According to a new study of homeless shelters in four major American cities, it is crucial to broadly test residents and staff for the coronavirus at the detection of the first case — or sooner — and to isolate those who test positive in order to prevent a wide spread of infection.
Even in two shelters in Atlanta, which had no known coronavirus cases, comprehensive testing of residents and staff found that 4 percent of the residents and 2 percent of staff members tested positive, suggesting that some people were spreading the disease through the shelters before they had symptoms.
And in 12 Seattle shelters, which each had only one known previous case, 5 percent of the residents and 1 percent of staff members tested positive.
But by the time there was a cluster of cases (meaning two or more) at shelters, the extent of the spread was far greater, according to the study released on Wednesday by the Centers for Disease Control and Prevention.
The study’s researchers took data from tests administered at shelters in Atlanta, Boston, San Francisco and Seattle from late March through early April. The program sought to test every resident and staff member at 19 shelters in those cities, but in some cases residents were not available or declined to participate.
After a cluster was detected in a shelter in Boston, testing found that more than a third of the residents had the virus, while two-thirds tested positive at a shelter in San Francisco.
“If you wait for a cluster to develop, you are almost too late,” said Dr. Jim O’Connell, the president of the Boston Health Care for the Homeless Program, who took part in the study. “The lesson is that from the beginning of this epidemic, we should have been testing people in nursing homes, prisons and in shelters because that is where it is spreading asymptomatically, and it can be deadly before you know it.”
Dr. O’Connell noted that some shelters outside of the study have reported very high rates of positive test results with little prior indication of infection. He said that 49 residents out of 114 (43 percent) tested positive at a shelter in Worcester, Mass., and very few of them had noticeable symptoms beforehand.
“That was before they had a cluster,” he said. “What we really need to be doing is testing early and often in shelters.”
The C.D.C. study said that approximately 1.4 million people nationally use homeless shelters each year. It added that infection-control practices at shelters are especially important, but they can be challenging because of overcrowding, the older age and underlying health conditions of many homeless people, a lack of hand sanitizers and the difficulty of keeping guests sheltering in place.
Dr. Georgina Peacock, a leader of the C.D.C.’s Covid-19 Response At Risk Task Force, noted that many of the study’s findings also apply to the population at large: Early and regular testing, followed by isolation, is vital to controlling the spread of the disease.
But she said these issues are magnified at shelters because of the conditions found at many of them.
“This is a vulnerable population,” she said, “and it is difficult in homeless shelters to do some of the things that we know are important to contain spread.”
In all, officials tested 1,192 residents and 313 staff members. Five of the 19 shelters had clusters of two or more people known to have contracted the virus before the testing was done. Those were the ones with high rates of infection.
Dr. O’Connell noted that the overall rate of infection in a region should be factored in to any conclusions. The more cases there are in a particular city, the more cases would be expected to be found in shelters there. The study said that Boston had the highest rate of infections among the four cities, at 14.4 cases per 100,000 residents at the time of the testing period. San Francisco had the lowest rate, at 5.7.
Another study, also released Wednesday by the C.D.C., examined how public health officials, in coordination with the C.D.C., tested 233 residents and staff members at three affiliated shelters in Seattle, where staff and residents had been moving regularly from one shelter to another. It found 18 percent of the residents at those shelters and 21 percent of the staff members tested positive for the virus.
Aggressive isolation measures after the testing helped to mitigate the spread, Dr. Peacock said. Residents who tested positive were transported to hospitals or brought to isolation areas away from the shelters to avoid further contamination.
Both reports recommended that service providers apply physical distancing measures where possible, including ensuring that residents’ heads are at least six feet apart while sleeping, and promote the use of cloth face coverings for everyone in a facility.
“Sometimes you don’t have that ability to do some of that social distancing,” Dr. Peacock said, “so you have to plan for how to reduce crowding.”
Sharon Bogan, a spokeswoman for Public Health — Seattle & King County, said there are currently 74 people in isolation, quarantine and recovery sites around King County.
According to the C.D.C. report, 147 of 408 residents at a Boston shelter had tested positive for Covid-19. Dr. O’Connell said that, like the shelter in Worcester, many of them were asymptomatic. Once they tested positive, they were moved to other locations, including a temporary field hospital in a Boston convention center with 500 beds set aside for homeless people. Other homeless guests, who did not test positive, were moved to dormitories at nearby Suffolk University to ease overcrowding.
“Large cities need to make sure they incorporate the homeless shelters and homeless population into their overall plan to treat the virus,” he said. “Cities have not done that, so this is a good clarion call for everyone.”