FOCUS: Management of an epidemic requires surveillance monitoring

By Fred Palm, contributing editor  |  Our South Carolina emergency plan is an all-event skeleton.  Depending upon the particular threat, customization to the plan is made. Even in the overall skeleton plan, any epidemic event is a second thought found in an appendix (14-1) to the general model for action.  In these plans, there is no pandemic appendix, so with COVID-19, we presumptively start as if the virus is an infection.

An epidemic requires a swift model that leaps ahead of the presenting of requests for medical services. State-level authorizations and equipment requests passed up the line will not be delivered in time. In fact, little time exists if an infection doubles every five days.

Social science surveys estimate the size of something, the incidence in the population. In this election season, for example, we are bombarded with polling data about what percentage of voters, likely Republicans or Democrats, are expected to behave in a particular way. We are even told the error or range of the estimate.

Epidemiologists, however, model statistically the progressions that yield this critical understanding: A leap ahead in time and space to deploy resources and institute policy actions relevant to the situation on the ground.  For example, a varying disease incidence of 0.2 percent, 2 percent, 12 percent or 22 percent implies vastly different control strategies. Closing schools is not needed at the lowest end of the range and too late at the upper end. The anticipatory leap ahead forecast supports the needed adaptation policy choice.

Gov. Henry McMaster’s plan is to test only people with symptoms. It is good to know who is positive, so they isolate themselves. But if we stop there and do not systematically sample the entire population using a big fat surveillance program, it will be too late for actions needed to stanch COVID-19.  Instead, late actions will be out-raced by the fast spread of the virus.

Here’s what could help now:  Panels of volunteer citizens (say 300,000) could report their condition day by day for serial tracking using a free tool like SurveyMonkey. Out of those aggregated individual reports are identified the hot COVID-19 locations. Over time, we will learn how fast the spread rate is and the policy response that is needed there. This is called survey reporting, which hospital arrivals can confirm.

The novel coronavirus is a public health emergency that needs to use a war model, not the flood model.  War models anticipate the enemy while flood models respond to the situation. McMaster, apparently taking the lead of his political patron in Washington, D.C., to stay in his good graces, needs to do more.  He should call Dr. James Clements, the president of Clemson University, to gear up the Clemson Survey Team, the Palmetto Cluster Computing Center and the Clemson/MUSC Artificial Intelligence Project team. Then he should promptly hire a general who will win at war. State Epidemiologist Linda Bell needs to get test kits distributed widely and start surveillance monitoring as if our lives depend on it. Why? Because they do.  We need to know how big this storm is and where it is with better accuracy than day by day recording counts of people infected. 

Edisto Island resident Fred Palm is a retired professor of analytics at the John Jay College Graduate School of Public Management. Earlier in his career, he was a developer of the New York City Mayor’s Management Report, survey research director and academic research director who used multivariate clustering and regression modeling to predict behavior.