Fluoroquinolones and the aortaBMJ 2018; 360 doi: https://doi.org/10.1136/bmj.k988 (Published 08 March 2018) Cite this as: BMJ 2018;360:k988
All rapid responses
Jeffrey Aronson draws a distinction between guidelines and criteria. I purposefully employed the latter because, as he has observed (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2677430/), the phrase "Bradford Hill criteria" is in common usage. This is especially true among clinicians. Because my commentary was written primarily for a clinical audience with the goal of helping readers feel more comfortable when evaluating an association generated by an observational study, reference to the "Bradford Hill criteria" is justifiable on the grounds of familiarity. In my view, the concepts and their application matter more than what we call them.
Competing interests: No competing interests
In his commentary on the research report by Pasternak et al. on a proposed association of fluoroquinolone antibiotics with aortic damage, David Juurlink refers to the "Bradford Hill criteria". But the items in Bradford Hill's 1965 list (which he called "aspects of an association" and "nine different viewpoints") are, all but one, guidelines, not criteria. Juurlink also writes that "temporality is Bradford Hill's only essential criterion"; this is a pleonasm--a criterion is essential by definition ("A test, principle, rule, canon, or standard, by which anything is judged or estimated", OED)--it requires to be met for the test to be passed. Temporality is the only item in Bradford Hill's list that is a criterion; the rest are guidelines.
This is not a mere verbal quibble. It emphasizes, as Bradford Hill did, that the items in his list do not definitively point to a cause and effect association.
It is also important to note that Bradford's Hill's items are asymmetrical, and that not only their presence but also their absence should be taken into account. For example, the absence of temporality is strong evidence against an association; its presence is of little confirmatory value. Conversely, the presence of similar effects produced by analogous interventions provides strong evidence in favour of an association; its absence is unhelpful. The presence of gradient (i.e. dose responsiveness) supports an association, but its absence is strong evidence against (https://www.ncbi.nlm.nih.gov/pubmed/26119837). Unfortunately, dose relations in pharmacological studies are, as here, rarely investigated.
Competing interests: JKA has written elsewhere about the Bradford Hill guidlines (e.g. https://www.ncbi.nlm.nih.gov/pubmed/19417051) and has published widely on adverse drug reactions.