As the low-risk group develops a so-called 'herd immunity' to covid-19, release the high-risk people from quarantine, suggest two Israeli computer science professors
When this corona interlude is finally done, what will be our most vivid memory of it? My answer, I’m coming to believe, is reading. Specifically, reading about corona. Every day there’s stuff being written and circulated about the pandemic: people wearing masks or not, the mathematics of growth and lockdowns, the search for a vaccine, the competence or otherwise of various world leaders as they face up to the disease, plenty more. In fact, I’ve given up trying to keep up with everything that comes my way. I’d need 30 hours a day, minimum, to do so, and which of us have that kind of time?
But I’ve been holding on to one article that someone sent me early in the pandemic, perhaps in early April. Something told me it would be good to revisit it in several weeks, if the pandemic did indeed last that long. That “if" is itself a testament to how we viewed covid then, back in those days when we had only a couple of thousand cases in India and it was possible to imagine we’d never be as badly hit as China or Italy.
Today, of course, India has over half a million cases, and over 16,000 deaths. We’ve long overtaken China and Italy, not forgetting Spain and South Korea, Madagascar and Mexico. Our case count is the fourth-highest in the world, and will move to third within the next week.
I can’t help thinking, should we pay attention to this article at all? Should we have paid attention back when we had only 2,000 cases, when we had just started on our lockdown? What about now, as we are apparently “easing" the lockdown?
The article was by two Israeli computer science professors, Shai Shalev-Shwartz and Amnon Shashua, and it is titled ‘Can we contain covid-19 without locking down the economy?’. To me, it was a good example of a phenomenon we’ve seen again and again during these last several months: mathematicians and computer scientists applying their skills to an epidemic, trying to find answers that purely biological or chemical analyses might not.
It begins by identifying three “models for handling the spread of Covid-19":
* A country-wide lockdown “until the spread of the virus is under control." This, the professors write, is the safest route but there’s the danger of a second wave of the disease when the lockdown is eased—as the US is finding out and parts of India are experiencing.
* A selective quarantine based on actual cases: locate everyone who has been infected and put them in quarantine. This is hard enough with just the cases already detected. But given the incubation period of the disease, it will also need authorities to do extensive contact-tracing to locate all those who might be infected over the next two weeks.
* A selective quarantine based on risk: divide the population into those at low risk of infection and those at high risk. Quarantine the high-risk section. Let those at low-risk go about their lives with measures like masks and distancing. As the low-risk group develops a so-called “herd immunity" to covid-19, release the high-risk people from quarantine.
The rest of the paper explores the third option, clearly the one the authors believe is the most efficient and will cause the least damage to a country’s economy. They use mathematics with a certain relentless logic. While some of that mathematics is beyond the scope of this column, I really want to give you a flavour of the logic.
Obviously, to answer many of the issues the analysis bumps into, we will need to understand how the disease spreads, which is a hard task. Instead, the authors make some assumptions and proceed with what they call a “worst-case" analysis. That’s a phrase familiar to every computer scientist: in writing software in particular, you are always trying to identify and allow for the worst case — or the extremes, or the boundaries — in any given situation. It’s when you do that that you can be sure your software works for the more routine cases.
To start, how do we make this division into high-risk and low-risk sections? Age is the easiest marker. Shalev-Shwartz and Shashua choose a cut-off of 67 years. These are essentially the people in society who have already retired. They are more easily quarantined too, because they anyway move around less than their younger compatriots.
Now if people at low-risk are going about their lives more or less as usual, some will certainly be infected and will need hospitalization. Two assumptions here in the paper: that all the low-risk people get infected and some fraction of them have symptoms severe enough to need hospital care; and that all of those need hospitalization at the same time. Both assumptions are unlikely to hold, but that’s just what’s meant by working with worst case scenarios.
If we measure the capacity of the healthcare system simply by the number of ICU beds available in the country, we have to be sure that this number is higher than the number of people who will need the beds. And what is the number of people who will need beds? That is, what is that fraction with severe symptoms?
Equivalently, what is the chance that someone from the low-risk group becomes severely ill?
To work that out, the paper takes into account the incubation period of the disease too. Why so? Because there are people who are actually infected today but, because of the incubation period, will only develop severe symptoms over the next several days.
Not only that, the paper also allows for those who are not infected today but might be infected tomorrow— the disease is spreading, after all— but show severe symptoms only days from now. Once more, this is the worst case: assume larger numbers than is obvious. Assume that the number of severe cases a week from now, say, includes both people who are already infected today and people who will be infected tomorrow.
If this is confusing, never mind. The point is, this analysis gives us a worst-case estimate of the chance of a low-risk person becoming severely ill. Multiply that by the size of the low-risk group, and we have a worst-case estimate of how many from that group will need hospitalization. This is the number that certainly needs to be less than the number of beds we have available. If we can assure that, we know that the healthcare system will be able to cope with the smaller severe case counts that we are more likely to face.
So let’s see: in March, India chose to go into a fairly tight lockdown. This has not stopped the spread of the disease—it never could have—but it has certainly slowed its spread. (The lockdown caused plenty of unnecessary misery too—the tragedy of migrants—but I’ll leave that for another column). That slowing was intended to “buy time" for us to ramp up our healthcare facilities to where we can handle all that this pandemic throws at us.
But now that we are looking at ways to open up, is this risk-based selective quarantine a reasonable way to proceed? That is, we quarantine our older population. We let the younger people function more or less as normal so that our economy starts recovering. Eventually, this younger fraction will have developed the famous “herd immunity". That’s when we let the older folks out of quarantine—into a population that is largely immune, in which the disease will have difficulty spreading. In effect, we have lowered the risk for this high-risk group.
Should we try this as we ease this lockdown?
Well, we need to be sure that as covid-19 spreads through the lower-risk group that won’t be in quarantine, our now ramped-up healthcare system will be able to cope with the severe cases we’ll get from that low-risk group. We need to be sure that it can cope with a worst-case number of younger people needing hospital care.
Nothing is ever sure, of course. But the logic Shalev-Shwartz and Shashua offer in their paper gets us close. If we can get numbers for how many beds are now available for potential covid patients, it’s worth running through their analysis to see if this is a possible route to take in dealing with the pandemic.
All because they considered, every time, the worst-case.
Once a computer scientist, Dilip D’Souza now lives in Mumbai and writes for his dinners. His Twitter handle is @DeathEndsFun
Subscribe to Mint Newsletters
* Enter a valid email
* Thank you for subscribing to our newsletter.
Never miss a story! Stay connected and informed with Mint.
our App Now!!