We Still Don’t Know What Works Best to Slow the Spread Of COVID-19

At the start of 2020, most of the world was terrified. Faced with a novel, deadly pandemic virus, something most of us had not expected to experience in our lifetimes, and witnessing the carnage the virus reaped in Wuhan, China, and Lombardy, Italy, countries worldwide went into protection mode.

To reduce viral transmission and save lives, nations implemented pandemic control policies. These included “test and trace,” isolation of infected people, quarantining of those exposed, indoor mask mandates, and closing various venues to try and reduce contact between individuals. Daily life in many countries changed drastically.

Since those first, bleak days of the early pandemic, we’ve had plenty of time to reflect on the steps taken at the start of the crisis, when governments and their public health advisers were making emergency decisions armed with very little data and information on an entirely new illness. This was the era before we had developed the powerful vaccines and medicines that have transformed the outlook for COVID-19. While there is certainly evidence that these early community mitigation strategies, which scientists call “non-pharmaceutical interventions” (NPIs), reduced the spread of the virus, what might surprise you is how little effort there has been to fully assess their impact.

Because of a lack of research on NPIs, we still can’t answer important questions like: which government measures had the greatest and the least impact? How did the sequencing and timing of these NPIs affect their effectiveness? Which measures caused more harm than benefit? We need answers to these questions so we can prepare for the next pandemic, armed with better knowledge.

When it comes to NPIs, every angry person online has a strong belief that if only we had spent more time promoting mask wearing, been more like Sweden with its government-sponsored healthcare and incredibly generous paid sick leave provisions, or done something, anything, better than we did, we could have averted the mass death, disability, and orphanhood that COVID-19 caused. However, given the lack of data, it’s remarkably hard to know exactly how we could have used NPIs more effectively.

Read more: The Pandemic Will Be Over When Americans Think It Is

The most strident critics of government interventions and of public health measures during COVID-19 go so far as to say that the “cure was worse than the disease”—that is, they think NPIs killed more people than COVID-19 itself. Our research found no evidence for this assertion; we found that letting the virus rip through the population in an uncontrolled way was much deadlier, at least in the short term, than the most stringent NPIs, such as shelter-in-place orders.

Nevertheless, as we previously argued, highly restrictive NPIs clearly caused harms. For example, prolonged shelter-in-place orders were linked with an increase in harmful alcohol use and domestic violence. However, there has been little in the way of research on the trade-offs—that is, on understanding the balance between the harms of uncontrolled viral transmission versus those of NPIs. And it can also be very difficult to distinguish the impacts of the pandemic itself from the harms of NPIs. There’s no doubt, for example, that prolonged school closures affected children’s mental health, but so did losing a parent or other caregiver to COVID-19.

With all NPIs, when you start digging into the research evidence, the picture isn’t always clear cut. Take masks. From a basic science perspective, masks work—they filter the particles that we breathe. High filtration masks, like N95s, work better than surgical or cloth masks. Masking provides quite a bit of protection for the people wearing them against respiratory diseases, and can also help reduce transmission from an infected person to others.

In theory, then, if every person in the world had worn a high-quality mask 24/7 for a few weeks the COVID-19 pandemic would have been, if not over, then at least substantially slowed. But in practice, the intervention that we implemented was not some perfect ideal of mask-wearing, in which everyone consistently wore a well-fitting N95 in every situation. During surges, not everyone masked indoors, not everyone wore N95s, and those who did wear a mask may have worn them imperfectly (we’ve all seen people wearing masks with their noses uncovered, or even with their masks hanging around their necks).

When researchers have assessed mask wearing under “real world” conditions, the impacts have been smaller than studies done under perfect conditions. The biggest real world randomized trial ever run, in Bangladesh, studied the impact of giving people free surgical masks combined with promotional activities at mosques, markets, and other public places. The intervention led to mask usage more than doubling (from 13% in villages without the intervention to 42% in villages with the intervention) while the reduction in COVID-19 cases was only 9%. This modest reduction in infections is in line with the reductions seen in other, smaller real world studies.

What about other NPIs like large event bans or shelter-in-place orders? Many people aren’t aware that the effectiveness of such NPIs reduced dramatically between 2020 and 2021, even though the NPIs were often stricter in 2021 than they were before. As people reported lower compliance with government restrictions, the number of cases that each NPI prevented fell. It’s quite likely that implementing, say, a ban on large gatherings, was more effective in 2020 than in the following years simply because people were already changing their behaviour in response to the pandemic anyway.

Then you can add additional complexity on top of that. A new independent report from Australia into the country’s pandemic response shows precisely how complicated evaluating our decisions can be. As the report notes, Australia has seen some impressive successes during the last two years, but there are also many areas where the pandemic response was implemented poorly. While shelter-in-place orders (“lockdowns”) were effective, some of these orders and border closures were avoidable. Disadvantaged people across Australian society were the most heavily impacted both by the virus and the NPIs put in place to mitigate it. One of the key arguments in the report is that even the most effective NPIs had costs, and those costs were not only unfairly distributed but also could probably have been avoided. We could have reduced the harms of NPIs while also maximizing their benefits.

Now, this report is based on Australia, but it’s easy to see how the same idea applies the world over. School closures were in part harmful because low-income children often did not have ready access to laptops and high-speed internet, which is something that governments could have addressed. Many outbreaks across the world disproportionately affected essential workers who could not stay home, including health workers, bus and train drivers, and people working in the production and processing of food, but governments often did little to improve conditions in their workplaces until it was too late. The lack of federal paid sick leave in the U.S. was a massive hindrance to controlling COVID-19. In some countries, people who had to isolate or quarantine were not given financial or food support, making it much harder for them to comply. Too few places instituted what Tufts University epidemiologist Ramnath Subarraman and colleagues call “humane shelter at home,” a term that highlights both the public health benefits of shelter in place and also the need to provide social protections—such as income assistance—that help vulnerable populations weather the storm.

But the problem with all this complexity is that it is anathema to the tedious simplicity that surrounds most COVID-19 retrospection. It’s easy to argue that ill-defined “lockdowns” have caused unimaginable harm, or that even the most extreme, ongoing NPIs are a great idea. It is, however, far harder to ask difficult questions like “When is it reasonable to close schools due to infectious diseases?” or “Do stay-at-home orders have a marginal benefit or harm when coupled with a range of other NPIs?” or even “Could we have achieved the same reduction in cases with less damaging interventions?”.

Unfortunately, difficult questions don’t win any political points, even though they are the most important ones to answer. Imagine if the next pandemic comes along, and it turns out to be uniquely harmful to children, we have no choice but to close schools, but we’ve made no progress on how to mitigate the harms of school closures—it would be an entirely preventable disaster. Until we can start having public discussions that focus on figuring out the best way to combat a pandemic rather than assigning blame, we’re never going to know what to do when the next novel virus comes along.

Which is a problem, because one thing virtually every expert agrees on is that we will face another pandemic just like COVID-19, or even more deadly, at some point in the future. Hopefully, we can get ready for it.