Appendix 3: Establishing Work-Related Causality

Occupational Health professionals will, from time to time, be required to make a judgement call regarding a putative exposure and an adverse health outcome and the question which arises is whether or not the exposure really caused the disease. Some times this is simpler than in others such as where a very close relationship exists between an adverse health outcome and a particular exposure eg. mesothelioma and asbestos dust exposure. However, do we establish this with entities for which many possible explanations exist such as anaemia, asthma, dermatitis and other target organ effects such as hepatitis, neuritis etc. the answer does not lie solely in statistics but also in applying a certain degree of logic common sense. The Bradford Hill criteria for assessing causality were established to attempt to address this very issue. These are listed as follows:

1. Strength of association
2. Consistency
3. Specificity
4. Appropriate time relationships
5. Biological gradient
6. Biological plausibility
7. Experimental support
8. Analogy
9. Coherence of the evidence

For ease of committing this to memory a suggested acronym is aggaccept, a= strength of association, g = biological gradient, a = analogy, c = consistency, c = coherence, e = experimental evidence, p = plausibility, t = time relationships.

So what do all these entities mean?

1. Strength of an association is generally measured in epidemiology by estimates of relative risk (RR). By convention a RR of greater than 2, points to the existence of a real relationship that does not occur due to chance alone.
2. Consistency: This term refers to the degree to which a finding can be reproduced in other situations such as different countries, different times and different work situations.
3. Specificity: This term refers to the theoretical "ideal" requirement that links an adverse health outcome with one specific exposure. The nearest example to this would be the relationship between asbestos and mesothelioma which, although very strong, is nevertheless not 100% specific. Specificity therefore is a theoretical criterion and is never likely to be fully met.
4. Appropriate time relationships: A proposed cause must precede an effect and the latent period between the earliest exposure and the effect manifestation should be what would be expected on the basis of the known biology of the disease in question.
5. Biological gradient: This term refers the dose response relationship with which the exposure and the outcome can be linked. Therefore, the likelihood of developing the adverse effect increases with the increased exposure. For example, this is one of the strongest elements of evidence linking lung cancer with tobacco smoking.
6. Biological plausibility: This term refers to the conformity of the outcome with the existing knowledge of the pathophysiology associated with the exposure in question. Therefore, this requires a detailed knowledge of the toxicology of the agent to which exposure takes place.
7. Experimental evidence: This term refers to the finding in the laboratory setting which ideally should support the findings in the clinical setting. This would include laboratory tests for carcinogenesis for example. One should bear in mind that the animal model or test tube circumstance is not necessarily true for the human model.
8. Analogy: This term refers to the evolution of similar adverse health outcomes under similar situations with other related chemicals. For example, in the case of benzene exposures, similar aromatics such as xylene and toluene should have not very different effects profiles.
9. Coherence of the evidence: This term refers to the coherence of all the elements of these criteria when they are summed up together. The more coherence that is found, the greater the strength of the association. However, notwithstanding this, the final decision is always a judgement call informed by epidemiological and other evidence and consequently retains an element of uncertainty.