JMIR Publications

ABSTRACT

Background:

Patients hospitalized for a given condition may be receiving other treatments for other contemporary conditions or comorbidities. The use of such observational clinical data for pharmacological hypothesis generation is appealing in the context of an emerging disease, but particularly challenging due to the presence of drug indication bias.

Objective:

With this study, we aimed at developing a fully data-driven pipeline that would address this challenge, and assess its clinical relevance on patients with COVID-19. Therefore, we applied the so-called pharmacopeia-wide association study, to analyze the associations between early - within 48h from admission - drug prescription and 28-day mortality in patients with COVID-19.

Methods:

We performed a multicenter retrospective cohort study using electronic medical records from 16 University hospitals of the Greater Paris area. We included all adult patients hospitalized in conventional wards during the first three waves of COVID-19. We investigated the association between early - within 48h from admission - drug prescription and 28-days mortality. We implemented a fully data-driven procedure based on adaptive LASSO to determine drug specific adjustment sets. We systematically computed several measures of association including robust methods based on propensity scores to control indication bias, which allowed the report of automated diagnosis. We validated against an expert-knowledge based pipeline on three treatments of references for which experts agreed on the expected association with mortality.

Results:

A total of 5,783 patients were included in the analysis. Median age at admission was 69.2 [IQR, 56.7 - 81.1] years old, and 3,390 (58.6%) of them were male. The performance of our automated pipeline was comparable or better for controlling bias than the expert-based adjustment set for three reference drugs: Dexamethasone, Phloroglucinol and Paracetamol. After correction for multiple testing, four drugs were associated with increased in-hospital mortality. Among these, Diazepam and Tramadol were the only ones not discarded by automated diagnostics.

Conclusions:

Our innovative approach proved useful in rapidly generating pharmacological hypotheses in an outbreak setting, without requiring a priori knowledge of the disease. Our systematic analysis of early prescribed treatments from patients hospitalized for COVID-19 showed that Diazepam and Tramadol were associated with increased 28-days mortality. Whether these drugs could worsen COVID-19 needs to be further assessed.