Health Services, Policy

Errors blight new Medicare data on Covid-19 in nursing homes  

The Centers for Medicare and Medicaid Services released new data showing the number of Covid-19 deaths and infections in nursing homes. But as it currently stands, the data has too many errors to be helpful for families trying to decide between individual facilities.   

One of the first Covid-19 outbreaks in the U.S. was reported at the Life Care Center of Kirkland in late February. Photo credit: David Ryder, Getty Images.

In early February, a sudden respiratory illness swept through the Life Care Center of Kirkland. Later confirmed to be Covid-19, the new disease caused at least 37 deaths among the nursing home’s residents.

Since then, nursing homes have become a major focus of the Covid-19 pandemic. The Centers for Medicare and Medicaid Services recently said it would crack down on infection control with penalties for repeated violations. Nursing homes also must report Covid-19 cases to the Centers for Disease Control and Prevention. So far, roughly 88% of Medicare-certified facilities — or 13,600 nursing homes — have reported this information, according to CMS.

The federal agency released new data last week showing at least 32,000 nursing home residents have died from Covid-19. It also named facilities, showing which nursing homes had seen the most cases and deaths among their staff and residents. But numerous flaws in the data make it difficult to determine which nursing homes actually had the most cases and are adequately prepared to care for patients.

For example, the Life Care Center of Kirkland doesn’t have any listed Covid-19 cases or deaths in the dataset. This is because CMS only requires facilities to report cases dating back to May, though some may opt to cumulatively report data back to January.

 

Improbable outliers

Other outliers point to clear errors in the data. A search of which nursing homes saw the most resident deaths from Covid-19 reveals several figures that shouldn’t be possible, given the number of beds.

For example, New Jersey-based Dellridge Health and Rehabilitation Center was listed as having 753 Covid-19 deaths, a figure several times larger than the most devastating known outbreak at any facility. In fact, the 96-bed nursing home had 16 residents die from Covid-19, according to state-reported data.

Other, similar cases stand out as outliers in the data. Complete Care at Summit Ridge, a 152-bed nursing home in New Jersey, was listed as having 311 Covid-19 deaths. In reality, 20 of its residents had died from Covid-19, administrator Israel Kanarek wrote in an email. He said the nursing home had already contacted CMS about the erroneous numbers.

Southern Pointe Living Center, a nursing home in Colbert, Oklahoma, has not had any Covid-19 cases, which is reflected in both the CMS data and an email from administrator Heather Mitchell. Yet despite that, the facility is incorrectly listed as having 339 Covid-19 deaths.

Though CMS acknowledged the data was preliminary, and subject to fluctuations, it didn’t specifically address these outliers in the data.

“As with many of the agency’s new reporting programs, CMS expected that the precision of the data would need to be refined over time,” a CMS spokesperson wrote in an emailed statement. “CMS released the data collected by the CDC while keeping in mind the balance between the need for transparency and speed against potential issues with the completeness of the initial data.”

 

Access data doesn’t tell the whole story

Lissy Hu, founder and CEO of care coordination software company CarePort, had hoped to use the data to help patients and their families make better decisions about where to go. Some patients might need to stay at a skilled nursing facility for a period of time as they recover from Covid-19. But only a fraction of nursing homes, roughly 13%, are taking recovered Covid-19 patients.

“People saw that there were all these deaths in nursing homes. I think there was a lot of public pressure for CMS to bring more transparency to what was happening in nursing homes, for good reason,” she said. “What we’re seeing in terms of what’s released, there are so many holes. People are really struggling with, what do I do with this data?”

Hu and her team looked at whether facilities said they had access to testing and a week’s worth of PPE. But they found that whether a facility reported having access to masks wasn’t necessarily correlated with star ratings, but rather with where there was an active Covid-19 outbreak.

While 96% of facilities said they had access to testing, those figures also can’t be taken at face value.

“You think, oh great, 96% have access to testing,” Hu said. “But do they have enough tests for residents, and the reagents they need to complete those tests? These nursing homes don’t have access to those supply chains.”

For the roughly 500 nursing homes that indicated they did not have access to any Covid-19 testing, states could use that information to act quickly, Hu said.

She also said she’d like to see more consideration of patient risk with the nursing home data. For example, does the facility primarily care for long-term residents, some of whom have dementia or other serious conditions? Or does it primarily see short-term rehabilitation referrals? That information could allow for more of an apples-to-apples comparison.

“I think CMS is doing the right thing in terms of trying to get more transparency. For that, they should be applauded,” Hu said. “I think we can do better in terms of the data that’s going out there so we can trust it to make decisions around which nursing home is most appropriate for our loved ones.”

Shares1
Shares1