Continuously adapting to track the epidemic’s dynamics as closely as possible

The surveillance system established by Santé publique France is constantly adapted to produce indicators that reflect the dynamics of the epidemic as closely as possible to reality and at the most granular levels of the territory. These adaptations are made taking into account the most recent scientific knowledge about the virus, changes in the measures taken to curb the epidemic, and international epidemiological standards. They may concern the system for reporting data to the agency—for example, the integration of TAg results into SI-DEP—or the methods for calculating indicators—taking into account results over a “rolling” 7-day period over time.

Integration of TAgs: Ensuring the Continuity of Information

Antigen tests have been authorized and reimbursed for the diagnosis of COVID-19 since October 17, with a phased rollout beginning on that date. They are now performed in clinical laboratories (LBM) or by other healthcare professionals (doctors, pharmacists, nurses, physical therapists, midwives, and dentists). Results of antigen tests conducted outside of medical laboratories have been entered into the SI-DEP system since November 16. Today, following preliminary analyses to verify the quality of the reported data, all test results—whether RT-PCR or antigen tests—are now included in the calculation of SI-DEP indicators (incidence rate, positivity rate, and testing rate). Note that while the inclusion of antigen tests automatically increases the incidence rates reported on Géodes as open data, the trend in the epidemic’s dynamics remains unchanged.

Number of people tested: what changes and what impact on the indicators?

Santé publique France’s epidemiological approach prioritizes person-centered indicators (incidence, positivity, and testing rates). Thus, SI-DEP data, which pertain to testing, have always been reprocessed to remove duplicate results from the same individual. Today, as the epidemic drags on, it is common for the same person to take multiple tests, particularly when previous tests were negative. Furthermore, our understanding has evolved, and the risk of reinfection—which is now considered very low but possible after 60 days—must be identifiable. The method for counting the number of people tested must therefore adapt to these changes to reflect the epidemic as closely as possible to reality.

This is why Santé publique France is now modifying the calculation method:

• the number of people tested: calculated over a given period (7 days, for example), it corresponds to the number of people who have had at least one test during that period and who have never tested positive in the preceding 60 days;

• the number of people testing positive: a person who tests positive either for the first time or more than 60 days after a previous positive test will be counted as a new case.

This new, person-centered calculation method is therefore more accurate for estimating the prevalence of the virus in the tested population. It results in changes to the calculation of the positivity rate and the testing rate.

The mathematical formula that combines the three main indicators (incidence rate, positivity rate, and testing rate) remains unchanged, however (Figure 1).

Figure 1: IR: incidence rate; PR: positivity rate; SR: screening rate.

In practice:

• Previously, when calculating the indicators:

  • Only people who tested positive for the first time since May 13 and those who tested negative for the first time since May 13 were included.

  • Thus, people who had tested negative multiple times were excluded, resulting in an increasing underestimation over time of the number of people tested.

  • This led to an overestimation of the positivity rate and an underestimation of the testing rate.

• Today, in the calculation of indicators:

  • On the one hand, people retested positive for the first time in more than 60 days will be taken into account, and on the other hand, all people tested according to the new definition.

  • The positivity rate according to this new calculation automatically decreases compared to that resulting from the old calculation; the testing rate, conversely, automatically increases. However, the trends over time for these two indicators are the same as those reported to date.

  • The change in calculation has no significant impact on the incidence rate, as this depends solely on the counting of people who have tested positive. The only change in this regard concerns people with two positive tests more than 60 days apart, who are very rare.

Before implementing this change, Santé publique France recalculated all indicators (incidence, positivity, and testing rates) produced since May 13 using these new methods, at every relevant territorial level (region, department, municipality, IRIS).

The curve shown (Figure 2) illustrates, for example, that positivity rates were very similar regardless of the calculation method used up to week 32 (August 9), after which a discrepancy began to emerge due to the very large number of tests conducted and the exclusion of individuals who tested negative multiple times.

Figure 2: Comparison of positivity rates calculated using the new and old definitions of a tested person.

“In a very short period of time, the SI-DEP system had to open up to new users and adapt to new testing strategies as well as to the prolonged nature of the pandemic. These changes took place without decision-makers or the French public ever being deprived of the information essential for managing the pandemic. Today, the system is fully operational, with the added benefit of facilitating international comparisons.”
Laetitia Huiart, Scientific Director at Santé publique France.