Thousands of deaths tied to U.S. nursing homes could have been prevented if operators would have slightly adjusted their staffing model, academic researchers assert in a new study.

The scientists tracked smartphone usage among all of the nation’s nursing homes for six weeks after federal guidance restricting visitors was issued in mid-March. They found that a statistically significant 7% of the 500,000-plus phones tracked pinged as cross-traffic among facilities.

The network indicators are “extremely predictive of cases,” tweeted Judy Chevalier, the William S. Beinecke Professor of Finance and Economics in the Yale School of Management, on Friday. “Obviously, there are limits to any observational study, but our point estimates suggest that shutting down all of these cross-connections could have led to a 44% decline in nursing home cases.”

The results provide significant evidence that paying nursing home employees to work at only one facility and otherwise limit “cross-traffic across homes” would deliver a “magnitude” of benefits, said Chevalier and fellow researchers M. Keith Chen, professor of behavioral Economics and Strategy,  and Elisa F. Long, both of the UCLA Anderson School of Management, in concluding study remarks.

Their work is currently under fasttrack review as a working paper with the Proceedings of the Journal of the National Academy of Sciences of the United States of America. The goal is that it will be released as a National Bureau of Economic Research paper by the end of this week, said a person familiar with the study. An abstract can be found here.

Ultimately, the hope is to have it published by the National Academies of Sciences, which could then lead to recommendations by the Centers for Disease Control and Prevention or the Centers for Medicare & Medicaid Services.

Investigators were “blown away” by the results, according to an expert familiar with the study. “It’s much worse than we thought. It’s insane,” a person close to the investigation said. 

Harvard medical policy expert Professor David Gabrowski called the study results “fascinating” in a tweet, adding “Wow!” about the 44% projection.

The research built on study findings from the CDC’s examination of the first known U.S. outbreak of COVID-19. That investigation focused relatively narrowly on Seattle-area facilities, which were found to have shared staff who spread the virus asymptomatically. Since the late-February outbreak, roughly 40% of all U.S. COVID-19 deaths have been related to nursing homes. Nursing home coronavirus death estimates vary widely, from about 40,000 to 60,000 or more, and rising.

The smartphone researchers used satellite imagery to outline the rooftops of all 15,400 nursing homes in the U.S. They then tracked phone signals from phone service providers (not cell towers). Veraset provided anonymized data. While some of the phones could have belonged to residents, the overwhelming majority are assumed to have not. A 15% to 30% sample of the cell phone traffic was more than enough to reliably sample results, a person familiar with the study explained.

The findings are seen as a “positive” for providers, who aren’t being blamed, the person pointed out. Rather, the findings could point to a “very practical, though logistically expensive but achievable way of decreasing COVID deaths in the U.S.” In the long run, asking nursing homes to have their staff live on campus as much as possible “may be the most cost-effective way to combat” COVID-19.

The study’s findings also could have implications for “cross-linkages” among other congregate settings, researchers wrote in their concluding remarks. These could include assisted living facilities, prisons and “large workplace facilities such as food-processing plants.”