The Hill Criteria for Establishing Causation

Last week’s blog discussed the continuum of cause and effect from sources to adverse health effect from human exposure to chemicals.  The focus of that piece was to explore the situation where you have an adverse health effect in the work place but no understanding or even a signal of any of the elements that preceded it in the above line of cause and effect.   Within that situation, we are first charged with confirming that the adverse reaction(s) are real and caused by a workplace exposure.   Once we do that we are absolutely on the hook to figure out what is happening in order to determine a fix. 

This first step, that is determining that the effect is real and work-related, is critical and very often not obvious.   This week’s blog talks about making that connection.

My friend and colleague, Dr. Andy Maier, recently taught me about what are commonly called Hill’s Criteria of Causation [Hill 1965].  These are the minimal conditions needed to establish a causal relationship between potential disease agents and human diseases. They were originally presented by Sir Austin Bradford Hill (1897-1991), British Professor Emeritus of Medical Statistics of the University of London, as a way to determine the causal link between a specific factor and a disease.  I am not an epidemiologist but I now understand that Hill’s Criteria form the basis of modern epidemiological research and have been used in epidemiological science for sixty years.  

Clearly, the Hill criteria can also be used to establish causation in the above situation; that is, we have an adverse health effect in a worker or workers that we believe is linked to their work place exposure but need to establish (or deny) that link.   The individual Hill Criteria are listed below with a brief explanation for each:

TEMPORAL RELATIONSHIP

Exposure always precedes the outcome.  Note:  This is perhaps the most important of the criteria.

STRENGTH

This is defined by the size of the association as measured by appropriate statistical tests.  I think we all know that, strictly speaking, correlation is not causation; however, the stronger the association, the more likely it is that the relation of "A" to "B" is causal.

DOSE-RESPONSE RELATIONSHIP

An increasing amount of exposure to the putative cause increases the risk. If a dose-response relationship is present, it is strong evidence for a causal relationship.   

CONSISTENCY

The association is consistent when results are replicated in studies in different settings using different methods.  That is, if a relationship is causal, we would expect to find it consistently in different studies and among different populations.  If more than one worker experiences the same effect in a similar manner in similar circumstances.

PLAUSIBILITY

The association agrees with currently accepted scientific understanding of pathological processes. In other words, there needs to be some rational and theoretical basis for positing an association between a vector and disease.

CONSIDERATION OF ALTERNATE EXPLANATIONS:

In judging whether a reported association is causal, it is necessary to determine the extent to which you have taken other possible explanations into account and have effectively ruled out such alternate explanations. In other words, it is always necessary to consider multiple hypotheses before making conclusions about the causal relationship between any two items under investigation. 

EXPERIMENT

The condition can be altered or fixed (i.e., prevented or ameliorated) by an appropriate experimental regimen. 

SPECIFICITY

This is established when a single putative cause produces a specific effect.

COHERENCE

The association should be compatible with existing theory and knowledge. In other words, it is necessary to evaluate claims of causality within the context of the current state of scientific and technical knowledge.

Ref: Hill, A. B. 1965. “The Environment and Disease: Association or Causation?” Proceedings of the Royal Society of Medicine, Section of Occupational Medicine 58, 295 – 300.

As mentioned above, if you use the above criteria and determine that the untoward worker health effect is indeed related to their exposure you are then committed to hunt it down with an eye toward control.  Use the judgmental discussion of causality to form other hypotheses to reveal the driving determinants of this exposure and then test them.