Limits and Opportunities of Federal Reporting on COVID-19 in Nursing Facilities

In response to the widespread concerns about the high numbers of coronavirus cases and deaths in long-term care (LTC) facilities, CMS recently implemented new COVID-19 data reporting requirements for all federally certified nursing facilities. The newly-released federal data builds on state-reported data that has been included in many, but not all, states’ regular coronavirus reporting for several weeks. However, preliminary analysis indicates that the federal data will not reveal the full extent of the crisis in LTC facilities, as most states are reporting higher numbers of cases and/or deaths than appear in the federal data. This discrepancy is most likely due to the fact that the federal data only requires states to report cases and deaths after May 8^th and is limited to cases and deaths in nursing facilities. State-reported data primarily reports cumulative cases and deaths across multiple types of long-term facilities. However, not all states have reported data, and data are not easily comparable across states. This brief explores the extent to which state and federal data on coronavirus in LTC facilities differ and discusses the likely causes for these differences. While federal reporting offers promise for analyzing the intersection of COVID-19 and LTC facilities, state data will continue to play a role in highlighting the scope of the crisis across facility types and trends in COVID-19 within these facilities over time.

Comparing federal and state data on COVID-19 and LTC

The federal government implemented a new reporting system for certified nursing facilities, requiring facilities in all states to report coronavirus cases and deaths among residents and staff. The federal data aims to produce current and uniform facility-level data for all states for both residents and staff, making it more comparable across states than the state-published data (Table 1). According to the federal requirements, nursing facilities must report data on confirmed coronavirus cases and deaths for residents and staff in all certified Medicare skilled nursing facilities and Medicaid nursing facilities. These data are to be reported at least once a week to the CDC using a standardized reporting portal called the CDC National Healthcare Safety Network (NHSN). This data is then shared with CMS and publicly reported. Facilities must also report outbreaks to residents, representatives, and families by next calendar day at 5pm. These notifications must follow a single confirmed infection in the facility, or 3+ residents and/or staff with respiratory symptoms that occur within 72 hours of each other.

Federal data released by CMS also includes additional information on facility resources and characteristics relevant to COVID-19 (Table 1). In addition to cases and deaths, CMS reports suspected coronavirus infections among residents and staff, total non-COVID deaths, personal protective equipment (PPE) supplies, ventilator capacity and availability, bed count, resident census, and testing access, among other variables. Many variables reported in the federal data are not reported by most states, making the federal data valuable for understanding more details about the crisis on the ground.

However, the federal data is not cumulative, which understates the full extent of the pandemic. The federal data does not require states to report data prior to May 8^th, though facilities may report cumulative data prior to that period in their first reporting if they wish. Most state-published data report cumulative cases since the onset of the public health crisis (Table 1). Given high numbers of cases and deaths in nursing facilities as far back as March and April, the requirement to report as of May 8^th may exclude a large number of cases that occurred during the “peak” of the curve in many states.

Additionally, federal reporting does not capture cases and deaths in other congregate settings, including assisted living facilities and other LTC settings. In contrast, most states include data on cases and deaths in other types of LTC facilities, including assisted living facilities, intermediate care facilities, group homes, and others (Table 1). Analysis of state data shows a notable number of cases and deaths in other LTC facilities.

Federal and state data also differ in their inclusion of recovered or resolved cases and outbreaks and suspected cases (Table 1). Seventeen states report data on recovered cases and resolved outbreaks, which is not reported in the federal data. Recovered cases refer to residents who previously tested positive for the virus, but have since recovered. Resolved outbreaks refer to facilities where there are no active cases for a specific period of time (in most cases, 14 days). Several of these states have also flagged facilities that are no longer undergoing an active outbreak. Understanding which facilities have successfully contained outbreaks can be helpful in understanding successful infection control policies. The number of recovered cases is also a valuable measure to understand the extent of the immediate crisis in facilities. Federal data does include suspected cases or deaths, while most states do not report this data.

Table 1: State and Federal Data Reporting on COVID-19 Cases and Deaths in Nursing Facilities
	State Reporting (46 states + DC)	Federal Reporting (50 states + DC)
Number of states with publicly available data on cases	42 states + DC	50 states + DC
Number reporting facility-level data on cases	29 states + DC	50 states + DC
Number of states with publicly available data on deaths	40 states + DC	50 states + DC
Number reporting facility-level data on deaths	26 states + DC	50 states + DC
People included in case and/or death counts	Residents + staff:  30 states + DC Residents only: 9 states Unspecified: 4 states	Residents + staff
Types of long-term care facilities included	Multiple LTCFs:  44 states Nursing facilities only: 2 states + DC	All certified Medicare skilled nursing facilities and Medicaid nursing facilities
Time period reported	Cumulative:  41 states + DC Active:  5 states	Varies, but at a minimum weekly cases and deaths starting May 8th
Frequency of data update	Daily: 23 states + DC At least once/week:  18 states Unclear:  5 states	Weekly
Includes suspected cases or deaths	8 states	50 states + DC
Includes recovered cases or resolved outbreaks	17 states	0 states

As of May 31st¹, the federal government is reporting about one-third fewer cases and deaths in long-term care facilities compared to the state data, reflecting differences in the scope of information included. Among the set of states for which there is both state and federal data, the federal data reports 37% fewer cases and 33% fewer deaths than state data. State data reports about 43,700 deaths for the 40 states reporting this data, while the federal data reports about 29,200 deaths among the same 40 states. State data reports about 217,400 cases for the 43 states reporting this data, while the federal data reports about 137,000 cases among the same 43 states.

Even including all states in the federal data, the federal data for 50 states plus DC is reporting less than three quarters of what the state data for 44 states reports nationally. Federal data reports 32,500 deaths and 155,700 cases among all states – about one-quarter fewer than the cases and deaths totals reported by states.

Five states report higher cases in federal data (Arkansas, Delaware, Michigan, North Dakota, and Wyoming); four states report higher deaths in federal data (Mississippi, Oklahoma, Oregon, and West Virginia). In most of these cases, the state data may be lower because it includes only resident (not staff) cases (AR, DE, MI) or only active cases (ND, MI, OR), and/or is based on a very small number of facilities in which a minor case undercount can sway percentages (WY). Other states, such as Oklahoma and West Virginia, may have data errors in the federal data, as the federal data currently reports a greater number of COVID deaths in nursing facilities than COVID deaths in the whole state. All other states report more cases and/or deaths than CMS reports. Most likely, this discrepancy is due to the reporting period (and whether cases/deaths prior to May 8^th are reported) and types of facilities reflected in federal and state-published data.

Differences in the number of cases and deaths between the federal and state data lead to different conclusions about the extent to which the crisis is concentrated in LTC facilities. These differences are largely driven by the differences in time period and types of facilities reflected in the federal and state-published data. Based on the numbers reported in the state and federal data, as well as the total number of cases and deaths from our daily cases/deaths tracker, we calculated the share of total COVID cases and deaths in the state that can be attributed to LTC facilities. State-reported data shows that LTC facility cases account for 16% of coronavirus cases across 43 states, while the federally reported data shows that nursing facility cases account for 9% of cases. State-reported data shows that LTC facility deaths account for 45% of coronavirus cases across 41 states, while the federally reported data shows that nursing facility deaths account for 31% of deaths. This comparison varies across states. The following interactive maps show state-by-state comparisons of the share of total cases and deaths that can be attributed to LTC/nursing facilities reported in the state and federal data, respectively.

Notably, the federal government reports more COVID-19 deaths in Oklahoma and West Virginia’s long-term care facilities than total COVID deaths in those states as of May 31^st. The analysis caps the share of deaths that can occur in LTC or nursing facilities at 100% of all COVID deaths in those states.

Interactive 1: 

Interactive 2: 

Looking Ahead

While early state reports on cases and deaths in LTC facilities were crucial in highlighting the scope of the problem in those settings and targeting policy to these hot spots, state-reported data has several limitations. Data on cases and deaths in long-term care facilities is still not reported for all states, and even among states reporting, is not directly comparable. Moreover, the state data generally does not include the additional indicators available in the federal data, including information on PPE availability and staffing shortages, which can help target policy responses.

Federal data offers great promise to address some of these limitations. At the same time, since the federal data was released, there have been several reports of inaccurate data, including issues of reporting fatalities where there were none and undercounting fatalities in facilities with known fatalities. The federal data also reports two states that report a greater number of COVID deaths in long-term care facilities than total COVID deaths in the state. Analysis of facility-level data, paired to account for timing and case inclusion criteria, can shed light on the scale of data discrepancies between state and federal data. To some extent, if data errors in the federal data are present, it is unsurprising given the inherent challenges with setting up a new data reporting system. Fielding a large-scale, nationwide data collection effort can often take months, or even years. Over time, it is possible that data errors will be resolved in the federal data. However, the federal government has limited levers to require reporting for LTC facilities it does not regulate, and state data will continue to be valuable to understanding the full scope of the COVID-19 crisis in congregate settings.

Overall, given the fast-moving nature of the COVID-19 crisis and need for accurate, timely data, both the federal and state-reported data on COVID-19 and LTC will likely play an important role in understanding the trajectory of the pandemic.