Key Questions About the Impact of Coronavirus on Long-Term Care Facilities Over Time

This analysis is based on data from 38 states plus Washington DC, for a total of 39 states. Within these 39 states, we were able to trend long-term care cases in 35 states and long-term care deaths in 36 states. Data was trended as far back as internal records and publicly available historical data allowed. States were chosen based on where we could reliably trend data. States were excluded from this analysis if they do not report data on cases and deaths in long-term care facilities, if their data is sourced from sporadically released media reports, or if there were data quality issues. The 12 states excluded from the analysis were excluded for the following reasons:

Alaska, Arizona, Hawaii, Missouri, Montana, New Mexico, South Dakota – Not reporting cases and deaths in long-term care facilities

Maine, Nebraska, New Hampshire, Wyoming – Data on cases and deaths in long-term care facilities are sourced from sporadic media reports

West Virginia – State-reported data has severe data quality issues

For all states, we trended the subset of data that would give us the longest reliable trend line. Notable examples of this include Louisiana, where data from non-nursing home long-term care facilities were excluded because they were not consistently reported. In Delaware, data excludes staff cases because that data was not reported consistently. For this reason, this analysis should not be used to identify state-level or national data on total long-term care cases and deaths. The most recent data on total cases and deaths in long-term care facilities can be located here.

Table 1: Long-Term Care Coronavirus Cases and Deaths Per 100,000 US and State Residents As of August 20th, 2020

This table presents the burden of long-term care cases and deaths that each state has experienced as of August 20^th, 2020. Total population data was taken from 2019 state population estimates from the US Census Bureau. The latest long-term care cases/deaths data available was used to calculate the burden of cases and deaths experienced by each state. National long-term care case burden was calculated by summing US population from 35 states that reported case data, dividing total cases by total population in those 35 states, and multiplying by 100,000 to find the value per 100,000 US residents. National long-term care deaths burden was calculated similarly by using data from 36 states that report LTC deaths.

Table 2: Average New LTC Cases and Deaths Per Week Per 100,000 US and State Residents, By Month

Total population data was taken from 2019 state population estimates from the US Census Bureau. The first week of available data for each state was not included in this analysis since the first week of data does not reflect a single week of cases/deaths, but rather all cases and deaths that have occurred up to that point. New cases and deaths were calculated for each week thereafter, and then averaged for all of the weeks within the month. April, June, and July reflect 4 weeks of data. May reflects 5 weeks of data. August reflects 3 weeks of data. These average new cases were converted to represent cases and deaths per 100,000 state residents to allow for easier comparison across states. National new cases and national new deaths were calculated by averaging new cases and new deaths across states. See limitations for more details on this process.

Limitations

There were several possible approaches to this analysis, all of which posed major limitations. This analysis could be limited to the time period where most major states are reporting, which would limit the time period of this analysis to mid-June to present day. However, this approach would miss the major peaks in states such as Massachusetts, Connecticut, and New Jersey. Another option was to limit the analysis to states where we had the earliest weeks of data available. However, this approach would exclude Texas, Michigan, Maryland, and several other states where data was not available until at least several weeks after other states began reporting. Deaths in the states that would have been excluded in this approach make up 30% of all long-term care deaths due to COVID-19. This analysis could have also used the federally reported data, but this would have limited the time period of analysis as well.

Due to data availability and quality issues, we were unable to include all states in this analysis. Thus, all national calculations in this analysis are subject to data availability. In particular, national calculations of new cases and deaths in Table 2 are limited by varying numbers of states included in each week and month, the differences in policies across states, and the lack of reporting comparability. However, given the data limitations, this is the best approximation of new cases and deaths per month.

We conducted a sensitivity analysis for the analysis in Table 2 to see how much using different states in each month affected both the raw Ns and the direction of the trend. Our sensitivity analysis found no notable differences. When looking at the same set of 23 states for new cases and 19 states for new deaths from April to August, we found similar trends as when looking at all available states in respective months. As mentioned above, this method excludes several major states, such as Texas, Michigan, and Maryland.

Another key limitation is related to data on cases. Two large states – Florida and New York – do not report data on cumulative cases. Thus, the national numbers we present for case data are not truly nationally representative. These states report a large number of deaths, so they have likely experienced similarly large numbers of cases.