8 Bradford Hill Examples You May Not Know

Posted by on May 15, 2014

When challenging general causation arguments in toxic tort and product liability personal injury cases, more times than not, defense lawyers find themselves relying upon the criteria developed by Sir Austin Bradford Hill (see our related previous posts: 5 reasons to apply Bradford Hill criteria in your next case and Applying the Bradford Hill criteria to assess GMO safety). While these criteria should not be considered a checklist or a strict algorithm for determining whether an exposure is causally related to a disease, they certainly provide one of the most widely cited guidelines to evaluate causation based on observational epidemiological studies.

In making the case that the Hill criteria have not been met, defense lawyers often find themselves arguing over whether the Hill criteria are indeed the methodology of choice. These arguments often arise in admissibility hearings such as Daubert and Frye hearings. In order to support their general acceptance of the Hill methodology, it is often helpful to demonstrate instances where the Hill criteria have been used effectively in different scientific scenarios.

We provide below 8 instances where the Hill criteria have been used by scientists to make determinations about a causal inference (there are many others — this list is meant to be a sampling). It should be noted that this list is not meant to “endorse” any of the analyses in the examples provided below as reliable. Performing a reliable Hill criteria analysis is an extremely detailed endeavor and the devil is always in the details. Clearly, if a systematic and reliable method is not employed in selecting the individual studies that have gone into the analysis, the Hill analysis is not reliable.

Indeed, based on a cursory review of the evidence, we would strongly take issue with the conclusions reached in some of the examples we have provided below. However, the examples are provided to demonstrate that reliance on the Hill criteria is widely accepted by scientists to assess causation across a wide range of controversies and topics.

8 Instances Where Bradford Hill was Used to Assess Causation

Cobigo 2012Applying Hill’s Criteria to the Study of Autism Spectrum Disorders (ASD) and Exposure to Mercury

Cobigo et al. conducted a systematic review to examine the hypothesized link between ASD and mercury exposure.

To critically appraise the reviewed studies and to make an assessment of causation, Hill’s criteria for causation were applied. According to the authors, the reviewed studies failed to demonstrate strength and consistency of association as well as establish a temporal link between the onset of ASD symptoms and mercury exposure.

Thus, the authors concluded that the risk of developing and being diagnosed with an ASD as a result of mercury exposure remains unclear.

Cresswell 2012 – Dietary traces of neonicotinoid pesticides as a cause of population declines in honey bees: an evaluation by Hill’s epidemiological criteria

In this study, the authors use Hill’s criteria for causation as a structured process for making a judgment about the proposition that traces of dietary neonicotinoids in nectar and pollen cause population declines in honey bees.

Overall, the authors report that despite the “absence of decisive experimental results, the analysis shows that, while the proposition is a substantially justified conjecture in the context of current knowledge, it is also substantially contraindicated by a wide variety of circumstantial epidemiological evidence.”

And the authors conclude that dietary neonicotinoids cannot be implicated in honey bee declines. 


Freeman 2012 – Plasma levels of polychlorinated biphenyls, non-Hodgkin lymphoma, and causation

In a study by Freeman et al., the authors conducted a meta-analysis (of published case-control studies) to assess the strength of association between NHL and PCBs, followed by Hill’s criteria for causation to evaluate the possible causal relationship between NHL and elevated PCB levels.

According to the authors of this study, they believe their results indicate a general causal association between NHL and PCB exposure.

Davidson 2010 The Bradford Hill criteria and zinc-induced anosmia: a causality analysis

Investigators applied Hill’s criteria for causation to assess the relationship between intranasal zinc gluconate therapy and anosmia (inability to perceive odor). The authors conducted a small clinical study and a literature review before applying the Hill criteria.

According to the authors, clinical, biological, and experimental data support the Bradford Hill criteria to demonstrate that intranasal zinc gluconate therapy causes hyposmia and anosmia.

Perrio 2007 Application of the bradford hill criteria to assess the causality of cisapride-induced arrhythmia: a model for assessing causal association in pharmacovigilance.

The authors of this study applied Hill’s criteria for causation to assess the causality of cisapride-induced QTc interval prolongation/arrhythmia.

Literature searches were conducted, and investigators culled down the studies from about 200 to 70. Hill’s criteria for causation were evaluated using the seventy publications which consisted of 24 case reports, case series or spontaneous report summaries; eight epidemiological studies; 22 clinical studies; and 16 experimental (in vivo and in vitro) publications.

The most compelling evidence for an association between cisapride use and QTc interval prolongation/arrhythmia came from case/spontaneous reports and biological plausibility. However, the evidence provided by clinical studies was inconsistent, and epidemiological studies failed to demonstrate an association.

This study also demonstrated how different types of evidence found in pharmacovigilance can be evaluated using the Bradford Hill criteria.

Holt 2006 – Antipsychotic drugs and diabetes–an application of the Austin Bradford Hill criteria

Investigators assessed whether the association between antipsychotic drugs and diabetes is causative using Hill’s criteria for causation.

According to the authors, there is support of a causative relationship with regard to temporality for some cases of diabetes, and there is biologically plausibility. However, it was noted that the strength of association was weak, there is lack of consistency or specificity, and there is little evidence to support a biological gradient.

Overall, the authors concluded that the evidence surrounding a causative link between antipsychotic drugs and diabetes is inconclusive. Moreover, the risk is probably low and the attributable risk of developing diabetes is greater for traditional risk factors such as family history, ethnicity, obesity and ageing than it is for receiving an antipsychotic drug.

Morgan 2004 – Aspirin and colorectal cancer?

Morgan evaluated the use of aspirin in the reduction of colorectal cancer risk using Hill’s criteria for causation.

Overall, Morgan concluded that the evidence is suggestive of a reduction of colorectal cancer risk by perhaps 20-30% because some of the causality criteria were convincingly fulfilled.

Bergsagel 1998 – Benzene and Multiple Myeloma: Appraisal of the Scientific Evidence

In this study, the investigators sought to update, review, and summarize studies of the association of benzene with myeloma and to apply Hill’s criteria for causation in the evaluation of the role of benzene in the causation of multiple myeloma.

Overall, the authors concluded that based on their analysis, there is strong evidence linking high levels of exposure to benzene with an increased risk of developing acute myelogenous leukemia and the evidence for this association satisfies all of the Hill criteria. However, there is “no scientific evidence to support a causal relationship between exposure to benzene or other petroleum products and the risk of developing multiple myeloma.”

Did you like this post? We are here to help! Schedule a consult with one of our experts.

Just fill out the form bellow and we will contact you with more information.