020.


Linda Lacewell, a member of the Executive Chamber Covid Task Force (the “Task Force”) who was designated as a “counsel,” noted that Gov. Cuomo and Melissa DeRosa were unaware of the March 25 Advisory when the subject was brought up at the April 20, 2020, press conference. Ms. Lacewell cited lawyer-client privilege in a number of aspects of her testimony, including with respect to a purported conversation with Task Force member Larry Schwartz about the March 25 Advisory. However, as the Staff Memo noted, Larry Schwartz testified that he was unaware of the March 25 Advisory and has no recollection of a conversation about it with Ms. Lacewell.


Special Counsel to the governor, Beth Garvey, testified that she (not the top aide to the governor, Melissa DeRosa) would have been the final Executive Chamber approval for agency guidance until sometime in April 2020. Ms. Garvey said that she did not recall the March 25 Advisory but speculated that it was reviewed by a member of her staff before being formally approved by Ms. Garvey.


Given the later significance of the March 25 Advisory, the fact that none of the Transcribed Interviewees other than Dr. Zucker and Brad Hutton were aware of the March 25 Advisory before the April 20, 2020, press conference, has seemed suspicious to critics of the Cuomo administration. However, as I said in my Re-examining Policies paper, the March 25 Advisory is best understood as part of a logical progression of public health responses to the pandemic, rather than an isolated initiative.


As for the origins of the March 25 Advisory, I have great respect for Dr. Zucker and know he is a person of the highest integrity. I’m sure his testimony was based on his best recollection, as was the lack of recollection by others. Memories can vary in the fog of war. Perhaps some new fact will emerge on this topic that will be dispositive.


But in any event, the most important point in this part of the narrative against the Cuomo administration is that there is no debate about the central issue of whether the March 25 Advisory was a directive from the Executive Chamber that overrode the concerns of the public health professionals at DOH. Dr. Zucker and Brad Hutton both made clear that they strongly supported the policy of the March 25 Advisory and were not concerned about the wording of the March 25 Advisory, since they believed it should be interpreted in the context of other State and federal guidance regarding transmission protocols.


Did the March 25 Advisory Compel Nursing Homes to Accept Covid-19 Patients?

The second major question I asked in my Re-examining Policies paper was whether the March 25 Advisory required nursing homes to accept transfers from hospitals under all circumstances. The Staff Memo made a Finding that the March 25 Advisory “was not consistent with applicable federal guidance regarding hospital to nursing home transfers and COVID-19 related infection control,” because the March 25 Advisory did not follow the CMS and CDC guidance in explicitly including language that conditioned the admission of a Covid-19 patient from the hospital on specific transmission protocols being observed.


Although much has been made of the semantic difference, Gov. Cuomo, Dr. Zucker, and Beth Garvey, the former Special Counsel to the Governor, strongly argued that compliance with transmission protocols was implied in i.e., understood to be applicable to – the March 25 Advisory, because such compliance was required by both existing New York law and regulations, as well as by previous DOH transmission protocol guidance.


Dr. Zucker and Brad Hutton emphasized in their testimony that the March 25 Advisory was not intended to compel nursing homes to accept transfers of Covid-19 patients from hospitals if they could not observe transmission protocols and adequately care for residents. In this context, it is relevant to note that Dr. Zucker believed, as he expressed in his Transcribed Interview, that these patients were no longer contagious given the length of time between infection and hospital discharge.


Nevertheless, as I pointed out in my Re-examining Policies paper, although it was not the intention of DOH to require nursing homes to accept transfers of Covid-19 patients under all circumstances, some nursing homes interpreted the March 25 Advisory as a mandate, notwithstanding DOH’s intention. That leads to the question of what impact such admissions may have had.


To What Extent Were Transfers Under the March 25 Advisory Associated with Nursing Home Deaths?

The significance of whether nursing homes were required to accept Covid-19 patients, as well as the importance of the semantic distinction between the March 25 Advisory and CMS and CDC guidance, is whether admission of patients pursuant to the March 25 Advisory led to increased nursing home deaths.  If New York nursing homes widely accepted contagious Covid-19 patients from hospitals who were contagious and for whom they could not follow transmission protocols of the type described in guidance from DOH, CMS, and CDC, then you would expect to see a very substantial number of nursing home resident deaths associated with the 8,579 transfers of Covid-19 patients from hospitals to nursing homes that occurred during the timeframe in which the March 25 Advisory was in effect.


Despite saying that the March 25 Advisory had “caused predictable but disastrous consequences,”[4] the Staff Memo made no Findings about what those “disastrous consequences” were. The Staff Memo Republican Subcommittee members simply took it as a given that the March 25 Advisory was significantly responsible for the deaths of nursing home residents, without citing any empirical evidence for the proposition. Simply put, the dog did not bark.


Instead of presenting empirical evidence of the connection between admissions of Covid-19 patients and nursing home deaths, the Staff Memo simply inferred that the March 25 Advisory must have led to "disastrous consequences" because Gov. Cuomo made edits to a July 2020 DOH report (the “DOH Factors Report”)[5] that made the report more definitive about the lack of connection between transfers and deaths of nursing home residents. Although, as I discuss in some detail below, there were a number of issues associated with the DOH Factors Report, these issues did not bear on the conclusion that nursing home deaths were not significantly affected by admissions of Covid-19 patients.


With respect to the empirical evidence about the extent to which March 25 Advisory was associated with additional nursing home deaths, I have been engaging in an analytical debate with Bill Hammond and Ian Kingsbury of the Empire Center about their conclusion that the March 25 Advisory was “associated with several hundred and possibly more than 1,000 additional resident deaths” statewide – an estimate they have since revised downward. To their credit, the Empire Center analysis included important caveats about the limitations of their analysis. Unfortunately, the tenor of the Staff Memo, and especially the questioning by Subcommittee Republicans, did not reflect the cautiousness of the Empire Center’s limited conclusions.


Bill and Ian have challenged my critique of their conclusions in my paper Debunking the Empire Center Analysis Attributing Nursing Home Covid-19 Deaths to DOH Policy. I address their rebuttal in Part II of this Commentary.


Did the July 6 DOH Factors Report Make Misstatements About the Impact of the March 25 Advisory?

The Staff Memo made several process-related Findings regarding the July 6, 2020, DOH Factors Report, including that Gov. Cuomo directed that the report be prepared, that the report was not “independently prepared by DOH and not peer-reviewed,” and that Gov. Cuomo edited the report “to make the Report’s findings more causal.” In addition, the Staff Memo found that the Executive Chamber made the decision not to include out-of-facility nursing home deaths in the report.


In my Re-examining Policies paper, I extensively discussed the substance of the DOH Factors Report. I also address the decision not to report out-of-facility nursing home deaths in the DOH Factors Report, an issue that is clarified in the Transcribed Interviews, is an issue that continued after the July DOH Factors Report and which I will address in a moment.


The first two of the Staff Memo Findings about the DOH Factors Report have never been in dispute. It is common practice for the Executive Chamber to request reports from agencies and to make comments on (i.e., edit) the final product. Given the importance of the issue of the cause of nursing home deaths, it is not surprising that the Executive Chamber would be heavily involved in the process.


In this case, primary responsibility for the task was given to Jim Malatras, a former Director of State Operations and Policy Director under Gov. Cuomo who became a member of the Task Force. Mr. Malatras worked extensively with others in transforming a paper about the sources of nursing home infections that was being developed within DOH for publication in an academic journal into a more accessible and less technical government report.


Beyond the issue of whether to include out-of-facility nursing home deaths, the most important question about the DOH Factors Report is whether the involvement of the Executive Chamber altered the substance of the DOH Factors Report with respect to the strength of its conclusions about the role of admissions of Covid-19 patients in causing additional nursing home deaths.


The Transcribed Interview with Dr. Eleanor Adams, the DOH staff member who was most involved in the earlier DOH documents that led to the DOH Factors Report and who communicated extensively with Mr. Malatras, as well as the testimony of Dr. Zucker, indicate there was not a difference in point of view about the conclusions of the DOH Factors Report, although there was a nuanced difference in language that should be addressed.


Moreover, as I wrote in the Re-Examining Policies paper, tensions that had developed between DOH staff and the Executive Chamber during the response to the pandemic carried over into the preparation of the final DOH Factors Report. As discussed below, the Transcribed Interviews also reflect different and strongly held views within the Task Force about whether to include out-of-facility deaths in the DOH Factors Report – but not about the central conclusion of the report.


Interestingly, the one new document made public by the Subcommittee in connection with the release of the Staff Memo is particularly relevant to this issue of the influence of the Executive Chamber on the conclusions of the DOH Factors Report. The document was an email from Dr. Adams to Dr. Zucker, dated June 7, 2024, before any Executive Chamber editing of what became the DOH Factors Report had begun. Dr. Adams, an epidemiologist who many of the Transcribed Interviewees went out of their way to praise, was providing talking points for Dr. Zucker. The email states in several places that the data “shows that readmission/admissions of residents with COVID-19 to nursing homes most likely was not a driver of the number of deaths in nursing homes.”[6] (Emphasis added)


The final DOH Factors Report uses language that is more direct than the more careful analytical language that Dr. Adams used in her email to Dr. Zucker. The DOH Factors Report said, “an analysis of the timing of admissions versus fatalities shows that it could not be the driver of nursing home infections or fatalities”[7] (emphasis added); and that “the data does not support [the] assertion” that “the admission of COVID-positive residents introduced COVID into nursing homes.”[8] 


While I acknowledge the nuanced difference in language between the email and the DOH Factors Report, I don’t believe the different formulations reflect a material difference in the substantive conclusion of the DOH Factors Report. The conclusion was that transmission from staff members was the primary source of Covid-19 infections in nursing homes. In combination with two other factors that were described in the report, this conclusion about the source of infections led to the second main conclusion of the DOH Factors Report, which was the lack of a connection between transfers of Covid-19 patients and additional deaths of nursing home residents.


As discussed in my Re-Examining Policies paper, these two factors were evidence that all but six nursing homes that receive transfers already had experienced a staff or resident Covid-19 infection; and the length of time between infection and discharge, which made it very unlikely that the patients transferred were still contagious.


Because of the importance the Subcommittee and others have put on nuanced distinctions in language about the relationship between the March 25 Advisory and nursing home deaths, it is unfortunate that the Hochul administration has not released the earlier DOH documents that formed the basis for the DOH Factors Report. A review of those earlier DOH documents would improve our ability to understand the extent of any differences in language and conclusions from those earlier documents and the final DOH Factors Report.


Why did the DOH Factors Report Not Include Out-of-Facility Nursing Home Deaths?

The Staff Memo made a Finding that the Executive Chamber, rather than DOH, was responsible for the decision not to include the out-of-facility nursing home death number in the DOH Factors Report. The Staff Memo’s Finding focused exclusively on the process by which the decision was made (i.e., by the Executive Chamber as opposed to DOH), rather than the importance and consequences of the decision itself.


While the Assembly Impeachment Investigation Report highlighted tensions between DOH and the Executive Chamber about whether to include out-of-facility nursing home deaths in the DOH Factors Report, the Transcribed Interviews offer a good deal of transparency about the internal disagreement within the Task Force on the subject. Jim Malatras strongly favored the inclusion of the out-of-facility nursing home death numbers in the belief that it would resolve the controversy over the issue, while Melissa DeRosa continued to question the validity of the numbers and the wisdom of putting out numbers that likely were not fully accurate. Dr. Zucker acknowledged making the final decision regarding whether to include out-of-facility deaths in the DOH Factors Report. He believed it was not necessary to include out-of-facility nursing home death data, because it did not affect the central conclusion of the report that the Covid-19 infection was introduced by staff.


Government officials frequently disagree over whether and when to release certain information. The real question in this case, of course, is why the decision was made not to include out-of-facility nursing home deaths at the time of the DOH Factors Report or even until February 2021. The Transcribed Interviews provide a much clearer picture of the different points of view on this issue, including the extent of uncertainty about the validity of the out-of-facility nursing home death data at the time of the DOH Factors Report and the subsequent legislative testimony by Dr. Zucker on August 3, 2020.


After the DOH Factors Report was released, Melissa DeRosa charged Gareth Rhodes, a member of the Task Force, to conduct an internal review of the “data integrity” of the data concerning out-of-facility deaths, which had been reported to DOH by nursing homes and in many cases had involved subsequent manual efforts at reconciliation with the hospitals. Gareth Rhodes described his efforts as “a common sense view of a data set.” [9] He was not tasked with reconciling the data with nursing homes and hospitals but rather went through the data with DOH staff “line by line [to] make sure there were no discrepancies or any inconsistencies.”[10]


One of the revelations in the Transcribed Interviews (which is reported in the Staff Memo) was that this review by Gareth Rhodes, which was not completed until late August 2020, found “maybe 600 inconsistencies” in the data out of roughly 3,500 cases that could be followed up on to assure accuracy.[11] It is not clear how many of those “inconsistencies” changed the ultimate count of out-of-facility nursing home deaths, but it does refute the notion that this was an easily determined statistic that the Cuomo administration was sitting on for political purposes.


I still believe, as I wrote in the Re-examining Policies paper, that:

“In hindsight, given the damage caused by the perception that New York’s convention of reporting nursing home deaths was a cover-up…the Cuomo administration clearly would have been wiser to provide an estimate of the number of nursing home residents who died in hospitals with the appropriate caveats about data limitations.”[12]

That said, the information in the Transcribed Interviews gives me a much better appreciation for why the most senior New York State officials were reluctant to release this data. Information in the Transcribed Interviews that bears on this point includes the following:


  • First, the decision to change the methodology in early May 2020 to exclude out-of-facility deaths was made because of the extreme unreliability of such data at that time.

  • Second, the fact of “maybe 600 inconsistencies” in the out-of-facility deaths data suggests that there were reasons other than political calculation for not disclosing the estimate of out-of-facility deaths in the July DOH Factors Report or the August testimony by Dr. Zucker to the legislature.

  • Third, Gov. Cuomo and Melissa DeRosa were clearly concerned in the fall of 2020 that the DOJ investigation made releasing an imprecise number legally fraught.

  • Fourth, Melissa DeRosa’s Transcribed Interview makes clearer than I had appreciated the dynamic of responding to the DOJ information request about nursing home deaths before fulfilling the legislative information request.

  • Fifth, I had not appreciated that the Attorney General’s report in January 2021, which estimated the number of out-of-facility nursing home deaths, intentionally preempted – rather than prompted – the disclosure by DOH of the full death total of nursing home residents, including confirmed and presumed nursing home deaths in the facility, as well as out-of-facility deaths.

  • And sixth, although it likely played a smaller part in the decision, President Trump had criticized New York for releasing probable or “presumed” nursing home deaths – a number about 2/3 as large as out-of-facility deaths – in May 2020 on the basis that it overstated the severity of the pandemic.


One can argue whether the administration’s decision was an excessive prioritization of data integrity and precision over an approach of reporting the most accurate available information with data caveats. But the assumption of so many critics of the Cuomo administration that the only reasonable interpretation of the decision to not report the out-of-facility nursing home death information sooner was a self-serving cover-up is much more confounded than supported by the accounts of the ten individuals most closely involved in the Covid-19 response in their Transcribed Interviews. Moreover, the Staff Memo provided no new evidence from its review of 550,000 pages of documents to support the cover-up narrative.[13]                       

                                     

Part II – Responding to the Critique from Bill Hammond and Ian Kingsbury of My Commentary: Debunking the Empire Center Analysis Attributing Nursing Home Covid-19 Deaths to DOH Policy


On September 11, 2024, Bill Hammond and Ian Kingsbury of the Empire Center posted a response to my commentary, Debunking the Empire Center Analysis Attributing Nursing Home Covid-19 Deaths to DOH Policy. Their response, titled What Paul Francis Got Wrong About the Empire Center’s Nursing Home Research (the “Empire Center Rebuttal”), asserted that my Commentary “was marred by significant factual errors and analytical mistakes.” Readers can judge that assertion for themselves after reviewing my reply here.


Two mistakes I acknowledge making are that I assumed that the Empire Center was using a more standard definition of Upstate as being comprised of the 54 counties north and west of Westchester. Instead, the Empire Center included another six counties in their definition of the Downstate region (including Sullivan County and Ulster County, which are not usually thought of as being Downstate), which corresponds to the counties covered by DOH’s Metropolitan Area Regional Office. I also stated that the number of Upstate nursing homes that received an admission of a Covid-19 patient was 27, when the actual number appears to be 45.[14] Neither of these adjustments changes the basis of my critique of the Empire Center correlation analysis.


As I have said before, I give Bill and Ian (to whom I will refer collectively as the Empire Center) a lot of credit for sharing with me the nursing home transfer information that was released by DOH pursuant to a FOIL request and which is not publicly available. I also give them credit for revising their regression analysis in the Empire Center Rebuttal based on my identification of three Upstate nursing homes with Covid-only units or facilities to reduce the number of additional nursing home deaths that their analysis suggests were associated with the March 25 Advisory.


I also noticed (since it was included in the one document released by the Subcommittee) that 55 deaths of nursing home residents occurred at the Elderwood at Amherst nursing home in Erie County. This was such a large number that I Googled that nursing home and found that it also had a fully separate Covid-only unit for residents and staff.[15] Presumably, this would result in a further downward revision of the Empire Center’s analysis of the impact of the March 25 Advisory in Upstate nursing homes.


Adjusting their regression model as described above had a significant impact on the number of deaths of nursing home residents associated with Covid-19 admissions. The Empire Center Rebuttal states:

“Statewide, analysis of the revised data set would indicate that each admission was associated with .07 additional deaths, plus or minus .02. Upstate, each admission was associated [with] .30 additional deaths, plus or minus .09.
“The statewide results of this alternative analysis — limited to exclude the three identified Covid-only nursing homes — would suggest the policy was associated with hundreds of additional deaths.”[16]

After maintaining for more than three years that the number of nursing home deaths statewide associated with transfers of Covid-19 patients from hospitals was “possibly more than 1,000,”[17] the Empire Center Rebuttal is now saying that statewide, the associated number of nursing home deaths is 442 and the associated number of nursing home deaths Upstate (the only region where there is any statistically significant correlation) is 98. The math is shown in the footnote.[18] I believe these estimates would be reduced further if adjusted for transfers to Covid-only units or facilities such as Elderwood at Amherst.


Especially if we are talking only about Upstate nursing homes, it’s difficult to see how the Empire Center’s conclusion that 98 deaths of nursing home residents out of 9,110 such deaths in the relevant timeframe (~1%) contradicts the statement in the DOH Factors Report that, “Admission policies were not a significant factor in nursing home fatalities.” 


The somewhat more difficult question, however, relates to the Empire Center’s presentation of any statewide number based on a regression analysis. My position, in short, is that because of the very different characteristics of the relevant data between Downstate and Upstate nursing homes, it is inappropriate to use a data set comprised of the Downstate and Upstate data to extrapolate a statewide total.


I should point out that the Empire Center Rebuttal is wrong in stating that I did not recognize that the Empire Center study ran a statewide regression analysis. Instead, I agreed with the conclusion of the data and policy analyst, John Bacheller, that a statewide regression analysis was not methodologically reliable or appropriate, for the reasons discussed in more detail below.[19] 


The issue is not whether the Empire Center study made a mathematical error. Instead, I believe the Empire Center study made a methodological error in presenting a statewide regression analysis – as opposed to two separate regional analyses – as being reliable, given the lack of correlation in the larger Downstate region and the vast differences in the way the Upstate and Downstate regions experienced the Covid-19 pandemic. As with any academic debate, different experts may reach different conclusions. But I think both common sense and the weight of academic literature support my view.


In addressing the appropriateness of a statewide regression analysis, as opposed to two regional analyses, I think it is helpful to again review the critique of the original Empire Center Study by John Bacheller. In addition, I address the most recent Empire Center post on documents released by the House Select Subcommittee. Finally, for those who want to have a better understanding of the technical aspects of the methodological issue, I include a discussion in the Appendix of this paper.


John Bacheller’s Critique of a Statewide Analysis

Before getting into a more technical explanation of why I do not believe a statewide regression analysis is valid, let me again present the relatively accessible analysis presented by John Bacheller.


To begin with, Bacheller arrived at the same conclusion as the Empire Center that there was no statistically significant correlation between the March 25 Advisory and deaths of nursing home residents in the Downstate region. He wrote:

“Given that the strength of the association between COVID-positive admissions is so weak — only one percent of deaths are associated with COVID-positive admissions — and that the relationship was not statistically significant, I conclude that the Health Department mandate that nursing homes accept COVID-positive residents did not result in more deaths in the New York metropolitan area.”[20]

Bacheller then found that because the transmission patterns between the Upstate and Downstate regions were so different, the statewide estimate was “unlikely” to be accurate. He explained:

“The regional models point to different transmission patterns in each region. In the downstate area, where more than 90% of COVID-related nursing home deaths occurred, the data does not show an association between the admission of COVID-positive new nursing home residents and COVID-related deaths. Upstate, a relatively strong relationship is present, with about 30% of regional deaths associated with the entry of COVID-positive residents, based on the model. 
“The large number of deaths that the statewide model estimates are associated with COVID-positive admissions — more than 2,000 is unlikely– given that in the region where a statistically significant relationship between admissions and deaths was present, only 730 deaths occurred.”[21] (Emphasis added)
September 9, 2024, Empire Center Blog Post

Also, before getting into the technical aspects of the statewide regression model, I should address the most recent post by Bill Hammond of the Empire Center, titled Internal Cuomo Administration Documents Showed Evidence of Harm from Nursing Home Order, posted on September 9, 2024. According to Hammond, “documents [released by the Subcommittee] would seem to undermine, if not contradict, the Cuomo administration’s long-standing claim that the March 25, 2020, order had no significant impact on the health of nursing homes residents during the pandemic’s first wave.”[22] 


The “documents” that Hammond is referring to are the June 7, 2020, email from Dr. Eleanor Adams to Dr. Howard Zucker, and two accompanying slides prepared by McKinsey & Company. The McKinsey slides are labeled as showing the “average mortality rate [in nursing homes] by level of admission/readmissions”[23] statewide and for New York City alone. Hammond is referring to a chart in the statewide data that shows a mortality rate of 8.1 percent for nursing homes with ‘some [Covid-positive] admissions or readmissions,’ compared to a 4.1 percent rate for homes with ‘no admissions or readmissions.’ The second slide is a comparable chart for New York City nursing homes, with a somewhat lower mortality rate for homes with admissions compared to homes with no admissions (of which there were only four).


While Hammond believes these charts are meaningful, it is essentially impossible to discern their meaning with any confidence based on the information presented. The slides themselves raise questions that would need to be addressed before understanding the significance, if any, of the charts. These questions include:

  • why it makes sense for the chart to exclude some 240 nursing homes that “had no positive Covid cases” (during what period of time?);

  • why the statewide mortality rate in the chart is higher than the New York City mortality rate, when an earlier slide shows that the highest mortality rate in the state was in New York City;

  • what constituted the data set about nursing home deaths given that the slide was “self-reported data from hospitals” (emphasis added); and

  • was the relationship between admissions and mortality distorted by facility-specific factors, such as nursing homes that had Covid-only units or facilities and thus would be expected to receive a higher percentage of admissions as well as a higher mortality rate?


Obviously, Dr. Adams did not consider this information in the slides to undermine the conclusions she expressed in her email to Dr. Zucker, already cited above, that the data “shows that readmission/admissions of residents with COVID-19 to nursing homes most likely was not a driver of the number of deaths in nursing homes.” (Emphasis added)


I have a good deal of respect for Bill Hammond, but the fact that he places significance on evidence as slender as these slides may reflect what is known as “confirmation bias,” or the tendency to search for, interpret, and remember information in a way that confirms one's preexisting beliefs or hypotheses, while giving disproportionately less attention to information that contradicts those beliefs. If that is the case for the Emperor Center, they would have plenty of company. Confirmation bias is, unfortunately, one of the plagues of our time.

                                                                                      ***

As I have noted before, the purpose of my three Commentaries on the subject of the Covid-19 nursing home controversy in New York is not to fact-find the way a lawyer would, or to prove the negative that no nursing home deaths were associated with transfers of Covid-19 patients from hospitals. Rather, my goal has simply been to objectively review both the public record about these issues and the empirical evidence regarding the impact of the March 25 Advisory to separate facts from deeply held, but not well-informed, beliefs.


In short, the conclusions of my three commentaries are the following:

  • Whatever its origins, the March 25 Advisory was strongly supported by the public health officials and staff at the Department of Health.

  • The empirical evidence strongly suggests that, at most, the March 25 Advisory had a small impact on the number of deaths of nursing home residents in New York, which may well have been offset by health benefits attributable to transferring more than 8,500 medically stable Covid-19 patients from hospitals, thus freeing up hospital beds and reducing the chances for hospital-based infections among the transferred population. There is a consensus among analysts that there is no correlation between transfers and nursing home deaths in the Downstate region, and reasons to question the reliability of the estimate of fewer than 100 nursing home deaths associated with transfers in the Upstate region.

  • The decision not to report an estimate of out-of-facility deaths of nursing home residents as early as the DOH Factors Report has done great damage to the reputation of the Cuomo administration. In hindsight, releasing such an estimate sooner would have been wise. However, such an estimate would not have been fully accurate, which was an overriding priority for the final decision-makers.

  • No one has made a plausible argument that releasing the out-of-facility nursing home death estimates sooner would have had any effect on the State’s actions on the ground in the pandemic response. The decision not to release an estimate, correct or incorrect, was without consequence as it related to nursing home residents or others.

  • The substance of the DOH Factors Report – other than with respect to including the estimate of out-of-facility nursing home deaths – was not materially altered through edits by the Executive Chamber.

  • The Transcribed Interviews of ten former state employees not only corroborate but strengthen these conclusions and nothing in the Staff Memo (which was also based on the review of 550,000 pages of documents) contradicts these conclusions.


I acknowledged at the outset that it would be difficult to change the minds of many people on the subject, given how deeply rooted the prevailing narrative has become and the intensity with which it is held by so many among elected officials, the media, and the general public.


However, the House Select Subcommittee on the Coronavirus Pandemic did the public a great service by making available the Transcribed Interviews – a philosophy and practice far too rare in government today. Reviewing the Transcribed Interviews in combination with a careful reading of what the rest of the public record actually says – and does not say – should enable anyone who is interested to develop a good understanding of this controversy. I recognize that not everyone who takes the time to engage in this exercise will make the same observations and come to the same conclusions that I have, but they will at least be able to make better informed arguments.


The willful ignorance of many of Gov. Cuomo’s critics as it relates to the March 25 Advisory and surrounding events exploits the families who lost loved ones in nursing homes and further coarsens an already toxic political environment. Whatever judgment people who examine these issues ultimately arrive at, I hope they will find that these Commentaries have made it easier to understand the record and reach their own conclusions.

 


Paul Francis

September 17, 2024


Paul Francis is the Chairman of the Step Two Policy Project. He served as the Director of the Budget in 2007 and as the Deputy Secretary for Health and Human Services from 2015-2020, among other positions in New York State government, before retiring in May 2023.


 

Appendix


Technical Reasons Why Using a Statewide Regression Analysis Is Not Methodologically Sound

Discussions of regression analysis very quickly move into the “my eyes glaze over” territory. I am not a statistician, but I think the concepts described below should be accessible to the general reader.


As discussed, significant regional differences in both COVID-19 transfers and deaths make it inappropriate to use a statewide regression model to assess the correlation between nursing home deaths and patient transfers. The following are some of the technical explanations of why I believe that is the case.


Asymmetry in the Data Distribution

The Downstate region accounts for 87% of the nursing home deaths and 95% of the admissions of Covid-19 patients from hospitals, meaning that the majority of the variance in both the independent variable (transfers) and the dependent variable (deaths) occurs within the Downstate data. Pooling Upstate and Downstate data into one model assumes uniformity in the relationship between these variables across regions. However, the massive imbalance in deaths and transfers across these regions suggests that the Upstate data, where correlation is more notable, could be drowned out by the Downstate data in a statewide analysis.


Since no significant correlation between transfers and deaths is observed in the Downstate data, the overall results of a statewide regression are disproportionately influenced by Downstate data’s lack of a statistically significant correlation. Blending two fundamentally different scenarios into a statewide model introduces aggregation bias, obscuring, rather than clarifying, the dynamics in either region.


Differential Statistical Significance Across Regions

When a key predictor (transfers) behaves differently across regions—insignificant in Downstate but moderately significant in Upstate—it is inappropriate to assume a common underlying relationship across the entire state. A statewide regression would produce an average estimate that reflects neither region accurately, potentially obscuring the effect in Upstate while overemphasizing the lack of significance in Downstate.


Violation of Model Assumptions

Statewide regression models assume that the relationship between the independent (transfers) and dependent (nursing home deaths) variables is uniform across observations, but the substantial differences between regions violate this assumption. The low R² value (~.2) of the statewide model further indicates that the model fails to explain much of the variability in the data.


The dramatic regional imbalance in deaths and transfers and the variation in statistical significance strongly imply the presence of regional “interaction effects” in a statewide model. If Downstate and Upstate nursing homes are affected by transfers differently, a statewide model may violate the assumption of homoscedasticity, where errors should have constant variance across observations. 


Heterogeneity of the Data Between Upstate and Downstate

The Upstate and Downstate regions differ in several critical respects that make pooling data inappropriate. They experienced the pandemic at different times and at very different severity levels, with community spread being much greater Downstate than Upstate. The Empire Center acknowledged this when it observed that because of relatively higher community spread, downstate transfers were not significantly correlated with nursing home deaths but in upstate, relatively lower community spread increased the likelihood that transfers increased the likelihood of infection.


These contextual differences introduce additional confounding factors that make pooling the data inappropriate.


Conclusion

In conclusion, a statewide regression model obscures critical differences between the Upstate and Downstate regions. The disparity in deaths and transfers, the varying levels of statistical significance, and the potential differences in healthcare and policy responses all point toward the inappropriateness of a simple pooled analysis. Using region-specific models or a hierarchical approach yields more accurate, regionally relevant insights that better reflect the true dynamics between hospital transfers and nursing home deaths than a statewide model can.


 

Endnotes


[1] Staff Memorandum: Findings from the Select Subcommittee’s Investigation into the Cuomo

Administration’s March 25 Directive admitting COVID-positive patients into

Nursing Homes. United States Congress Select Subcommittee on the Coronavirus Pandemic Majority Staff. September 9, 2024. P. 2. https://oversight.house.gov/wp-content/uploads/2024/09/2024.09.09-SSCP-Staff-Memorandum-Re.-Investigation-of-Governor-Andrew-Cuomo.pdf 

[2] House Committee On Oversight and Accountability, Select Subcommittee On The Coronavirus Pandemic, U.S. House Of Representatives. Transcribed Interview. Howard A. Zucker, M.D. See p. 88. https://oversight.house.gov/wp-content/uploads/2024/09/Zucker_Final_Redacted.pdf 

[3] Ibid.

[4] Staff Memorandum: Findings from the Select Subcommittee’s Investigation into the Cuomo

Administration’s March 25 Directive admitting COVID-positive patients into Nursing Homes. P. 2.

[5] New York State Department of Health. Factors Associated with Nursing Home Infections and Fatalities in New York State During the COVID-19 Global Health Crisis. July 6, 2020 (revised July 17, 2020, and February 11, 2021). “DOH Factors Report.”

[6] Email from Dr. Eleanor Adams, Special Advisor to the Commissioner, New York State Department of Health, to Dr. Howard Zucker, Commissioner, New York State Department of Health. June 7, 2020.  https://oversight.house.gov/wp-content/uploads/2024/09/NYSDOH-7403-7424_Adams-June-7-summary-of-NH-findings-admissions_Redacted.pdf 

[8] DOH Factors Report, p. 16

[9] House Committee On Oversight and Accountability, Select Subcommittee On The Coronavirus Pandemic, U.S. House Of Representatives. Transcribed Interview. Gareth Rhodes. See page 116. https://oversight.house.gov/wp-content/uploads/2024/09/Rhodes-Final_Redacted.pdf 

[10] Ibid.

[11] Ibid.

[12] DOH Factors Report, p. 17.

[13] If only for completeness, I should refer to one of the more curious Findings in the Staff Memo, which was that “the Executive Chamber made the decision to change the methodology of nursing home fatalities to not include out-of-facility deaths” on May 3, 2020. This Finding is surprising because the record seems clear that the out-of-facility nursing home deaths was particularly unreliable in the first two months of the pandemic because systems were being built on the fly, and the out-of-facility fatality numbers were based on secondhand information from nursing homes about deaths in hospitals.

[14] This change is based on revised analysis using data the NYS DOH originally provided in September 2020, which contains no redactions. The analysis I used for my previous two Commentaries on this topic used a data source from October 202, which was more recent but contained redactions that obscured (ostensibly, as a matter of cell suppression) whether some facilities received transfers.

[15] WBFO NPR. Elderwood at Amherst nursing home taking in COVID-19 patients from hospitals. Tom Dinki. March 30, 2020. https://www.wbfo.org/local/2020-03-30/elderwood-at-amherst-nursing-home-taking-in-covid-19-patients-from-hospitals 

[16] The Empire Center. What Paul Francis Got Wrong About the Empire Center’s Nursing Home Research. Bill Hammond. September 11, 2024.

[17] The Empire Center. COVID-positive Admissions Were Correlated with Higher Death Rates in New York Nursing Homes. Bill Hammond and Ian Kingsbury. February 18, 2021.

[18] Statewide, the total number of admissions of COVID-19 patients from hospitals to nursing homes during the relevant timeframe was 6,236. That figure, multiplied by 0.07, equals 442. In Upstate nursing homes, the total number of such admissions was 328. That figure, multiplied by 0.30, equals 98.

[19] Policy by Numbers Blog. Was Cuomo’s COVID Mandate Associated with Nursing Home Deaths? A Correction. John Bacheller. Published March 11, 2021, and updated July 22, 2022.

[20] Ibid.

[21] Ibid.

[22] The Empire Center. Internal Cuomo Administration Documents Showed Evidence of Harm from Nursing Home Order. Bill Hammond. September 9, 2024. https://www.empirecenter.org/publications/documents-showed-evidence-of-harm-from-nursing-home-order/ 

[23] Email from Dr. Eleanor Adams to Dr. Howard Zucker. June 7, 2020. See slides 10 and 11 of the accompanying McKinsey & Company PowerPoint. https://oversight.house.gov/wp-content/uploads/2024/09/NYSDOH-7403-7424_Adams-June-7-summary-of-NH-findings-admissions_Redacted.pdf 


27 views0 comments

Commentaires


bottom of page