Data mining can support quality assurance

Abstract

Doctors Hockey and Marshall’s essay on the need to train doctors in quality improvement, citing its successful use by manufacturers and US doctors1 is timely as it is now nearly a century since the first suggestions that hospitals might use industrial methods of management and surgeons provide formal quality assurance. In the meantime manufacturers have evolved into world-class companies, able to meet global demand, by paring processes down to provide goods of a narrow albeit protean quality, ones which satisfy customer requirements at an acceptable price. Their lean processes are rigidly defined and incorporated into comprehensive systems of Total Quality Management (TQM) so no activity, be it lean manufacturing or continuous improvement, functions in isolation. But despite the opinion that TQM concepts apply equally to manufacturing and service industries no company providing personal healthcare has succeeded in joining the world-class paragons and there is a corresponding absence of TQM at unit level in healthcare. In the case of surgery, a service in some respects close to the ethos of the craft workshops which developed into factories, there are fewer than a dozen reports containing the terms ‘surgery’ with ‘TQM’ in Pubmed. Of these only one used outcome of introducing TQM as its topic (it found little effect).2 This paucity is in contrast with an abundance of publications about clinically-driven surgical quality. The difference in activity in a lively discipline with an obvious commitment to quality is likely to reflect difficulty of extending a linear model into an environment which is complex and adaptive.3 As this healthcare environment is also knowledge-based and creates vast amounts of data doctors with an interest in quality assurance should familiarize themselves with data mining for patterns of care as an aid to discovering better processes4 and meeting inequalities in care.5