AbstractThe reporting of blood cell morphology, by biomedical scientists using microscopy, is a subjective and relatively uncontrolled process; morphology reports impact directly upon the clinical care of patients, however, no large studies of the processes morphologists employ to reach their conclusions have been undertaken.
This thesis chronicles the collaborative process with a national provider of quality assessment services (UK NEQAS(H)); detailing the pioneering developments that culminated in the creation of a national scheme, incorporating digital images of peripheral blood cells and accredited for continuous professional development. Annual exercises, using digital images distributed via the internet, were used to develop, test and create the scheme. Two workshops provided early assessment and feedback from participants. The aim of this research was to then evaluate the responses of the large number of professionals who completed cases, to give insights into how they interpret the blood cell morphology to produce their succinct report.
The responses of between 732 and 1,018 participants (median 878) were examined for five digital morphology cases specifically selected to cover a range of morphological features. The subsequent data examination shows that patterns of error and bias were found in the responses that have not been described in blood film reporting before. Where a single morphological abnormality existed (glandular fever or Pelger-Huët anomaly), the ability to identify the feature of interest was high (97% and 84% respectively), however, errors in knowledge-based classification were seen. For complex cases, with multiple abnormal features, additional errors of inattention and premature completion were found; in the case of lymphoma with oxidative haemolysis 68% correctly reported the acute haemolysis, however, only 17% correctly reported both abnormal pathologies. Heuristic methods of decision-making, not considered in morphology reporting before, help the understanding of these patterns of error and bias.
Following this research the national scheme will be adapted to support participants by indicating the potential common forms of error found in morphology reporting.
|Date of Award||Sep 2017|
|Supervisor||Graham Mills (Supervisor) & Gavin Knight (Supervisor)|