Therapy and prognosis are entailed by the diagnosis: the holistic success of the EBVM approach therefore firmly and critically rests on diagnostic accuracy.
Unfortunately, medical professionals do not appear to be very accurate with diagnoses. In human medicine, there is 30-50% discordance reported between doctors’ ante- (presumptive) and post-mortem (definitive) diagnoses, with no significant change in the last 100 years (Goldberg et al 2002). This is attenuated by attaching a degree of certainty – ‘very certain’ shows 16% discordance, ‘probable’ 33% and ‘uncertain’ 50% – and some body systems are more difficult (e.g. respiratory) than others (Shojania et al 2002; Sington and Cottrell 2002).
Veterinary surgeons do not perform much better, although it is a chronically under-researched area. The single study that exists – from a respected referral institution – shows discordance between ante- and post-mortem diagnoses of ranging from 15% (oncology) to 45% (ECC), with internal medicine (44%), neurology (35%), surgery (33%) and cardiology (21%) lying in between (Kent et al 2004). Incorrect diagnoses are therefore common; the potential for subsequent incorrect or harmful therapy and/or prognosis is great; the quality of interventional evidence is immaterial if the wrong disease is being treated.
How can we do better? Human EBM shows that technology, big data and further evidence does not guarantee improvement; these are unlikely realisations for EBVM in the near future in any case. The answer may lie in the fields of psychology and social science. Studies indicate that diagnostic success may rest largely with the individual: expert clinicians consistently perform better. But how?
Experts are marked out by the use of ‘illness scripts’, which are mental knowledge networks against which the presenting patient – history, signs, clinical data – are checked and hypotheses entertained or refuted until (with the addition of more clinical data) a diagnosis is reached (Custers 2015). Experts rapidly appraise relevant data, embracing and exploring inconsistencies in the clinical picture, largely ignore pathophysiology and make far fewer errors than novices.
It appears we would do well to harness and teach these skills – however, current problem-based learning methods fail to do so. To maximise the impact on clinical performance, it seems explicit recognition, teaching and integration of illness scripts and pathways to expertise are fundamental steps for EBVM. This would require a ground-shift change in veterinary teaching and professional development, but may well prove to be the greatest positive difference EBVM can make to the veterinary profession.