The prickly topic of causation

By Dr John Wilson

In the doctor-patient relationship the question will often arise “Doc, what would have caused this?”  Of course, as doctors we are all expected to have an immediate recall of possible culprits. Causation is important, because if we can treat or remove the causative agent, we have gone a long way to solving the patient’s problem.  Often there are multiple culprits.  Consider ischaemic heart disease.  We can’t alter the genetic predisposition, but there are a number of other risk-factors that can be managed.

Note that I have introduced the concept of “risk factors”.  Risk factors apply to populations.  We know, that if we consider a thousand people who smoke and eat lots of fast-foods, they will fare worse than a sample matched for age and gender, who are non-smoking vegetarians.

But if we are to look at one of those persons with a poor lifestyle and outcome, can we confidently declare that eating fast-food has been the actual cause?  We could, however decide that it was a contributory cause, and we could add in another half a dozen possible culprits; and if we were really honest, we could finish off by saying there are probably dozens of yet to be discovered risk factors.  When we are dealing with a scientific (including medical) question of “what caused this?” it is an open-ended matter, because whilst we can approach 100% certainty, we can never achieve it.

Enter now the lawyer into the doctor-patient relationship.  Our patient, rightly or wrongly, has now got it into his/her head that ultra-violet radiation from the computer screen at their work-station has caused the coronary event.  This brings in another new concept, that of “attribution bias”. More about that in a future article.

Your patient’s solicitor is not concerned that there may be other causative agents – the legal process is constrained only in addressing the question at hand – has the UV exposure caused the coronary event? Or to be more precise, has on the balance of probabilities, the UV exposure caused the coronary event?

That would be the extent of it if your patient was self-employed and intent on suing the computer manufacturer at common law.  But he is an employee, and therefore eligible to claim under the worker’s compensation system. What I am talking about here are conditions with gradual onset. Obviously there is a different situation when considering acute injuries arising out of a single traumatic incident.  Then there are aggravations, which are another “can of worms”.  This will be dealt with in a future article. Most Australian jurisdictions require that the causative factor must “be a significant (or substantial) contributing factor”.  The present legal interpretation of “significant” is something that is “more than trivial”.

Now it turns out that your patient has been seeing an alternative practitioner, and that your patient is number six in a cluster of like cases, which is the basis of an as-yet unpublished paper, and has been spruiking the theory that the UV causes thrombi of the retinal veins, which embolise via a minutely patent foramen ovale.

Your patient is persuasive and demanding. You write a WorkCover certificate citing the cause as “possible thrombotic embolus from UV exposure from computer screen”. In so doing, you have taken on the role of being an expert witness, and have even offered an opinion on the putative UV causation of your patient’s condition.  You could have saved yourself some pain if you had written “My patient claims...I do not have an opinion regarding causation”.

Barristers love “expert witnesses” who concede that it is “possible”; it is a small step then to having you agree that “if it is possible, then it could be probable?”.  And so on!

And so it is that the court process can start with a scientific nonsense and declare it to be the truth.  Court decisions like this (it was a deliberately silly fictitious scenario) sometimes leave those of us with scientific minds utterly gobsmacked, but we have to take a step back and understand that causation is a word that means one thing to the scientific mind, and something else to the legal mind.

I recently had a case of a shop worker, who had been a checkout operator for many years, aged in her 50s, and with quite advanced osteo-arthritis of both knees.  She claimed it was the standing work that had caused her knees to degenerate, and this premise had been accepted by the general practitioner, who was able to complete the certificate.  But this worker had a BMI of 47. I was asked to comment on the causation.  And when we use the term “law” in this context we are referring to common law such a civil claims and motor vehicle accident claims.  With statute legislation, such as workers’ compensation, things can get a little more complex – more about that later.

At this point it is appropriate to consider the methodologies available for ranking risk factors in these circumstances. Near the top of the list is the Randomised Control Study. This methodology works well in a situation where evaluating an agent which can also be provided in placebo form. Consider a trial to measure whether Vitamin D supplements result in less days away from work. Give half the workforce Vitamin D, and the other half sham-Vitamin D. But how do you do such a trial to assess the effects of standing?  Sham standing?

Cohort studies can have a high level of reliability.  Enlist volunteers so that there are equal numbers with matching BMI profiles, into two cohorts, one working at sitting jobs, and the other working on standing jobs.  All have plain X-rays at the beginning, and again at the end of the study, ten years later. But, hey, the report that I had to do was supposed to be with the insurer within 10 working days!

Which leaves the epidemiologists to save the day with a case-control study.  They can study a population of sitting workers, then look at a matching number of standing workers, iron out the confounders (such as footballers and netballers), and come up with a pretty good idea if one lot have better knees than the other.

In the case of my shop-worker, all the best studies will not tell us what caused, or what didn’t cause this particular worker’s knee joints to degenerate.  The statistics only give us figures on relative risk.

Finally there are meta-analysis studies, which critically review the studies that have been done, and rank them according to their strength, and discard any studies based on inferior methodologies.   Good meta-analyses occupy the top level of epidemiological opinion. How do you pick a good meta-analysis?  Frankly that is venturing outside of my expertise, and I suspect most readers too.  This is where we have to trust that reputable journals have sorted the good from the ordinary.

The American Medical Association is a reputable publisher, best known for its AMA Guides to the Evaluation of Permanent Impairment, and more recently has ventured into the field of causation.

The AMA Guides to the Evaluation of Disease and Injury Causation, by Melhorn and Ackerman is now in its second edition. This book outlines the basic principles of causation, and then presents a series of meta-analyses where the authors have done all the hard yards and reviewed numerous scientific studies related to risk factors and causation, of some common conditions. 

In the case of osteo-arthritis of the tibio-femoral joint they found insufficient evidence supporting standing as a risk factor.  They had also looked for articles that studied non-occupational risk factors, and noted the following: Age: very strong evidence; risk increases with age. Overweight or obese: Accepted risk factor; very strong evidence; evidence for dose-response relationship amongst women.  Female: Accepted risk factor, further analysis required.  Family history, previous trauma and menisectomy were all accepted risk factors, although were not issues in the worker that I had assessed.  Note the recurring use of that word risk.

Given that standing did not even get a look in, and as there were other risk factors that strongly featured in this worker’s situation, I concluded that standing was not a cause or even a trivial contributing factor.

But it seems that the legal status of those AMA guides may be on thin ice, at least in Australia. I was recently asked to give an opinion on a causation matter but with the stipulation that asked me not to use the AMA Guides.  To achieve that, I had to go back to basics, which meant that I found myself referring to Sir Austin Bradford Hill’s classic essay on causation that dates from 1965. More about the Bradford Hill criteria in a future article

I suspect the matter is not finished there, as the lawyers have different ways of looking at causation.  As one solicitor put it, if there was an alternative way of looking at the subject, and one that made common sense to the average person, the law could accept a premise of causation on that basis.

Should we then worry ourselves, trying to lower our reasoning processes to that of the average person?  How convinced would that average person be when confronted with my patient with the retinal vein micro-emboli?  No.  Give them sound medical opinion based on well-researched facts, and leave it to the lawyers to do the squabbling.

The AMA Guides deals with to subject according to US law. For a local perspective there is Causation in Law and Medicine by Ian Freckelton and Danuta Mendelson, published by Ashgate 2002.  The lead authors are both from Melbourne.  There are chapters by other authors.

In a nutshell.  As doctors our everyday work is driven by evidence-based medicine; in other words we are making decisions on the basis of what has been scientifically proven.  By understanding the cause of a disease (e.g. H. pylori and duodenal ulcer) we can effectively manage the disease.  When lawyers start sniffing around, and asking questions about causation it is all to do with smell of money.

In the next article I shall deal with bias in medical reporting and the related sin of partisanship, as well as dealing with conflicts of interest when they arise.  This will be of particular relevance to AFOEM trainees who will be attending their Faculty’s Annual Training Meeting in Adelaide in May 2016.  ANZSOM’s SA Branch is awarding a prize for an essay competition on this subject.  The winning entrant will have the opportunity to present the paper at ANZSOM’s ASM to be held in Newcastle in August.