Hampering the debate over what action to take on the coronavirus, and over which epidemiological model is the most accurate, is a shortage of evidence. Evidence includes the infectiousness of the virus, how readily it’s transmitted, whether infection confers immunity and, if so, for how long. The answers to such questions can only be obtained from individual testing. But testing has been conspicuously inadequate in most countries, being largely limited to those showing symptoms.

We know the number of deaths, those recorded at least, but a big unknown is the total number of people infected. This “evidence fiasco,” as eminent Stanford medical researcher and epidemiologist John Ioannidis describes it, creates great uncertainty about the lethality of COVID-19 and means that reported case fatality rates are meaningless. In Ioannidis’ words, “We don’t know if we are failing to capture infections by a factor of three or 300.”

The following table lists the death rate, expressed as a percentage of known infections, for the countries with the largest number of reported cases as of April 16, and the most recent data for testing rates (per 1,000 people).

As Ioannidis emphasizes, the death rate calculated as a percentage of the number of cases is highly uncertain because of variations in testing rate. And the number of fatalities is likely an undercount, since most countries don’t include those who die at home or in nursing facilities, as opposed to hospitals.

Nevertheless, the data does reveal some stark differences from country to country. Two nations with two of the highest testing rates in the table above – Italy and Germany – show markedly distinct death rates – 13.1% and 2.9%, respectively – despite having not very different numbers of COVID-19 cases. The disparity has been attributed to different demographics and levels of health in Italy and Germany. And two countries with two of the lowest testing rates, France and Turkey, also differ widely in mortality, though Turkey has a lower number of cases to date.

Most countries, including the U.S., lack the ability to test a large number of people and no countries have reliable data on the prevalence of the virus in the population as a whole. Clearly, more testing is needed before we can get a good handle on COVID-19 and be able to make sound policy decisions about the disease.

Two different types of test are necessary. The first is a test to discover how many people are currently infected or not infected, apart from those already diagnosed. A major problem in predicting the spread of the coronavirus has been the existence of asymptomatic individuals, possibly 25% or more of the population, who unknowingly have the disease and transmit the virus to those they come in contact with.

A rapid diagnostic test for infection has recently been developed by U.S. medical device manufacturer Abbott Laboratories. The compact, portable Abbott device, which recently received emergency use authorization from the FDA (U.S. Food and Drug Administration), can deliver a positive (infected) result for COVID-19 in as little as five minutes and a negative (uninfected) result in 13 minutes. Together with a more sophisticated device for use in large laboratories, Abbott expects to provide about 5 million tests in April alone. Public health laboratories using other devices will augment this number by several hundred thousand.

That’s not the whole testing story, however. A negative result in the first test includes both those who have never been infected and those who have been infected but are now recovered. To distinguish between these two groups requires a second test – an antibody test that indicates which members of the community are immune to the virus as a result of previous infection.

A large number of 15-minute rapid antibody tests have been developed around the world. In the U.S., more than 70 companies have sought approval to sell antibody tests in recent weeks, say regulators, although only one so far has received FDA emergency use authorization. It’s not known how reliable the other tests are; some countries have purchased millions of antibody tests only to discover they were inaccurate. And among other unknowns are the level of antibodies it takes to actually become immune and how long antibody protection against the coronavirus actually lasts.       

But there’s no question that both types of test are essential if we’re to accumulate enough evidence to conquer this deadly disease. Empirical evidence is one of the hallmarks of genuine science, and that’s as true of epidemiology as of other disciplines.

Next: Does Planting Trees Slow Global Warming? The Evidence