Screening tools for work-related asthma and their diagnostic accuracy: a systematic review

Introduction One in four cases of asthma in adults is caused or worsened by work (work-related asthma: WRA). Early detection of WRA could prevent poor health and employment outcomes, but clinical diagnosis is often missed or delayed. The standardisation and effectiveness of screening tools have not been well established. We aimed to summarise and compare the performance of screening tools for identifying WRA in both clinical settings and workplaces. Methods We searched for studies that used structured questionnaires or prediction models (with/without physiological tests) to identify WRA in clinical settings or workplaces with individuals aged≥16 years in MEDLINE, Embase, other bibliographic databases and grey literature between 1975 and 2024. Two reviewers independently screened titles, abstracts and full texts for inclusion, extracted data and assessed risk of bias using QUADAS-2 tool (Quality Assessment of Diagnostic Test Accuracy 2) or PROBAST (Prediction Model Risk of Bias Assessment Tool). Screening tools and their indices of accuracy were summarised with paired forest plots of sensitivities and specificities. Results Of 17 504 identified studies, 7 were included. All were implemented in tertiary hospitals (n=5) and specialist centres (n=2). The screening tools comprised questionnaires alone (individual questions n=3 and multiple questions n=2), questionnaire with methacholine challenge test (n=1) and diagnostic models (n=4). The question ‘improvement off work’ had sensitivity=74–87% and specificity=15–58% for identifying WRA. Multiple questions had sensitivity=80–100% and specificity=8–55%. Addition of the methacholine challenge test to one questionnaire improved specificity to 75% with sensitivity=65%. Diagnostic models reported area under the curve (AUC) between 0.69 and 0.89, and AUC was improved when adding demographic variables or objective tests. Discussion A single item ‘improvement off work’ and multiple questions have high sensitivity but low specificity for WRA, which are sufficient for screening purposes to enhance WRA diagnosis. Adding demographic variables and objective tests can improve specificity or AUC. However, studies on screening tools for WRA are limited and inadequately reported; further evaluations of performance are needed in primary care populations and workplaces. PROSPERO registration number CRD42021246031