(263) Using claims data to identify the exact indication of use for biological drugs approved for Immune-Mediated Inflammatory Diseases: The VALORE project experience
Assistant Professor Department of Diagnostics and Public Health, University of Verona, Verona, Italy
Background: Claims data have been used worldwide to identify both acute and chronic health conditions thanks to disease-specific case-identification algorithms combining information coded in claims databanks. Several algorithms were developed to identify Immune-Mediated Inflammatory Diseases (IMIDs) from claims data. However, biological drugs are usually approved for more than one IMID and an algorithm that identifies the exact indication of use for biological drugs approved for IMIDs treatment has never been developed.
Objectives: This study aims to develop and validate an algorithm capable to identify the exact IMID indication for incident biological drug users in the context of the Italian VALORE project.
Methods: All subjects with at least one dispensing of biological drug approved for the most frequent IMIDs such as rheumatoid arthritis (RA) psoriatic arthritis (PsA), psoriasis (PsO), ulcerative colitis (CU) and Crohn’s disease (CD) were identified from claims data of the Lazio region in Italy (observation period: 2010-2020). The date of the first biological drug dispensing was considered as the index date (ID) and incident biological users were defined as subjects without any dispensing of a biological study drug within the 365 days prior to ID. An algorithm that identifies the exact indication was developed using a combination of ICD-9-CM diagnostic codes, co-payment exemption and/or pharmacy claims. Positive predictive value (PPV) and sensitivity (SE) were estimated for each indication using as reference standard subjects with IMIDs as registered in the electronic therapeutic plans of Lazio region filled by the specialist during study period.
Results: Overall, 11,468 incident biological drug users with a single IMID indication were identified from Lazio region claims database (4,054 with RA indication, 1,973 with PsA, 2,698 with PsO, 1,562 with UC and 1,199 with CD). Estimated PPV and SE against the reference standard were 0.87 and 0.79 for RA, 0.84 and 0.66 for PsA, 0.45 and 0.87 for PsO, 0.64 and 0.95 for UC and 0.79 and 0.85 for CD, respectively.
Conclusions: Overall, the algorithm reported good PPV and SE. In particular, the higher estimates were found for RA and CD. However, to enhance the validation estimates and to confirm such results, sensitivity analyses are still ongoing and negative predictive value as well as specificity will be calculated.
