View Abstract

A6071 - A Novel Clinical Score for Diagnosing Occupational Asthma
Author Block: M. Taghiakbari1, J. Pralong2, C. Lemiere3, G. Moullec4, P. Saha-Chaudhuri5, A. Cartier3, R. Castano6, J. l’Archeveque1, E. Suarthana7; 1Research Center, Hôpital du Sacré-Cœur de Montréal, Montreal, Quebec, Canada, Montréal, QC, Canada, 2Institute for Work and Health, Epalinges-Lausanne, and Division of Pulmonary Diseases, Geneva University Hospitals, Geneva, Switzerland, 3Department of Medicine, Université de Montréal, Montreal, QC, Canada, 4Département de médecine sociale et préventive, Université de Montréal, Montreal, QC, Canada, 5Department of Epidemiology, Biostatistics and Occupational Health, McGill University, Montréal, QC, Canada, 6Department of Otolaryngology, Université de Montréal, Montreal, QC, Canada, 7Département de médecine sociale et préventive, Université de Montréal, Research Center, Hôpital du Sacré-Cœur de Montréal, Montreal, QC, Canada.
RATIONALE: The diagnosis of occupational asthma (OA) is challenging since the specific inhalation challenge (SIC) as the reference test is not widely available. We aimed to develop models to predict the presence of OA (defined as positive SIC) by using clinical and exposure characteristics as well as objective tests other than the SIC.
METHODS: This retrospective study analyzed data from workers with suspected OA, who were exposed to high-molecular-weight (HMW, n=160) and low-molecular-weight (LMW, n=340) agents and still worked one month before SIC. They were referred to OA clinic at Hôpital du Sacré-Cœur de Montréal between 1983 and 2016. Logistic regression models were developed in each exposure group. Model discrimination was determined by the area under the receiving operator characteristic curve (AUC). The calibration and internal validity of the models were evaluated. The final models were translated into clinical scores and stratified into risk groups.
RESULTS: The prevalence of OA was 52.5% in the HMW and 23% in the LMW group. In the HMW group, the final model included sex, age (>40 vs. ≤40 years), symptoms duration (≤1 vs. >1 year), rhinoconjunctivitis (presence/absence), inhaled corticosteroid use (yes/no), non-specific bronchial hyper-responsiveness (NSBHR, with methacholine challenge test), and specific sensitization (with skin-prick test or SPT, yes/no). It had a reasonable calibration and internal validity. The AUC was 0.901 (95%CI: 0.85-0.95), which was statistically significantly higher than the combined NSBHR and specific SPT alone. The top tertile of the clinical scores (i.e. the high probability group) had a higher specificity (90.8% vs. 78.9%) and positive predictive value (90.1% vs. 82.0%) than the combined objective tests.
The final model in the LMW group consisted of sex, rhinoconjunctivitis, symptoms duration (≤1 vs. >1 years), exposure duration (≤7 vs. >7 years), and NSBHR. It had an AUC of 0.686 (95%CI: 0.63-0.74); statistically significantly higher AUC than NSBHR alone. The model was not well-calibrated, but had reasonable internal validity.
CONCLUSIONS: Our novel risk prediction tool showed excellent discrimination for predicting OA for HMW group. The discrimination is also reasonable for the LMW group. Our user-friendly risk score could be extremely helpful for physicians at the secondary care level (i.e. specialists) in identifying subjects who are at a high risk of having OA. Therefore, our tool will aid clinicians in deciding whether their patients should be referred to a tertiary center. Before implementing this tool in routine clinical use, external validation will be necessary.