My dear old statistics teacher used to say that relying on any model, however good but founded on past data, was like driving by looking at the rearview mirror; fine as long as the future looked like the past. As governments struggle with their response to the COVID-19 pandemic, the $64 trillion question is "which past does the future look like?"
Science writes how models have become supremely important:
Entire cities and countries have been locked down based on hastily done forecasts that often haven't been peer reviewed. ... The Netherlands ... Prime Minister Mark Rutte rejected “working endlessly to contain the virus” and “shutting down the country completely.” Instead, he opted for “controlled spread” of the virus while making sure the health system isn't swamped with COVID-19 patients.
Other governments have chosen different models and in place of "controlled spread," bet on total lockdown.
Just how influential those models are became apparent over the past 2 weeks in the United Kingdom. ... a group at Imperial College London [recommended an approach] not unlike the strategy the Netherlands is pursuing. ... But on 16 March, the Imperial College group published a dramatically revised model that concluded—based on fresh data from the United Kingdom and Italy—that even a reduced peak would fill twice as many intensive care beds as estimated previously, overwhelming capacity. The only choice, they concluded, was to go all out on control measures.
But as new data becomes available the "fog of war" is slowly lifting and rekindling the debate between the models. "This week, the number of coronavirus tests in the United States surpassed those in South Korea and Italy — two countries that had been testing more aggressively," the New York Times reported. With this and other new data, the true parameters of the COVID-19 outbreak will be re-estimated providing a new "rearview mirror picture" to steer by. White House coronavirus task force member Dr. Deborah Birx said:
I'm sure you have seen the recent report out of the U.K. about them adjusting completely their needs. This is really quite important. If you remember, that was the report that says there would be 500,000 deaths in the U.K. and 2.2 million deaths in the United States. They've adjusted that number in the U.K. to 20,000. Half a million to 20,000. We are looking at that in great detail to understand that adjustment.
So these are the things we are looking at, because the predictions of the model don't match the reality on the ground in China, South Korea or Italy. We are five times the size of Italy. If we were Italy and did all those divisions, Italy should have close to 400,000 deaths. They are not close to achieving that.
Models are models. We are -- there is enough data of the real experience with the coronavirus on the ground to really make these predictions much more sound. So when people start talking about 20% of a population getting infected, it's very scary, but we don't have data that matches that based on our experience.
Our simulations are in agreement with other studies that the current epidemic wave in the UK and Italy in the absence of interventions should have an approximate duration of 2-3 months, with numbers of deaths lagging behind in time relative to overall infections. Importantly, the results we present here suggest the ongoing epidemics in the UK and Italy started at least a month before the first reported death and have already led to the accumulation of significant levels of herd immunity in both countries.
The single most important difference in the two paradigms is the relative weights placed upon the extent of pre-outbreak exposure. Sweden's scientific advisers believe that the exposure effect is substantial enough that the expected cases will not overwhelm its health care system.
"The government has only banned events with more than 500 participants...those who visit pubs and restaurants should be seated at a table rather than mingling at a bar, and that people taking public transport should "think carefully" about whether it's necessary. Those who fall ill with coronavirus-like symptoms need only wait two days after they feel well again before returning to work or school."
Time will tell whether the Swedes (and Dutch) have made the right choice. "The corresponding number of deaths in Sweden predicted using the UK [Imperial] simulations are much higher than the Swedish government’s simulations suggest. The reason appears to be that Swedish authorities believe there are many infected people without symptoms and that, of those who come to clinical attention, only one in five will require hospitalisation. At this point, it is hard to know how many people are asymptomatic as there is no structured screening in Sweden and no antibody test to check who has actually had COVID-19 and recovered from it."
It is possible that some worst-case scenarios envisioning deaths in the millions and hundreds of thousands will be revised downwards in the coming days and although these may be disputed, the availability of sufficient reliable data will reduce the uncertainty. With growing information in hand, a reasonable policy balancing work and quarantine can be devised on a sound basis. That won't make it easy but it will make it sane.
No comments:
Post a Comment