The COVID-19 nursing home data released in late May by the Centers for Medicare and Medicaid Services contains a large number of inaccuracies, an analysis by the Washington Examiner found.
In its press release announcing the data, the agency stated that there were “over 95,000 confirmed COVID-19 cases and almost 32,000 deaths” in U.S. nursing homes. Those numbers are now in doubt, as is whether the data can yield insight into how the virus was transmitted to nursing home residents.
At least 33 nursing homes in the data show more residents dying from COVID-19 than they have beds to accommodate residents. For example, the data show that the Southern Pointe Living Center in Colbert, Oklahoma, had 339 resident deaths, but the facility only has 95 beds. The center told Medpage Today that no resident had died from COVID-19.
The data also show many nursing homes with more residents and staff with COVID-19 than the facilities have beds. The number of staff members is usually lower than the number of residents at long-term care facilities.
The most egregious case of this was Smith Village in Chicago, which was listed as having 1,105 residents and 955 staff members with COVID-19. Yet it has only 100 beds, according to the data. A spokeswoman with Smith Village confirmed that it had 37 residents and 30 staff members who tested positive for the coronavirus. At present, only one resident still has the virus and is asymptomatic, while all of the staff have returned to work.
In all, the data listed 641 nursing homes with more residents infected with COVID-19 than they had beds and 702 nursing homes with more staff members infected than they had beds.
The total number of cases and deaths in the data may be inflated. Many nursing homes are listed twice. The aforementioned Smith Village is one such facility, and its second entry lists 851 cases among residents and 745 among staff.
“Data, in general, is going to have discrepancies. Any rushed attempt like this is going to have even more errors,” said John McHugh, a professor at the Mailman School of Public Health at Columbia University. “Especially when you have facilities that are simultaneously trying to fight a virus and comply with new data reporting requirements.”
The data collection process was rapid. On April 19, CMS required that nursing homes report coronavirus data. About 88% of the 15,600 nursing homes in the U.S. reported data by the due date of May 17. Two weeks later, CMS released the data to the public.
Another likely inaccuracy in the data is the number of nursing homes that have infections among the staff but none among the residents. There are 164 facilities in the data that have over 10 staff with COVID-19 but zero infected residents. Many experts believe that staff was a potent source of transmission of the virus to residents.
The Lakeside Rehabilitation Center in Lubbock, Texas, was listed as having no infections among residents but 658 among staff. A letter to residents and families posted on the facility’s website said 42 residents and 44 staff members had COVID-19.
Vince DeChellis, co-founder of the healthcare data technology company DarkMatter2db, said the problem lies in the federal government’s lax standards.
“The usual cause is the lack of standards and oversight,” he said. “The credentialing, by which we mean, ‘Who is the individual [entering the data]? What is their specialty? What is their degree?’ There are very few standards for that.”
CMS did not respond to a request for comment. In an email to MedPage Today, CMS said, “As with any new reporting program, there can be data submission errors in the beginning. In an effort to be transparent, CMS made the data collected by the CDC public as quickly as possible balancing transparency and speed against the potential of initial data errors.”
DeChellis suggested that CMS is by no means alone among federal agencies when it comes to quality problems in the healthcare data it collects.
“This is a huge topic,” he said. “You are only scratching the surface.”
