Performance of the Self‐Controlled Case Series With Active Comparators for Drug Safety Signal Detection Using the French Administrative Healthcare Database (SNDS)

TL;DR

This study evaluates how well a statistical method called SCCS detects drug safety issues in a French healthcare database, finding that adding active comparators improves accuracy but lowers sensitivity.

Contribution

The study introduces the use of active comparators in SCCS for drug safety signal detection in the SNDS database, a novel application in this context.

Findings

Using active comparators increased specificity to 0.91 but reduced sensitivity to 0.52.

The SNDS database is effective for detecting drug safety signals, especially for hospital-captured outcomes.

A reference set of drug-outcome pairs improved the SCCS design's performance for signal detection.

Abstract

The self controlled case series (SCCS) is one of the most promising methods for drug safety signal detection using real world data (RWD), and incorporating active comparators could potentially improve its performance by addressing time‐varying confounding by indication. The ‘Système National des Données de Santé’ (SNDS) is a large nationwide administrative claims database, which has not been used extensively for drug safety signal detection. While comparable in size to other RWD sources, it is unclear to what extent the performance of SCCS correlates with that in other sources.

This study aims to evaluate the performance of the SCCS with and without active comparators for signal detection in the French administrative healthcare database SNDS.

We applied the SCCS to macrolide and fluoroquinolone antibiotics, using amoxicillin as the active comparator. Amoxicillin was chosen as an active comparator with similar indications. In total, 7 drugs and 30 outcomes from all organ classes were selected. We developed a reference set of 104 positive controls and 58 negative controls, using a taxonomy framework to ensure the selected drug outcome pairs are theoretically well suited to the SCCS design. The observation period lasted 2 years, with a 30‐day risk window after each dispensing. Diagnostic performance was measured using sensitivity and specificity with respect to the product labels.

The sensitivity and specificity of the SCCS without active comparator were 0.89 and 0.43, respectively, when limited to pairs with satisfactory power. Specificity increased up to 0.91 with active comparators; however, sensitivity decreased to 0.52.

The SNDS is a useful data source for signal detection, particularly for outcomes captured in hospitals. Using a carefully designed reference set of drug‐outcome pairs well suited to the study design, the SCCS achieved satisfactory performance for signal detection in this database. In this study, the use of active comparators improved overall performance at the expense of greatly reduced sensitivity.