Spirometry quality control: man versus machine

ABSTRACT

OBJECTIVE: To compare the efficacy of local quality control (QC) of BOLD (Burden of Obstructive Lung Disease) Trinidad and Tobago study with spirometer machine quality grades as well as with BOLD Pulmonary Function Reading Centre QC standards at Imperial College, London. METHODOLOGY: Quality of the randomly chosen two hundred spirometries was independently assessed by the local team based on the 2005 ATS/ERS spirometry standards. Central QC quality assessment results were blinded to the local team. However, the machine quality grades were readily available to both the central and local teams during interpretation. Local and machine QC performances were correlated with the central QC which was taken as the gold standard for this study. RESULTS: According to central QC, 17.5% of the spirometries were declared as unacceptable. The noted common errors were prolonged peak expiratory flow time(14.5%), submaximal blasts (11.5%), variable efforts (11%) and poor reproducibility (10%). Cronbach’s alpha assessment revealed an overall reliability of 0.82 and a correlation of 0.73 between central and local QC. The correlation between central and machine QC was poor(0.548). Local QC exhibited 89.1% sensitivity and 94.3% specificity relative to the central QC. Though the machine QC displayed a better sensitivity (97.6%), its specificity was significantly lower 45.7%). CONCLUSION: Local QC was effective when compared with the central QC and executed better performance than the machine’s QC. This expertise can now be utilized for the future spirometry based local research studies. Due to limitations in identifying various errors, machine QC should not solely be used as the determinant of acceptability.