AUTHORS
Shaniqua Johnson Anum, MD, Qin SUN2, Nicole Herrera, MPH3, Matthew Fenchel, MS4, Travis Cleaves, CSM5, Jeffrey Lebensburger, DO6
AFFILIATIONS
1 Children's of Alabama, Birmingham, Alabama, United States, University of Alabama at Birmingham, Birmingham, Alabama, United States
2 Cincinnati Children's Hospital Medical Center, Cincinnati, Ohio, United States
3 ASH Research Collaborative, Washington, District of Columbia, United States
4 Cincinnati Children's Hospital Medical Center, Cincinnati, Ohio, United States
5 ASH Research Collaborative, Washington, District of Columbia, United States
6 Children's of Alabama, Birmingham, Alabama, United States, University of Alabama at Birmingham, Birmingham, Alabama, United States
ABSTRACT
Introduction: Acute chest syndrome (ACS) remains a leading cause of morbidity and mortality in sickle cell disease (SCD). Landmark cohorts defining ACS epidemiology pre-date hydroxyurea, newer disease-modifying therapies, and current transfusion and antimicrobial practices, limiting contemporary applicability. No consensus definition exists for real-world identification of severe ACS across institutions, limiting early risk stratification and potentially contributing to disparities in care. The ASH Research Collaborative® Data Hub (ASH RC Data Hub) is a high-quality, real-world clinical resource containing electronic health record-derived data from hospitals across the United States, including more than 14,000 individuals with a physician-attested SCD diagnosis. This multicenter resource provides an opportunity to develop a scalable definition of severe ACS in a contemporary cohort, addressing a critical gap in the current treatment era. We hypothesized that older age, physician-attested HbSS subtype, and prior venous thromboembolism (VTE) would be associated with severe ACS.
Methods: We performed a retrospective cohort study using ASH RC Data Hub data (2019-2025) from individuals of all ages with a physician-attested SCD diagnosis and ≥1 inpatient ACS hospitalization during the observation period. We developed an ACS computable phenotype requiring SCD-specific ICD-10 codes, pneumonia or respiratory proxy codes, or antibiotics associated with ACS treatment (azithromycin, moxifloxacin, or levofloxacin) during an eligible hospitalization; ACS hospitalizations not meeting severe criteria were classified as non-severe. We defined severe ACS as an ACS hospitalization requiring endotracheal intubation or mechanical ventilation (invasive or noninvasive). Exchange transfusion was not included in the severe ACS criteria because it could not yet be reliably ascertained across sites. At the patient level, cases were anchored at their first severe ACS admission; controls, who had no severe ACS admission during the observation period, were anchored at their first non-severe ACS admission. Covariates were assessed during available observation up to 24 months before each patient’s anchor date; patients with less than 24 months of observable data before their anchor date were included and identified for sensitivity analysis. VTE history was assessed at any time before the anchor date using a PI-reviewed, ASH RC-maintained ICD-10/SNOMED code set. Severe ACS frequency was calculated at the event level (proportion of ACS hospitalizations meeting severe criteria) and patient level (patients with ≥1 severe ACS hospitalization among patients with ≥1 ACS hospitalization). Bivariate comparisons used chi-square tests for categorical and Wilcoxon rank-sum tests for continuous variables.
Results: We identified 3,586 patients with SCD and ≥1 ACS hospitalization, contributing 9,700 ACS hospitalizations. Severe ACS occurred in 2.5 per 100 ACS hospitalizations (event level); 176 patients (4.9%) experienced ≥1 severe episode (patient level). Patient-level severe ACS frequency was highest among adult patients (7.9%; ≥23 years) compared with transition-age (3.5%; 13–22 years) and pediatric (2.3%; 0–12 years) patients. In case-control analysis (176 cases and 3,410 controls), patients with severe ACS were older (median age 29 vs 19 years) and more often adult at anchor (67.6% vs 39.9%) (both p<0.001). The severe ACS cohort was also more often male (56.2% vs 48.1%; p=0.034) and more likely to have prior VTE (38.1% vs 6.4%; p<0.001). The distribution of physician-attested SCD diagnosis types differed between groups (p<0.001); HbSS subtype was less common among cases (64.2% vs 73.4%; p=0.007), contrary to our hypothesis. Admission white blood cell count, hemoglobin, and platelet count did not differ significantly between groups.
Conclusion: In this contemporary, multicenter cohort, ventilatory-support-defined severe ACS occurred in roughly 1 in 20 patients with ACS overall, rising to 1 in 13 among adults. Cases were older, more often male, and more likely to have prior VTE, while HbSS subtype was less common among cases. Exclusion of exchange transfusion may have underestimated the frequency of severe ACS and reduced power to detect associations. These findings support the feasibility of characterizing real-world severe ACS in the ASH RC Data Hub; further phenotype validation and adjusted analyses are needed before use for risk prediction.
DISCLOSURES
Shaniqua Johnson Anum, MD: No relevant conflicts of interest to declare.