ABSTRACT
Pulmonary function tests (PFT) are sometimes monitored during treatment with known pulmonary toxic drugs to detect asymptomatic drug-induced interstitial lung disease (DILD). We conducted a systematic review to assess the accuracy of PFT, including the diffusing capacity for carbon monoxide (DLCO), for early detection of DILD in a range of drugs. Using a pre-specified, registered review protocol, OvidMEDLINE and EMBASE were searched from 1946 to February 2018. Two reviewers independently screened abstracts and reviewed full-text articles for inclusion. Included studies were assessed for risk of bias using adapted QUADAS-2 domains and primary outcome data were extracted and entered into RevMan5 to estimate sensitivity and specificity with 95% confidence intervals (CI). The search identified 4065 citations and included 42 studies. The most commonly studied drugs were bleomycin and amiodarone. Due to clinical heterogeneity between studies, a pooled analysis was not performed. Sensitivity of monitoring with DLCO varied between 0 and 100%, with the majority of studies finding a sensitivity of <80%. CI were wide for the majority of studies. Specificity was less than 90% in all studies. Risk of bias was high for the majority of studies for the quality domain of reference standard. The findings of this review do not support routine PFT for early detection of DILD. Due to methodological limitations, the relatively small number of participants and the low prevalence of DILD in the included studies, there remains significant uncertainty about the sensitivity of PFT to screen for DILD.