PDF pdf 353KB

The Impact of Ambiguity Over Clinical Status As Predictor of Healthcare Treatment Outcomes

This paper considers the impact of \emph{Knightian uncertainty} or \emph{ambiguity} about the reliability of a patient's acquired comorbidities and risk factors as predictors of treatment outcome on the optimal time to initiate treatment. I show that high levels of such ambiguity is detrimental to patient welfare. Hence, learning about the clinical state (comprised of the comorbidities and risk factors) as a predictor of treatment outcome in order to resolve, at least partially, this ambiguity is crucial to improving their welfare.

The learning is achieved via a sequential hypothesis test in which the clinician will only treat if her ambiguity about outcome is sufficiently low; i.e., below some threshold which is derived based on the cost of making a wrong decision by treating. I show that learning in this way does indeed improve patient welfare with respect to the optimal treatment (timing) strategy.

The paper concludes with a discussion on the practical considerations for clinicians, how they can use these results in managing patient care, and notes that the results support specialisation across hospitals so that certain treatments are only carried out a small number of specialist centres.