Application of the DIVERSE framework to describe ten European data sources applying the ConcePTION Pregnancy Algorithm

In this Data Sources Table we describe 10 European data sources included in a study that aims to test the ConcePTION pregnancy algorithm (https://github.com/ARS-toscana/ConcePTIONAlgorithmPregnancies). The ConcePTION Pregnancy Algorithm is a meta-algorithm designed to extract from any instance (i.e., a data cut, Thurin NH, et al. Clin Pharmacol Ther. 2022 Jan;111(1):321-331) of a population-based data source the list of pregnancies that occurred to persons included in the instance. The test of the ConcePTION Pregnancy Algorithm is fully described in a manuscript, for which this document is supplementary material. The description of the data sources was based on the DIVERSE framework. This framework was introduced as a standardized tool to describe Real-World Data Sources (Gini R, et al. Pharmacoepidemiol Drug Saf. 2024 May;33(5):e5787). The framework consists of nine dimensions: Organization: Description of the organization that makes the data accessible for research Data originator: Description of the organization that collects the data and for which purpose Underlying population: Description of the event(s) that cause persons to be included in the data source population Prompts: Description of the event(s) that prompted the recording of the data Content: high-level description of the of the information captured in the data Data dictionary: Description of the data dictionary, including coding systems or free text Time span: description of the time span when the data source is available Healthcare and culture: Description of the healthcare system and/or the culture of the area where the data source is generated Data quality: Description of aspects of data quality of the data source In our table, we also applied the concept of data bank developed in the ConcePTION project: each data source is split into data banks, that is a collec­tion of structured healthcare data, sustained by an organiza­tion, which mandates that a record is prompted whenever a specified class of events occurs to persons belonging to a specific populati­on (Thurin NH, et al. Clin Pharmacol Ther. 2022 Jan;111(1):321-331). For our study, we restricted this report to the data banks that were needed to test the ConcePTION Pregnancy Algorithm. We then applied the nine dimensions separately to each data bank. We also focused the content of the dimensions to our study purpose: for example, in the case of the dimension ‘healthcare and culture’, we focused on aspects of the local healthcare system and culture that have an impact on the generation of data for pregnant people. By describing the European data sources participating in our study using such dimensions, we were able to provide a complete picture of the strengths and limitations of the data sources for our specific purpose and support interpretation of results based on possible biases induced by the nature of the data. To populate the nine dimensions of the DIVERSE framework, we used information stored in the ConcePTION and VAC4EU Catalogues (https://vac4eu.molgeniscloud.org/VAC4EU/catalogue; https://conception-acc.molgeniscloud.org/Conception/catalogue), and in the recently updated HMA-EMA Catalogues of Real-World Data Sources and Studies (https://catalogues.ema.europa.eu). The Data Experts and Access Partners of each data source reviewed the information and provided expert opinion.