(092) Ursodeoxycholic acid may be a new protective factor for severe COVID-19 outcomes – evidence from a nationwide Swedish register-based cohort study within the SCIFI-PEARL project

Publication Number: 429

Background: It was recently reported (Brevini et al. Nature 2022) that ursodeoxycholic acid (UDCA ) used to treat cholestasis in liver or biliary disease, which can reduce the expression of angiotensin converting enzyme 2 (ACE2), could provide a protective effect for severe COVID-19 outcomes.

Objectives: We aimed to evaluate the potential effect of UDCA treatment on severe COVID-19 outcomes by comparing those outcomes in a group of UDCA exposed versus non-exposed with similar underlying diseases. We used the linked nationwide Swedish register database SCIFI-PEARL (Swedish Covid-19 Investigation for Future Insights – a Population Epidemiology Approach using Register Linkage).

Methods: In this nationwide cohort study, we included all SARS-CoV-2 test positive (TP) subjects aged 18 or above at TP during 2020-2022, who had been diagnosed with primary biliary cholangitis (ICD-10 code K74.3), cirrhosis of liver (K74.6), autoimmune hepatitis (K75.4), or cholangitis (K83.0) from 2015-01-01 to TP. The UDCA exposed group consisted of those who had filled prescription within 6 months before TP. Studied outcomes were COVID-19-related hospitalization, 30-day mortality after TP, and COVID-19 cause-specific death. Subjects were followed from TP to the earliest of: outcome, end of 2022, 1 year after TP, emigration, or death.
Cox regression was used for each outcome to estimate adjusted hazard ratios (aHR) with 95% confidence intervals (95%CI), including adjustments for vaccination status, COVID-19 waves, prior comorbidities, and sociodemographics, providing a conditional ATE (average treatment effect in everyone) effect estimate. Weighted propensity score analyses (WPSA) were used as an alternative to estimate the average treatment effect in the treated (ATT). Doubly robust method was applied by additional inclusion of all the covariates in the WPSA.

Results: The study cohort comprised 833 exposed and 3638 non-exposed individuals.
The Cox regression showed increased risk of hospitalization (aHR 1.19, 95%CI 0.92-1.53), and decreased mortality (0.62, 0.34-1.13) and death (0.70, 0.37-1.32) in the exposed. WPSA achieved adequate balance and showed similar results for all outcomes (hospitalization 1.15, 0.82-1.61; mortality 0.78, 0.40-1.53; death 0.86, 0.43-1.71). Doubly robust showed also similar results (hospitalization 1.16, 0.83-1.62; mortality 0.79, 0.43-1.44; death 0.91, 0.43-1.91).

Conclusions: The results are supportive of a moderate protective effect mainly for mortality and death, but not statistically significant. The results from different analytical methods were consistent. Further analyses are planned to use various cohort definitions.