Integration of ISO 15189 and external quality assurance data to assist the detection of poor laboratory performance in NSW, Australia

A systematic survey (SS) of the peer-reviewed literature was conducted to identify the key international themes that govern quality management for laboratories. Informed by the survey findings, integrated models utilising assessment results against the ISO 15189 standard, and data from external quality assurance (EQA) programs, were developed to predict laboratory performance. Via the PubMed database, a SS of the international pathology quality literature identified over 100 articles, which were subsequently subjected to text mining and meta-analyses via R statistical programing. Word patterns were examined for indicators of current best practice in quality assurance. Random Forest (RF) and ANCOVA models were subsequently developed with combined ISO 15189 standard and EQA data obtained from 21 anonymous pathology laboratories in NSW. The SS and associated text mining showed no consistent international consensus, but a significant minority (15%) of articles suggested root cause analysis as a means of exploring quality problems. Using the RF algorithm, an integrated ISO 15189 external audit—EQA model was developed, with results further supported by ANCOVA. The combined RF—ANCOVA method succeeded in identifying EQA markers [e.g., serum potassium (K+)] that correlated with ISO 15189 external audit results, providing an integrated predictive model of laboratory quality more statistically robust than proposed initiatives to apply root cause analyses as a means to systematically monitor laboratory performance.