Making Sense of COVID-19 Statistics
by Park Dietz, MD, MPH, PhD
The news is filled with statistics about COVID-19, many of which do not mean what you may believe they mean, and some of which are devoid of meaning.
Any educated consumer of information—and certainly anyone who has taken an introductory course in statistics, sociology, political science, or epidemiology—would know that frequency data (e.g., the number of people with a disease) can’t be interpreted to estimate disease risk without considering the number of people in the population (e.g., the number of Covid-19 cases per 100,000 population). Yet the overwhelming majority of news reports announce the total number of confirmed cases, the total number of deaths, or the daily number of new deaths. Even when these numbers are reasonably accurate, which is not always true, they are far less informative than the rates per 100,000 population or rates per million population.
Once official guidance supports reopening, law firms with offices in multiple cities, counties, states, or countries need to compare rates per capita to have any chance of gauging which offices face higher or lower risks to inform their decisions about when and which offices to re-open. You should expect state, city, and county governments to require you to devise site-specific risk assessments for each office, to design a re-opening plan for each office, to provide training for all employees in each office, and to place masks, gloves, and approved sanitizing supplies in strategic locations in each office. In California, this is already required by the California Department of Public Health and CalOSHA, and we’ve been through this exercise in re-opening our own office earlier this week. Additional guidance from CDC is expected to be released soon and is unlikely to be less burdensome, particularly if it conflicts with local guidance.
The popular Johns Hopkins website favored by the media (https://gisanddata.maps.arcgis.com/apps/opsdashboard/index.html?mc_cid=a614d98a15&mc_eid=55126b62d9#/bda7594740fd40299423467b48e9ecf6) thus far provides only frequency data, not rate data, and thus has limited utility for comparing states or countries.
Comparisons between states are possible with data published here: https://www.worldometers.info/coronavirus/country/us/
Of the nine columns of data currently reported on this website, the single most useful column is the total deaths per million population, even though there are no doubt inconsistencies in the attribution of deaths to COVID-19. The second most useful column is cases per million population, even though this figure reflects how many tests have been administered, biases regarding which people have been tested, variations in which tests have been used, and the inaccuracies of all existing tests. These data are updated throughout the day and will continue to change, but for illustrative purposes let’s look at the data as of this writing (mid-day 5/14/20), comparing the 10 highest death rate states with the 10 lowest death rate states (while recognizing that large and populous states have wide variations among cities and counties):
The reasons the correlation between the death rate and the case rate is so imperfect include differences among states in the stages of the pandemic in which they find themselves, in part because of differences among states in population density, demographics, health, compliance with guidelines, and perhaps health care resources. For all the imperfections of the data, one cannot help but notice that as of today, New York has a death rate 117 times higher than that of Wyoming and a case rate 15 times higher. While certain precautions should be used universally to reduce transmission, it would be irrational to limit freedom of movement or economic activity universally without regard to these variations in rates.
Comparisons among cities or counties currently require calculating rates yourself, as these data are not currently published, with many counties publishing frequency data but not rate data. Comparisons between countries are even more challenging because of country-to-country differences in how cases and deaths are measured, and because of data manipulation for strategic and political purposes.
If your firm would like assistance in preparing site-specific risk assessments, plans, or training, or in comparing offices or geographic regions, PD&A’s epidemiologist, Ed Maes, Ph.D., is available for consultation.