To ensure the protection of test subjects’ personal data, an algorithm links each test result to a unique, anonymized pseudonym. This algorithm has recently been updated so that it counts a patient only once even if they are tested multiple times within a short period, as can sometimes occur with enhanced monitoring of variants. The indicators have been recalculated for the entire country, which has allowed for the elimination of duplicates. The difference between the indicators calculated using the new and old methods is 12% for the incidence rate and 8% for the positivity rate. This difference does not alter the assessment of the epidemic’s dynamics; these changes enable the SI-DEP system to produce more accurate data and enhance its effectiveness.

Key steps in producing SI-DEP indicators: anonymizing data by creating a pseudonym and removing duplicates

The SI-DEP information system provides data to various entities with different objectives and needs: Santé publique France and the Ministry of Health for monitoring the epidemic; the Health Insurance system and the Regional Health Agencies (ARS) for contact tracing.

To ensure the protection of the personal data of those tested, an algorithm assigns each person tested a pseudonym calculated from the patient’s identifying information. This step is called “pseudonymization.”

Epidemiological indicators are then generated by Santé Publique France from this anonymized database (using pseudonyms) and processed to count only a single person when they are tested multiple times within a short period. This is the deduplication step.

Since its launch and the implementation of a large-scale testing policy, the SI-DEP system has incorporated, in addition to RT-PCR tests, antigen tests, screening of positive tests to detect variants, and saliva tests. A person may therefore be tested multiple times, in different locations, with different types of tests, and within short time intervals.

When a patient’s personal data was not entered exactly the same way, two different pseudonyms could be generated for a single person tested twice, without it being possible to identify them as a duplicate.

A single format for entering information about tested individuals

The collaborative effort to develop the solutions necessary for producing robust data involved revising the pseudonymization algorithm. Standardization of data entry (automatic capitalization, special characters, etc.) and simplification of the identifying characteristics used (birth name, first name, age, and gender) to generate the pseudonym were implemented to ensure that the same pseudonym is always generated for the same person.

These improvements now enable the SI-DEP system to produce even more accurate data and enhance its effectiveness.

Before open data: Verifying the reliability of the new indicators

The indicators were recalculated across the entire country using the new pseudonymization method. The correction was applied to a 3-month historical dataset, which corresponds both to the retention period for personal data and to the ramp-up of variant detection. A verification period was also necessary before publishing these indicators as open data to ensure their robustness and reliability.

The indicators recalculated using this new method show:

• A 12% decrease in the incidence rate for all of France

• An 8% decrease in the positivity rate for all of France

• A 6% decrease in the total number of confirmed cases

A comparison of the indicators produced using the old and new pseudonymization methods shows similar curves and trends with no impact on the dynamics of the epidemic, its monitoring, or its interpretation.

A look back at the work carried out with all stakeholders contributing to the operation of SI-DEP

Resolving discrepancies in the indicators generated by SI-DEP has mobilized all stakeholders—the Directorate General for Health, Santé publique France, the Directorate for Research, Studies, Evaluation, and Statistics (DREES), the Regional Health Agencies (ARS), the Paris Public Hospital System (AP-HP), and the Health Insurance Fund—who contribute to its operation to enhance its effectiveness.