Reflections on interpretative ‘external quality assurance’ in histopathology

Abstract

Participation in interpretative external quality assurance (EQA) schemes has come to be accepted as a duty of the diagnostic histopathologist. EQA consumes significant resources and with a disciplinary component to most schemes, it is appropriate to ask if they achieve what the name offers. EQA is not an adequate model of good diagnostic practice. Diagnosis is Bayesian: probability evaluations ('priors') integrating the available information (clinical history, endoscopic appearances, imaging, biochemistry etc.) are updated by histological evidence, including consultation with colleagues, and the diagnostic histopathologist has the option of hypothesis testing. In EQA, priors are highly skewed, hypothesis testing is unavailable and consultation is disallowed. Since only a minority of laboratory errors are purely errors of interpretation, EQA as a name misrepresents what such schemes can deliver. Even if 'poor performance' in EQA were to reflect actual poor diagnostic performance in day-to-day practice, EQA schemes generate too few events, i.e. errors of interpretation, to have sufficient statistical power to 'assure' anything. Furthermore, some 'poor' and 'persistently poor' performance labels generated will be noise, not a genuine signal. Nevertheless, EQA is valuable, as a vehicle for self-knowledge. Having to make up our mind on challenging cases can reveal ignorance which we might prefer, but have no right, to ignore. Having discovered and acknowledged a deficiency, we can correct it. With EQA there is nowhere to hide, and it is vital for the personal development of the honest and conscientious diagnostic pathologist. But we must be realistic about what it can and cannot do: interpretative EQA is only a small component of error reduction in pathology.