Epidemiology Professor Marc Lipsitch Talks Covid-19 Surveillance Strategies


Harvard epidemiology professor Marc Lipsitch discussed lessons from the Covid-19 pandemic regarding strategies for disease surveillance at a Harvard School of Public Health seminar on Wednesday.

Lipsitch, who has led the Center for Forecasting and Outbreak Analytics at the Centers for Disease Control and Prevention since last year, reflected on strategies used by different countries to track the spread of Covid-19.

Lipsitch said most countries, except for Luxembourg and the United Kingdom, relied on case reporting as the “basic unit of surveillance” during the pandemic.

However, he noted that the number of reported cases is an “unreliable measure of disease activity” because people may not be incentivized to take a Covid test or report a positive result.


“This is a very strange kind of quantity to report, because it’s not an epidemiological quantity such as incidence or prevalence or duration,” he said. “But it’s a complex function of those quantities.”

Lipsitch instead praised the surveillance strategies used in Luxembourg and the U.K.

He said Luxembourg randomly sampled a “substantial fraction” of its “small population” each week to measure the spread of the virus. This strategy allowed Luxembourg to track the effectiveness of its countermeasures against Covid-19.

The U.K. used two surveillance methods early on in the pandemic, according to Lipsitch. One was repeated at different points during the pandemic, while the other tracked households over time.

Lipsitch said the methods used in Luxembourg and the U.K. are based on the belief that the “best way” to track infection in a country is to measure the spread in a random sample.

Lipsitch also highlighted the importance of data completeness. For instance, Lipsitch said the U.S. is missing data examining the race and ethnicity of those infected with SARS-CoV-2.

Still, there are ways to “squeeze” additional details from “the admittedly imperfect data that is gathered in a place that does not have random sampling,” according to Lipsitch.

Lipsitch said studies from the past year show that taking data from hospital populations can serve as a cheaper alternative to random sampling from the general population.

Lipsitch highlighted that seemingly constant factors — such as genetic sequence, illness severity, and contagiousness — can change over time as the virus mutates and the population develops resistance.

“What Covid has taught us is that actually all of these are surveillance questions,” he said. “All of them are changing through the pandemic.”