IPHAM Director's Message: COVID-19

April 1, 2020

Dear IPHAM,

The Kaiser Family Foundation publishes regular updates of state policies for social distancing to slow spread of the COVID-19 epidemic. Illinois is one of 28 states under a statewide stay at home order, one of 16 states with a ban on all gatherings of any size, one of 46 with mandated school closures, and one of 44 limiting restaurant service to takeout or delivery. When and how will we know if they are working? Should we consider stricter policies? Can we relax the policies we already have enacted?

Daily counts of newly confirmed COVID-19 cases is one of the most common data sources being used to monitor and forecast spread of the epidemic. These data are publicly available for all states and can be plotted to show daily trends in new cases. Unfortunately, I have some serious concerns about using these data in this fashion. We must consider how to improve the reporting and use of these data, as well as to identify alternative data sources to inform thoughtful policy decisions.

First, it is important to recognize that no state in the US has been performing population screening for COVID-19. Most states perform targeted diagnostic testing, primarily among only a small subset of people who have moderate to severe flu-like symptoms. According to public health agency reporting of completed COVID-19 tests, only 2.9 tests have been conducted for every 1000 people nationally. Illinois ranks 31st among all states, with testing still being restricted in many locations because of limited testing supplies. In other words, case reports of confirmed COVID-19 testing in Illinois still reflects only a small fraction of all people infected. 

A second concern I have is the use of public health reporting data to estimate changes in daily rates of disease when there appears to be inconsistency in daily reporting. For example, Illinois has reported results for more than 1000 tests every day since March 19th, but the percent of tests reported as positive each day has ranged from 8.6% on March 21st to 48.7% on March 29th. The daily positivity rate over this time appears to vary randomly, suggesting either large daily fluctuations in the conduct of testing (i.e. some days testing only people with high pre-test probability, while other days testing more indiscriminately), or inconsistent reporting of test results (particularly negative test results) from health systems to the state. Unless we can explain or eliminate the high variation in publicly reported test results, we must consider alternative data sources to understand and forecast the spread of the virus.

One interesting idea has been to track the effects of social distancing more directly by using continuously available spatial data from sources such as smart phones and other personal devices to measure how far people are traveling away from their homes. Unacast is a Norwegian based company that has been using such data to publish maps that depict changes in personal travel distances before and after social distancing policies have been implemented. Visualizing such metrics at a state or county level is one way to observe if such policies are keeping people closer to their homes. Today’s plot for the state of Illinois is pasted below, with the colored line indicating a 30 to 50% decrease in the mobility of our population since the epidemic began in our state. This seems encouraging.


 
Another novel monitoring approach involves tracking symptoms. San Francisco-based Kinsa Health has sold or given away more than a million internet-connected smart thermometers to households throughout the US. The company integrates temperature readings by geography to publish a “US Health Weather Map” that shows the prevalence of fever by geographic region of the US.  Time trends in smart thermometer readings for Cook County indicate a surge in febrile illnesses around March 10th, with a return back to expected levels for this time of year on about March 24th.  This also is encouraging.



A final alternative data source that may prove useful for improving how we track COVID-19 cases are electronic health records. Internationally, somewhere around 5 to 15% of cases require treatment in an emergency department or hospital setting. Though hospitalization rates for a specific geography will depend on characteristics of the population and local standards of medical practice, we can reasonably assume that as long as hospital beds and staff remain below full capacity, the numbers of people hospitalized each day will reflect some fixed percentage of the total number of cases of COVID-19 present in the community. In Chicago, several health systems that collaborate in the Chicago Area Patient-Centered Outcomes Research Network (CAPriCORN) are working towards a process to share de-identified electronic health records for patients hospitalized with COVID-19 throughout Cook County.  It is hoped that analysis of trends in the numbers and outcomes of these hospitalizations should inform better estimates of future cases and their implications for public policy and health system capacity. More information about this initiative and ways to contribute to our city’s response to the COVID-19 epidemic can be found on our Chicago COVID Coalition website.

Best wishes and stay healthy,  

Ronald T. Ackermann, MD, MPH
Professor of Medicine & Medical Social Sciences
Senior Associate Dean for Public Health
Director, Institute for Public Health and Medicine