Orlistat and the risk of acute liver injury: self controlled case series study in UK Clinical Practice Research Datalink

Adam Schiff raises an interesting question. Ideally we would want to see whether liver function returns to normal whilst orlistat is continued (suggesting the event was not caused by orlistat, or was even improved by orlistat), or whether resolution happens once orlistat is stopped (suggesting the event might be caused by orlistat). Such dechallenge information can be useful, though the interpretation is not straightforward as medications are frequently stopped or interrupted during adverse events they didn’t cause, and some causally related events resolve on continued medication. Nonetheless, this analysis would be relatively easy to determine in the setting of a randomised trial with stringent procedures for patient follow up and the careful recording of resolution dates for adverse events.

However, the general practice records we used for our study do not allow us to make these kinds of inferences as resolution is not easily determined. It could be indicated by a normal test result, but also by a lack of further testing, in which instance a resolution date cannot be defined. Equally, a lack of further testing would not signify the event had resolved, as it could also indicate a continuing event with poor monitoring, or a patient who chose not to return for further testing. Unfortunately we cannot distinguish between these possibilities and so the suggested analysis is not possible.