A32-31 Prognostic Importance of Past Asthma Attacks on the Risk of Future Attacks: Findings From the Oracle2 Patient-level Meta-analysis

Rationale The strongest predictor of future asthma attacks is the history of an attack in the past year; however, the relative prognostic importance of specific features of a patient’s attack history—occurrence (none/any), number, or severity of prior attacks—remains uncertain. We hypothesized that number of attacks in the past year, rather than their occurrence or severity, would best predict future attacks after adjustment for key clinical and inflammatory risk factors. Methods We analyzed the OxfoRd Asthma attaCk risk scaLE (ORACLE2) dataset, a patient-level meta-analysis of the control arms of 22 randomized trials investigating the effects of fixed treatment regimens on asthma exacerbations over at least 24 weeks (PROSPERO-CRD42021245337). Multivariable negative binomial models were fitted to assess associations between asthma attacks in the past year, prior asthma hospitalizations, and prior ICU admission with the risk of future attacks. Multivariable models were adjusted for baseline blood eosinophils, exhaled nitric oxide, Global Initiative for Asthma treatment step, Asthma Control Questionnaire-5 score, forced expiratory volume in 1 second (FEV1), FEV1/forced vital capacity, sex, body mass index, smoking status, allergic rhinitis, chronic rhinosinusitis, and nasal polyposis. Pooled Wald tests and Nagelkerke R² served to determine whether finer categorizations of attack history variables improved model fit, and tests of trend were used to evaluate whether attack rates increased linearly with the number of prior asthma attacks and/or hospitalizations. Results Across 6,515 participants, the median [IQR] number of attacks in the year preceding enrolment was 2 [1]. Asthma-related hospitalizations and ICU admissions were reported for 15.4% and 4.7% respectively. After multivariable adjustment, the number of attacks in the past year remained associated with higher risks of future attacks, and more granular categorizations led improvements in fit (Figure). There was a linear relationship (ptrend<0.001), whereby every additional attack increased the rate of future attacks by 27% (RR 1.27 [95%CI 1.23-1.30]). In multivariable models, every additional asthma-related hospitalization was associated with a higher risk of future attacks (1.21 [1.14-1.28]); however, this association was no longer significant after adjusting for the number of prior attacks (1.02 [0.93-1.12]). In univariable and multivariable models, there was no association between prior ICU admission or intubation and asthma attack risk. Conclusions The number of asthma attacks in the past year provides better prognostic information on future risk than attack occurrence (none/any) or severity. Precisely quantified prior attacks, combined with other inflammatory and clinical characteristics, should inform risk stratification strategies in asthma. This abstract is funded by: Association Pulmonaire du Québec (APQ); Fonds de Recherche du Québec – Santé (FRQS), Québec AIRS Network; Academy of Medical Sciences (AcMedSci); National Institute for Health and Care Research (NIHR)