‘Inference to the best diagnosis’: HIV antibody-testing and regulatory expertise

I analyse United Kingdom (UK) regulatory guidance that since 1985 has governed diagnostic testing for HIV. I report on a STS analysis of these texts and associated sources which uses a theoretical framework of regulatory expertise based on the work of Julia Black and Harry Collins. The findings show how the interpretative flexibility in test design was addressed for optimal diagnostic inference. UK regulators nationally deployed United States (US) algorithms from the mid-1980s by virtue of their links with US contributory experts who had acquired interactional expertise for population-wide diagnostics in signal detection theory (SDT) and clinical decision analysis (CDA). Key variables like test sensitivity, specificity and exposure risk, all of which impact on interpretive flexibility, became prospective factors within the newly emerging diagnostic algorithms as UK regulators developed contributory expertise. By the 1990s this situation enabled UK regulators to innovate new diagnostic guidance which radically departed from its US counterpart in terms of deploying simpler tests inside an 'alternative' diagnostic algorithm still used in the UK. Based on these findings, I argue that this diagnostic guidance advanced a novel form of 'inference to the best diagnosis' using heterogeneous data, like risk categories and laboratory-readings which, although now adopted internationally, is still poorly understood by sceptical users. I further argue that the above theoretical approach could facilitate better understandings, especially concerning diagnostic regulation as a form of contributory expertise which may draw on interactional expertise in several specialist fields.