Retrospective diagnosis: Pitfalls and purposes

There are many arguments against the practice of retrospective diagnosis, also known as posthumous diagnosis or retrodiagnosis. These include the nature of the sources used (general absence of physical evidence; reliance on second hand evidence) and the context (historically contingent diagnostic categories; the subject’s different cognitive world), not to mention potential ethical issues (lack of “patient” consent), all of which render any conclusions speculative, even presumptuous. The clinical skill of pattern recognition may involuntarily suggest diagnosis, without sufficient attention to sources and context. Yet, both clinicians and historians still attempt retrospective diagnosis (in the interests of full disclosure, I must admit to being a serial offender, using both patient and literary narratives ¹ ), and pathography constitutes a well-recognised genre within medical biography. The matter is rendered urgent by Karenberg’s recommendation to journal editors to reject such papers when submitted for publication. ²

The issues involved have been carefully formulated by Muramoto, ³ who identified two challenges: the ontological and the epistemic. The ontological challenge questions whether disease entities persist over time. Evidently, diseases may come and go (e.g. epidemics) even though there is probably little change in human biology over historical time and culture (i.e. it is transhistorical and transcultural), albeit some predispositions or vulnerabilities to disease may vary between times and places. But if it be accepted, as seems likely to clinicians, that many disease entities did exist before their clinical description and incorporation into evolving nosologies, ⁴ the central question is therefore the epistemic challenge: how can diagnosis be empirically verified retrospectively?

This is, of course, rarely possible, although molecular genetic diagnosis using fortuitously preserved tissue (“molecular biography” ⁵ ) or archaeological findings (palaeopathology ⁶ ) may sometimes occur. However, Muramoto argues that all diagnoses are cultural constructs, hence liable to change over time and based on hypothesis-making and hypothesis-evaluation following iterative probabilistic Bayesian reasoning in circumstances of uncertainty, as used in day-to-day clinical practice. In consequence, retrospective syndromic diagnosis based on history alone is permissible as an explanatory device provided that overspecification is avoided.

Retrospective diagnosis does have purpose: To better understand an individual’s life and work and to correct or amend previous interpretations. Occasionally, an author may specifically indicate their writing is intended for scrutiny by clinicians (autopathography, e.g. Margiad Evans’ “desire to put into physician’s [sic] hands … clues to the feelings of such a sufferer [from epilepsy] as myself” ⁷ ), an invitation which should be accepted.

Hence, pace Karenberg, ² this editor will continue to consider papers suggesting retrospective diagnoses, applying the guidance suggested by Muramoto. ³ A good example of this approach has already appeared in the Journal.^8,9