# Applying High‐Dimensional Propensity Scores in a Study of Inhaled Corticosteroids and COVID‐19 Outcomes

## TL;DR

This study used high-dimensional propensity scores to reduce bias in assessing how inhaled corticosteroids affect outcomes in patients with COPD during the pandemic.

## Contribution

The study introduces the use of high-dimensional propensity scores to better adjust for unmeasured confounding in pharmacoepidemiologic research on ICS and COVID-19.

## Key findings

- Conventional PS-weighted models showed weak evidence of increased hospitalization risk among ICS users.
- HDPS-weighting reduced the estimated risk of hospitalization and death, moving results closer to no effect.
- HDPS results varied with the number of covariates, emphasizing the need for sensitivity analyses.

## Abstract

In pharmacoepidemiologic studies of COVID‐19, there were concerns about bias from residual confounding. We investigated the effects of inhaled corticosteroids (ICS) on COVID‐19 outcomes, applying high‐dimensional propensity scores (HDPS) to adjust for unmeasured confounding.

We selected patients with chronic obstructive pulmonary disease on 01 March 2020 from Clinical Practice Research Datalink (CPRD) Aurum, comparing ICS/LABA/(+/−LAMA) and LABA/LAMA users. ICS effects on the outcomes COVID‐19 hospitalisation and death were assessed through IPT‐weighted and unweighted Cox regression. HDPS were estimated from primary care observations, prescriptions and hospitalisations. SNOMED‐CT codes and dictionary of medicines and devices codes from CPRD Aurum were mapped to International Classification of Disease 10th revision codes and British National Formulary paragraphs, respectively. We estimated propensity scores (PS) combining prespecified and HDPS covariates, selecting the top 100, 250, 500, 750 and 1000 covariates ranked by confounding potential.

When excluding triple therapy users, conventional PS‐weighted estimates showed weak evidence of increased COVID‐19 hospitalisation risk among ICS users (HR 1.19 [95% CI: 0.92–1.54]). Results varied slightly based on the number of covariates included in HDPS (HR using 100 HDPS covariates excluding triple therapy 1.01 [95% CI: 0.76–1.33], HR using 250 HDPS covariates excluding triple therapy 1.24 [95% CI: 0.83–1.87]). Conventional PS‐weighted models showed weak evidence of a harmful association of ICS with COVID‐19 death when excluding triple therapy users (HR 1.24 [95% CI: 0.87–1.75]). HDPS‐weighting moved estimates toward the null (HR using 250 HDPS covariates excluding triple therapy 1.08 [95% CI: 0.73–1.59]).

HDPS may have better controlled confounding for COVID‐19 deaths in this case. HDPS results can be sensitive to the number of covariates included, highlighting the importance of sensitivity analyses.