(C-166) Safety of Transient Oral Corticosteroids Use in Mild-to-Moderate Asthma for Exacerbation Management: Methodological Review and Beyond

Background: Short-term oral corticosteroids (OCS) are the standard treatment for exacerbations in patients with mild-to-moderate asthma. With over 23 million adults living with asthma in the US, the safety of OCS remains a public health concern, yet observational studies have reported inconsistent findings for the risks of OCS use in asthma patients.

Objectives: We reviewed observational studies that examined the acute safety outcomes of OCS in patients with asthma, focusing on potential biases, to explain the variations in findings. We describe these biases and propose methodological improvements for robust causal inference to inform safer clinical practice.

Methods: We conducted a PubMed search to identify all population-based observational studies that examined the safety of short-term OCS use among asthma patients, published in the last 10 years. Each paper was evaluated for study methods, primary results, and biases, including confounding by indication, time-related bias, selection bias, and measurement error.

Results: We identified four observational studies that used patient data from the UK and Denmark, all adopting a cohort approach, with one incorporating a nested case-control analysis. All studies reported increased risks of several adverse events with OCS use, such as cardiovascular (CV) events, psychiatric events, pneumonia, and sepsis. For example, hazard ratios for CV outcomes associated with OCS use, compared with non-use, ranged from 1.4 to 5.1. We found that all four studies were potentially affected by confounding bias, with its extent varying with the choice of covariates used in adjustment. Three studies used cohort definitions that failed to distinguish short- and long-term OCS users, leading to potential effect-modification and exposure misclassification, undermining the estimation of acute effects. One study was also affected by immortal time and selection bias.

Conclusions: The four studies of the safety of OCS to treat exacerbations in patients with mild-to-moderate asthma have significant methodological limitations that can explain their divergent results. These include confounding, immortal time bias, and selection bias, hindering their ability to inform on safe clinical decision-making. More rigorous designs that avoid these issues should be considered. For example, the case-crossover approach can control for measured and unmeasured time-invariant confounders and minimize selection bias, as it does not require the identification of comparator individuals. Additionally, target trial emulation using, for example, a prevalent new-user design, can help mitigate time-related biases, enhance confounding control, and improve risk estimation when comparing OCS users with non-users.