Abstract
Rationale: Guidelines recommend systemic corticosteroids and inhaled β-agonists for patients with severe asthma exacerbation who are admitted to intensive care units. The benefits and utilization of adjunct treatments after guideline-recommended first-line treatments have been initiated are unclear. Objectives: Examine practice patterns of adjunct interventions in US intensive care units (ICUs) and their associations with outcomes for adults with severe asthma exacerbations. Methods: Using the multicenter PINC AI Healthcare Database of Premier Inc. (2016-2022), we sought to explore the use of adjunct interventions (medications [e.g., magnesium, leukotriene inhibitors, terbutaline, heliox] and procedures [e.g., invasive and noninvasive mechanical ventilation]) for adult patients admitted to U.S. ICUs with acute asthma exacerbations. We used hierarchical generalized linear models to calculate risk-adjusted rates of adjunct interventions and quantified between-hospital variation in adjunct interventions using the intraclass correlation coefficient (ICC; higher values correspond to higher between-hospital variation). We then used K-means clustering to identify groups of hospitals with similar risk-adjusted practice profiles of all adjunct treatments and examined associations between identified hospital clusters and patient outcomes. Results: We identified 62,392 patients from 961 hospitals for inclusion. Adjunct interventions with the highest between-hospital variation after risk adjustment were heliox (ICC, 91%), inhaled steroids (ICC, 23%), invasive mechanical ventilation (ICC, 21%), terbutaline (ICC, 22%), paralytics (ICC, 16%), and noninvasive ventilation (ICC, 15%). K-means clustering identified two distinct hospital clusters: Patients who were admitted to Cluster 1 hospitals (399 hospitals) had higher risk-adjusted rates of noninvasive ventilation (51% vs. 33%), compared with patients who were admitted to Cluster 2 hospitals (234 hospitals), which had higher risk-adjusted rates of invasive mechanical ventilation (63% vs. 30%). Cluster 2 was associated with fewer hospital-free days (β = -0.75 d; 95% confidence interval [CI] = -0.95, -0.55) and increased in-hospital mortality (adjusted odds ratio, 1.28; 95% CI = 1.17, 1.40). Conclusions: The use of adjunct interventions for patients with severe asthma exacerbations vary widely across U.S. hospitals; however, hospitals generally fall into two clusters differentiated primarily by the use of invasive or noninvasive mechanical ventilation. The cluster favoring noninvasive mechanical ventilation was associated with improved outcomes. Our results help to inform usual-care arms of future comparative effectiveness studies and efforts to standardize asthma practice.