Saturday, April 11, 2020

Covid-19 models miss the point: “It's not over until it's over”


Gordon L. Weil

People are fascinated by numbers.

Not surprisingly, the Covid-19 pandemic has become entangled with statistics. The problem may be that people focus on those numbers, so they lose sight of the real problem.

How many cases have there been in China or Italy or the U.S.? Where is the pandemic “epicenter” based on the case count? What's the number of ventilators, face masks, or protective gear?

Experts have been busy building models to create forecasts of the possible number of new cases, recoveries and deaths. Daily press briefings are mainly about the latest counts and the expected shortfalls in equipment generated by models. Models seem to be used to frighten people.

Suddenly, we are expected to understand enough math to know that an important goal is to “flatten the curve.” The models produce a curve. What curve? What does flattening do? Whatever, let's just do it.

New York Gov. Andrew Cuomo heroically fights the country's toughest challenge, but he seems to believe he can scare people into reducing the spread by citing forecasts. In Washington, the federal government publicly revealed a controversial forecast showing a stunning quarter of a million dead.

Dr. Nirav Shah, the Maine CDC head, says he uses the models not to produce forecasts as their main purpose, but to plan for a range of possible health challenges. He understands that the models tell him that what might happen depends on what people do to reduce the spread.

That's why he reluctantly released model results. People are likely to believe models yield reliable forecasts, when they are really only a tool.

The most important data from a Covid-19 model is how many cases can be expected. Differing assumptions about people's behavior produce a wide range of results, but no single, reliable forecast. People influence the model not the other way around.

Social distancing, using face masks and hand washing matter more than models.

The simple lesson of the models is that more spread means more illness. That hardly requires a lot of detail about the numbers. And knowing the exact number of people who will be affected is impossible.

Shah makes one mistake when he says the Covid-19 forecasts are like weather forecasts. Weather forecasts are famously inaccurate, because conditions beyond our control continually change. But people can control Covid-19 models by cutting down on current cases. In fact, right now, that's about all that can.

People would be unwise to take comfort from models or even from the belief that “flattening of the curve” is the goal. We simply know too little and numbers produced by models may tempt people to believe their hopes.

Flattening the curve does not necessarily mean that fewer people will get Covid-19. It means that the number will be spread over a longer period, which will stress hospitals less and provide time to find a helpful medication or vaccine. That could save some lives.

“It's not over, until it's over,” said Yogi Berra, the philosopher-baseball player.

With stay-at-home orders to fight the spread, the economy loses both producers and consumers. It slows down. The government keeps it alive by pouring out money to people and companies.

President Trump and others see the stock market as an indicator of the health of the economy. Investors have extreme reactions to each day's model numbers and data reports, which they treat as a daily forecast. The market swings wildly.

Trump's re-election is reportedly dependent on the state of the economy. Like everybody else for their own reasons, he wishes for a speedy economic recovery.

Falling stock market indexes – numbers, again – cause some policy makers to press for an early finding that there is a cure for the virus and that the crisis is ending, freeing people from protecting themselves so they can get back to work.

Peter Navarro, a presidential advisor, says his economics doctorate makes him as much an expert on the virus models as the medical doctors. He asserts the crisis is not so bad as they say, so we should test and simply declare, “It's over.” He admits that some people will die. Economic recovery is apparently worth lives.

“Dr.” Rudy Giuliani, Trump's personal lawyer, claims that some sketchy data is enough to show an unproven drug is the Covid-19 cure, though scientists are still far from that conclusion. Fall ill, take the drug, and go back to work.

Navarro is flat wrong. Giuliani is dangerous. The health of the economy depends on the health of people.

The health of people depends on their protecting themselves and others and, in the end, on science, not math.

No comments:

Post a Comment