HONG KONG — Lockdowns, quarantines and extreme forms of physical distancing work: They are curbing the spread of COVID-19. But they cannot last indefinitely, at least not without causing enormous damage to economies and compromising peoples’ good will and emotional well-being.
When governments decide to close schools (or not), for example, they are implicitly trying to balance these various interests. One major problem, though: Their calculus about the underlying trade-offs typically is unclear, and the criteria for their policy adjustments are unknown.
A formal framework is needed, with an explicit rationale grounded in science, for determining when and how and based on what factors to relax restrictions — and how to reapply some or all of them should another epidemic wave hit again.
Containment has failed everywhere. In some places — Wuhan in February; northern Italy in March — the epidemic spread so quickly that the relevant authorities had to focus mainly on mitigating its effects, on damage control. In other places, suppression has worked so far: Hong Kong, Singapore and Taiwan have not experienced sustained local epidemics. Not yet, at least.
But in the many more places now in the throes of full-on epidemics, notably in the U.S. and Western Europe, the pressing concern is how to suppress the virus’s spread so as to avert a Wuhan-like health disaster, but without destroying economies or undermining people’s resilience and their willing consent to very taxing social-distancing measures.
The first objective of any response anywhere must be to protect lives, and that means averting the collapse of the health care system. Hospitals are the last line of defense. When their capacity to handle emergencies is overwhelmed — as in Bergamo, northern Italy, or in areas of Spain — there is little point theorizing as I am about to: All one can do then is to roll up one’s sleeves, hook up patients to intravenous drips and ventilators, and try to save as many lives as possible with whatever means are available.
But past that point (or, preferably, before it), the ultimate objective must be to bring the epidemic down to a slow burn so as to buy time for the world’s population to acquire, one way or another, immunity to COVID-19. The COVID-19 pandemic can only be prevented from resurging when at least half the world’s population has become immune to the new virus. And that can happen in only one of two ways: After enough people have been infected and have recovered, or have been inoculated with a vaccine.
Allowing the first option to happen, unmitigated, would be a humanitarian catastrophe: It would mean very many deaths, mostly among the elderly and poor people with limited access to health care. The second option — developing a safe, effective vaccine and making enough of it for everybody — is a goal at least one year, perhaps two years, away. Massive lockdowns and distancing measures cannot be sustained that long.
(Note that it is very unlikely that any large community has acquired sufficient herd immunity to the novel coronavirus yet, even where it has hit hardest. Nor do we have enough information about the people known to have been infected and to have recovered so far — that’s why it is so urgent to conduct serology, or blood work, to study how many people in countries that have already experienced a first wave of infections, like China and South Korea, are producing antibodies.)
And so to see us through the next year or more, we must all prepare for several cycles of a “suppress and lift” policy — cycles during which restrictions are applied and relaxed, applied again and relaxed again, in ways that can keep the pandemic under control but at an acceptable economic and social cost.
How best to do that will vary by country, depending on its means, its tolerance for disruption and its people’s collective will. In all cases, however, the challenge essentially is a three-way tug of war between combating the disease, protecting the economy and keeping society on an even keel.
Here is a formal framework for how governments could monitor the state of this pandemic much more accurately than many seem to be doing now, and how then, acting on the evidence, they could tune their interventions quickly enough to stay ahead of the outbreak trajectory.
For starters, one needs robust data. Policy must not be determined based on the daily count of reported cases — the tallies you read about constantly in the news — because those are unreliable. What’s needed instead is the coronavirus’s real-time, effective reproduction number, or its actual ability to spread at a particular time. And one needs to understand that number properly, in context.
The rate at which a virus is transmitted — known as the R-naught (R0), or basic reproductive number — refers to the average number of people to whom an infected person passes on the virus in a population with no pre-existing immunity. The R0 can vary from place to place because of the population’s age structure and how frequently people come into contact with each other.
The “effective” version of that number, the Rt — or the reproductive number at time “t” — is the virus’s actual transmission rate at a given moment. It varies according to the measures to control the epidemic — quarantine and isolation protocols, travel restrictions, school closures, physical distancing, the use of face masks — that have been put in place.
Daily reported cases do not convey the true state of the virus’s spread. For one thing, there is so much heterogeneity in the per capita testing capacity of countries around the world that it would be foolhardy to try to draw any broad conclusion about the virus’s transmissibility from all that disparate data. For another, the figures for reported cases lag actual infections by at least 10 to 14 days.
That’s because the incubation period for COVID-19 is about six days. And because — partly given shortages of test kits in many countries — some people don’t ever get tested, and those who do probably don’t until they have displayed symptoms for a few days.
However, it is possible to bring the daily count of reported cases closer to the real-time Rt thanks to both statistical adjustments and digital analytics.
The School of Public Health at the University of Hong Kong has been estimating, and publishing, the real-time Rt for Hong Kong since early February. The chart is based on the epidemic curve corrected by established statistical methods to reduce the time lag between the onset of infection symptoms and the official reporting of new cases. (The result is called “nowcasting.”) We hope to soon be able to further enhance these estimates by incorporating location-based data from the Octopus card that many Hong Kongers use to pay for public transport or to shop.
In China, the location-based functions of the online payment platforms of Alibaba, Baidu and Tencent could be used to track people’s activity. In the West, data feeds from Facebook and Google could geo-code online searches and payments. Citymapper, a mapping and public transit app, follows people’s movements in major cities in real time.
Activity data mined from all these apps and platforms, as well as records from payment cards, could be used to determine how people mix — which in turn could be used to infer the likelihood of their passing the virus around. In a recent contribution to the journal Science, Caroline Buckee described how all this data could be marshaled to chart a real-time map analyzing how physical distancing policies are affecting people’s movements.
With a bit of ingenuity, existing digital tools can quickly be turned into epidemic-monitoring instruments — and without intruding into people’s lives. Those who, as a general matter, worry about invasions of privacy (and rightly so) need not in this case: The idea is to only study aggregate, and therefore anonymous, numbers — to look at big data, not at personal information or anyone’s identity.
Then, having determined what the Rt actually is, decisionmakers could more precisely adjust their interventions to keep that number at what is, for them and their constituencies, an acceptable level.
An Rt of 1 means that the epidemic is holding steady: For every person who is infected, another one becomes infected, and as the first one either recovers or dies, the second one replaces it; the size of the total pool of infected people remains the same. At a rate below 1, the epidemic will fade out. Above 1, it will grow, perhaps exponentially.
That said, an Rt of 1 or below will not do in all circumstances. Context matters, too.
An Rt of 1 might be acceptable in a place with 10 million people if, say, no more than a couple of dozen new infections are confirmed every day. But it wouldn’t be if an epidemic were raging there and several hundred or thousands of new cases occurred daily. In the face of an explosive outbreak, the authorities would first need to take a sledgehammer to the Rt to knock it down to a very low level — 0.1 or 0.2 — and maintain it there for as long as it took to bring the daily case count down to a manageable figure.
In other words: Each community must determine the real-time effective reproductive number it can accept given its own circumstances, in particular the stage of the epidemic it is at.
Still, for all communities that determination essentially requires doing the same thing: Figuring out the number of new daily infections that their health system can handle without imploding.
Imagine a city that has 1,000 beds in intensive care units. It cannot have more than 1,000 people on a respirator at any given time. If the average length of a patient’s stay in the ICU is 14 days, this city cannot provide intensive care for more than about 71 new patients a day (1,000 / 14 = 71.42). Assuming that about 5% of all newly infected cases are so severe as to require intensive care, then the city cannot afford to have more than a total of about 1,420 new infections a day (71 x 20 = 1,420). This is the true number of infections, only a fraction of which are reflected in the officially reported count.
The authorities, having established the number of new infections the city’s emergency health facilities can support, can then determine what Rt they should aim for and tune their interventions to reach it.
Next, once it is clear what the health care system can bear, one must ask what the economy and, separately, what the people, can accept.
Even if the health care system can just about tolerate 1,420 new infections a day, would Wall Street? Would the financial markets — and, more important, the real economy — be spooked? Or react as they do during a bad flu season?
And how long can the population accept the restrictions required to maintain that level of infections? Will people stop complying? Are their mental and emotional well-being being jeopardized?
There is no right or wrong answer about the best way to respond to a threat as great and as complex as this pandemic. One can imagine a variety of individual views: “I’d rather protect the economy and take a chance with the epidemic”; “I’d rather take no chance and allow the economy to tank, partly because I’m sure it will bounce back in a year”; “I’m already going crazy after one week of lockdown; I can’t see myself sticking to this for three months.” This range is the reason the general public, especially in Western democracies, should have a chance to shape this discussion.
And yet, even though different communities will strike a different balance between these interests, the “suppress and lift” strategy is generalizable to all.
After achieving a sustained decline in the Rt and bringing the number of daily new cases down to an acceptable baseline thanks to stringent physical distancing, a society can consider relaxing some measures (say, reopen schools). But it must be ready to reimpose drastic restrictions as soon as those critical figures start rising again — as they will, especially, paradoxically, in places that have fared not too badly so far. Then the restrictions must be lifted and reapplied, and lifted and reapplied, as long as it takes for the population at large to build up enough immunity to the virus.
Trying to see our way through the pandemic with this “suppress and lift” approach is much like driving a car on a long and tortuous road. One needs to hit the brakes and release them, again and again, to keep moving forward without crashing, all with an eye toward safely reaching one’s final destination.
Gabriel Leung (@gmleunghku) is an epidemiologist and dean of medicine at the University of Hong Kong. He is the founding director of the World Health Organization Collaborating Center for Infectious Disease Epidemiology and Control and is an adviser to the Hong Kong and Chinese governments on the new coronavirus.