The basic principles of epidemiology are summarised by CDCD: count, divide, compare, then discuss differences. Unfortunately discussion of COVID-19 has widely ignored this approach.

First, we must count something. Examples of things you might count include deaths, cases of disease, uses of a health service, occurrence of a symptom, or the presence of a risk factor.

Then we divide by the population at risk to produce a rate, or ratio. This could be an incidence rate, where new cases are divided by the population at risk. It could be point prevalence, where the total number of cases at a given point in time is divided by the population at risk.

We can determine mortality rate by dividing the total number of deaths by the population at risk of getting the disease. Fatality rate is the total number of deaths divided by the total number of people who have caught the disease. The standardised mortality ratio is the total number of deaths divided by the number of deaths to be expected, by applying expected mortality rates to the population at risk.

We then must compare different populations, and different subsets of the population.

Finally we discuss the differences, considering not just the explanation that fits a particular, preconceived theory, but the wide range of possible confounding factors.

To count COVID-19 properly, we would need to test widely. It would be extremely helpful to repeatedly test a stratified sample of the population to measure the true incidence and prevalence of infection, and the changes over time within subgroups.

Some antibody tests have been judged not to be accurate enough for diagnostic use because of false positives and false negatives. However, they may be accurate enough for statistical use if the results are not going to influence the behaviour or treatment of individuals, and if the inaccurate results have a predictable and unimportant effect on the statistics.

Until we have some measure of the incidence in the total population, it is impossible to calculate the fatality rate, or to assess the reasons for the gender and ethnic differences that are emerging in mortality rates. That in turn makes it impossible to consider a rational strategy for exiting the lockdown.

News reporters presenting accounts of the number of cases country by country may have remembered to compare, but they seem to have forgotten how to divide.

It is being widely reported that the US is performing worse than the UK in the coronavirus epidemic because it has twice as many deaths. As it has five times the population this is actually a substantially lower mortality rate, not a higher one.

Of course, when we discuss this difference, we might conclude that the lower rate is not due to the US doing better but to it being behind us on the curve. It may also be because cases are not counted properly, or because the rate is higher in some parts of the country, but yet to spread to the rest.

Why is the fatality rate lower in Germany? It may be due to a greater proportion of cases being counted. However, that would lower the fatality rate but not the mortality rate, so why does Germany have fewer deaths?

If it is because the incidence rate is lower, why? It has been suggested that the lower case-fatality might be because a greater proportion of infections in Germany are in younger people, but if that is the explanation, we must consider whether Germany has a younger population (which it does not), or whether older people are better shielded (this may be the case).

Why do men have higher mortality rates from COVID-19 than women? Is it due to behavioural differences like smoking rates, or intrinsic sex differences? Are the apparent ethnic differences real, or are they due to confounding factors like deprivation or environment?

All of these questions have implications for how we manage the epidemic. Unless we remember to count, divide, compare and discuss differences we cannot answer them. There is a lot of comparing and discussing going on. It will be idle speculation until we remember to count and divide.

*Dr Steve Watkins is the vice-president of Doctors in Unite*

## Published by