NRPLUS MEMBER ARTICLE T he recent day-over-day increases in reported COVID-19 cases have generated public reactions that range from COVID fatigue to cautiousness to outright panic. In many places, the number of new cases being reported each day even eclipses the numbers seen at the prior peak of the pandemic in the U.S. this spring. Nationally, on June 26th, there were 40,588 new cases per day reported, compared to a peak of 44,726 on April 26th. The increases are being driven by certain states and regions. Two states that exemplify this increase are Texas and Florida. In Texas, on June 26th, there were 5,799 new cases compared with a previous peak of 1,801 on May 16. In Florida, the reported number of new cases per day increased from 673 to 8,942 over the same time period.
However, rising case numbers only paint part of the picture. Case numbers alone do not provide sufficient information on whether there is a true accelerating spread of infection or just greater detection through increased testing. Case numbers alone cannot reliably inform policy-makers whether the new cases will have the same impact on the health-care system and COVID-related deaths as cases detected earlier in the pandemic in previous hotspots such as New York City and Boston.
In order to know if there is truly increased spread, we need to supplement case data with information on the percentage of tests that are positive — the positivity. In an ideal world where we knew the infection status of the whole population, we would only need to monitor the percentage of the population who were infected each day to determine whether the spread of infection was accelerating. In the absence of these data, the most practical approach to assess true spread of infection is to look at the percentage of tests that are positive, and from this to infer the percentage of the population that is infected. Indeed, one of CDC’s gating criteria for opening the country is declining positivity over a 14-day period at a time when testing rates were stable.
However, the validity of any inference about the spread of infection based on changes in positivity depends on the degree to which the characteristics (e.g., age, severity of illness) of the people being tested do not change over time. The more the characteristics of people being tested change, the harder it is to determine whether changes in positivity are due to changes in the spread or changes in who is being tested (e.g., testing a higher percentage of symptomatic people).
There is good reason to believe that some of the decline we saw in positivity over the course of the pandemic was due to a change in those being tested. We know that, early on in the pandemic, when testing was very limited in the U.S. (~170,000 tests per day at the end of March), individuals with symptoms of COVID were prioritized in testing and many of those tested positive (~20 percent positivity). Earlier this month, as testing became much more available (~550,000 tests per day) the positivity rate was only about 5 percent. We assume that, with expanded testing, more asymptomatic people and people with milder illness were now making up a greater percentage of those tested than earlier in the pandemic. This is consistent with recent reports from several states showing that more younger people are being tested now than previously.
Of major concern is that, over the past few weeks, national positivity has increased from about 5 percent to more than 6 percent. According to the CDC, the number of tests conducted each day over the past three weeks has been relatively stable at between 500,000 and 600,000 tests per day. This suggest that the characteristics of those tested are not changing significantly, allowing inferences about the nature of the spread of infection to be more valid.
Applying this thinking to Texas and Florida, it seems that the characteristics of the populations affected by the spread of infection in Texas and Florida are similar, perhaps younger and healthier than reported earlier in the pandemic. This interpretation is supported by similar increases in positivity and limited increases in hospitalizations. The positivity increased from 6.6 percent to 13.7 percent over the past two months in Texas and increased from 4.5 percent to 15 percent in Florida. However, without information about who was tested in Florida and Texas, it is still hard to be fully confident in any conclusions about the nature of the spread of the infection in the two states.
In order to get even more useful information about who is being tested, public-health programs establish what are known as sentinel-surveillance systems in which health departments work with testing sites in key locations to support the collection of additional information on who is being tested. While sentinel-surveillance programs can only give a snapshot of all those being tested, if set up in the right places (e.g., urgent-care centers, primary-care practices, in communities with high numbers of infections and hospital emergency rooms), we can glean important information that supplements routine positivity information about who is being tested and furthers our understanding of where and how widely the infection is spreading.
The recent increase in COVID case numbers is very concerning, but limited in what it tells us about the nature of the current spread. The percentage of those tested and information on who is being tested are essential to effectively monitor on-the-ground conditions concerning the spread of COVID and its subsequent impact on both the general population and health-care. Armed with data on positivity from routine testing and from sentinel surveillance, along with data on deaths and hospital capacity, policy-makers can more appropriately scale up or scale back the extent of reopening and push for increased mitigation efforts, such as facial coverings and social distancing.