Various strategies are available to slow the spread of the dangerous coronavirus. These include enhanced hygiene, “social distancing,” wearing masks, and restricting travel outside the home. This last strategy, often called “shelter-in-place,” is now being enforced by many governments in the United States and around the world.

All these strategies work by reducing transmission rates, which protects individuals and “flattens the curve” to avoid overloading medical facilities and other resources. Preliminary evidence and common sense suggest that all of these strategies can work. If enough people frequently wash their hands, sanitize them when unable to wash, and avoid touching their faces, that will reduce transmission of the virus. Combine hygiene with social distancing and the likelihood of success becomes even greater. Many in Hong Kong believe the use of masks saved their city from a more serious outbreak.

Most, but not all, governments have gone further, issuing “shelter-in-place” decrees. These rules add even more pain amidst the pandemic, contributing to job losses, increasing social isolation, and criminalizing innocent behavior. If shelter-in-place produces much better outcomes than policies with less harmful side effects, it may be justified. Think of it this way: If shelter-in-place was a drug, would the FDA approve it? Unfortunately there has been little study, or even debate, about whether these rules work better than less harmful alternatives.

States that have not issued shelter-in-place orders are Arkansas, Iowa, Nebraska, and North and South Dakota. The data are preliminary, but how are these places doing so far?

New York is included to show the severity of the outbreak in that state—in the United States, more than a quarter of all cases and more than a third of all deaths have taken place in that one state. Otherwise, all of the states in the Midwest corridor, from North Dakota and Minnesota to Texas and Louisiana, are shown for comparison. (Data on cases and shelter-in-place orders are from the New York Times as of April 13.)

Louisiana, facing a serious outbreak of the virus, issued a shelter-in-place order on March 23. Minnesota, despite the small number of cases there, issued an order on March 27. Kansas did likewise on March 30. Oklahoma Gov. Kevin Stitt shut down “nonessential businesses” and required older residents to shelter-in-place in much of the state on March 24 and expanded his order statewide on April 1; the state’s two largest cities are under full shelter-in-place orders. Texas issued a statewide shelter-in-place order April 2 after a number of its largest counties had already issued similar orders in March. Missouri’s shelter-in-place order took effect on April 6.

Excluding Louisiana and comparing the five remaining shelter-in-place states with their free-state neighbors offers a very preliminary analysis of whether the orders work. As of April 9, the average number of cases per hundred thousand in the shelter-in-place states was 46.8. The average for the free states was 52.4. The death rate per hundred thousand people is slightly lower in the free states. In other words, this very preliminary data suggests no meaningful difference between states with shelter-in-place orders and those with less restrictive policies.

Other data also suggest that shelter-in-place orders have limited efficacy. A new academic paper (preprint—not yet peer-reviewed) finds that “social distancing measures were associated with a decrease in U.S. COVID19 epidemic growth.” It also finds “no additional decline in mean daily case growth after implementation of statewide restrictions on internal movement.” In other words, shelter-in-place or “lock down” orders had no effect.

This seems apparent in Seattle. The Wall Street Journal reports that new case numbers around Seattle peaked on March 26 and that “the state’s peak of Covid-19 deaths may have already passed.” Gov. Jay Inslee ordered the state’s residents to stay home on March 23, just three days before the peak. Because symptoms and then testing take several days, there is no way the Inslee’s order could be responsible for flattening the curve in Seattle. Instead, social distancing and corporate work-from-home policies appear to be responsible. (Singapore offers a similar example.)

Meanwhile, Sweden has remained radically open and free during the pandemic. As of April 13, Sweden had 103 cases and 9 deaths per hundred thousand people. Its neighbors are all under some version of shelter-in-place orders. Norway has 123 cases and 2 deaths per hundred thousand; Finland has 56 cases and 1 death per hundred thousand; Denmark has 109 cases and 5 deaths per hundred thousand; and Germany has 148 cases and 3 deaths per hundred thousand. In all these nations, the rate of growth is declining.

The question in Sweden is whether the virus’s toll there will actually be worse in the long run. Note that the idea of flattening the curve is not necessarily that fewer people will become infected, but that the pace of the pandemic will be slow enough that medical resources are not overloaded. Either way, Sweden has not even embraced minimally restrictive measures like social distancing or masks.

One important note is that population density matters. In the most dense areas, like New York City, a policy like shelter-in-place may be the only practical way to implement social distancing. But what passes for social distancing in a major city may be very much like ordinary life in a rural community or even a suburb. These are all important areas for further research that will, hopefully, guide better policymaking in the future. This is also a reminder of why the system of federalism, which puts most governing power in the hands of the states, is an important part of our constitutional system even amidst a pandemic.

Whatever policies governments adopt, the pandemic itself will cause job losses and other economic devastation. Yet one rule of medicine—and a good guideline for public policy—is to first do no harm. Just because a patient is already sick, that is no excuse to experiment with unproven remedies with known dangerous side effects. The preliminary data on shelter-in-place orders, compared to less harmful policies, suggest that they may offer little or no additional benefit in combating Covid-19 in most places.