Issue 24 / 13 December 2010

I WAS recently contacted by a Sydney Morning Herald reporter who was trying to understand one miracle of Mary MacKillop that paved the way for her canonisation.

Someone in Victoria prayed to her and a cancer declared hopeless subsequently went into remission.

What could be the explanation?

It was not hard to talk about possible explanations.

First, the numerator problem; cancers do sometimes remit spontaneously, unexpectedly, the tail end of the normal distribution of survivors of each type of cancer.

And second, the denominator issue; how many people pray to a would-be saint, with most not delivered from a dreadful outcome?

The question itself is interesting because it betrays a human weakness.

We are prone to overemphasise associations and believe them to be causal.

A person with cancer prays to a dead nun and gets better.

Intuition suggests it was the act of piety, right?

Astonishingly the story caught on and propagated through media channels.

The telephone rang for interviews.

Worse, the two-dimensional simplification of poorly informed journalists reduced the story down to Catholic vs anti-Catholic.

Not just anti-Catholic, I was being pushed into the corner of criticising all churches, all religions.

Impertinent (and irrelevant) questions followed.

Was I a Christian?

Did I believe in God?

My atheism was the focus of comment on a Catholic website.

So, why the wrong question?

Unlike Richard Dawkins, I don’t see science and mathematics as a bulwark against religion, something to be stamped out.

Rather, let’s marvel at the way scientists (including, of course, clinicians) are able to cope with the dissonance of having religious beliefs (for example, that there is a magic and mystic component to life) completely at odds with their scientific ones (the machinery of DNA can explain how this works) without any grating of sensibility.

Within this context, it is interesting to speculate how religion actually evolved, as we must presume it did.

This is not to say that religion cannot be intrusive to the delivery, and education, especially of science.

The debates in America about whether Darwinian evolution by natural selection should be taught in schools alongside creationist theories seem to be anachronistic by about 150 years.

But, in general, religion and science get along just fine.

Rather, the question we should be brooding over is how we can be more rational when it comes to addressing the distressing business of over-associating phenomena, particularly when it comes to treatment effects that can be turned into “miracles”.

This is very important in medicine; together with many other things, patients (and, all too often, their doctors) fail to appreciate such as the placebo effect and regression to the mean.

As an illustration, I remember as a very young GP counselling a patient with high cholesterol after screening.

He ate a hearty high-fat diet.

Could he cut down on the steak and eggs and bacon? And come back in 6 months with a new lipid screen?

When he did return, he showed a huge reduction.

I beamed at him at the follow-up visit: “Well done. Whatever you’ve changed in your lifestyle is working well!”

“Well, Doc,” he replied, “I ain’t changed an effing thing” ― except he didn’t say “effing”.

It was a decade before I appreciated the regression-to-the mean effect when two measures are imperfectly correlated.

Our job is to make sure we don’t fail to manage the interpretation of data as badly as our more credulous patients.

This matters.

Patients with poor health literacy have poorer health outcomes, which may be causal.

Which may mean we need more than a passing acquaintance with statistics and epidemiology, something we too often bungle, to explain “miracles”.

If we don’t really understand these effects, what hope is there for our patients?

Professor Chris Del Mar is Professor of Primary Care Research, Bond University, Queensland.

Posted 13 December 2010

10 thoughts on “Chris Del Mar: ‘Tis the season to suspend disbelief

  1. Anonymous says:

    Why Is It so Difficult for Physicians to Discuss

    Megan Best and colleagues from the Psycho-Oncology
    Cooperative Research Group at the University of Sydney,
    Australia, conducted qualitative interviews with 23 physicians in
    palliative care or medical oncology from Australia and New
    Zealand. Participants were asked in open-ended questions why
    spiritual conversations with advanced cancer patients were so
    difficult and what the underlying challenges were (prompted by the
    fact that these discussions seldom took place). Results indicated
    the following themes in physician responses: (1) confusion over
    differences between spirituality and religion; (2) peer pressure from
    other physicians who disapproved addressing these issues with
    patients; (3) issues of personal faith that cause confusion about
    their role as a healthcare professional vs. addressing religious
    issues with patients; (4) institutional factors such as the presence
    of chaplains who can address these issues; and (5) historical
    factors in their medical training having to do with the separation of
    science and spirituality. Researchers concluded that “the current
    suspicion with which religion is regarded in medicine needs to be
    addressed if discussion of spirituality in the medical consultation is
    to become routine.”
    Citation: Best, M., Butow, P., & Olver, I. (2016). Why do we find it
    so hard to discuss spirituality? A qualitative exploration of
    attitudinal barriers. Journal of Clinical Medicine, 5(9), 77,
    Comment: This study provides important insights as to why
    palliative care physicians and cancer doctors are not assessing or
    addressing spiritual issues that come up in clinical care. Such
    studies are needed in the United States and other areas of the
    world, particularly in those regions where a high percentage of the
    population is religious and will invariably have spiritual needs when
    they are dying.

  2. Stephen Kleid says:

    I’m a skeptic. In over 20 years of treating cancers, I have yet to see a miracle, or met an oncologist who has had a case of surprise survival from metastatic cancer with no classical treatment (except prayer, meditation, vitamins, coffee enemas, etc). I’ve seen some suprising cases of prolonged survival with cancer, but they all succumbed. I have also seen some supposed radiological recurrences of cancer that turn out to be biopsy negative – hence the need for a tissue diagnosis, to confirm actual recurrence of cancer. I wonder whether some “miracles” are actually incorrect diagnoses. I once read an interview with Ian Gawler, who had had only seen a handful of surprise recoveries, including his case.

  3. Robert Loblay says:

    Here is a prime example from a Time magazine article (24 January) about the recent shooting spree in Tuczon. Under the heading of ‘Looking for risk factors’ (p23) the journalist says: “Many people are blaming mental illness for the massacre, but a more reliable set of predictors of violent crime are age (…), gender (…), lower socioeconomic status and history of arrest.”
    These so-called ‘risk factors’ and ‘predictors’ are nothing more than statistical correlates, but the journalist is implying that they are a more important CAUSE of the massacre than the obviously psychotic state of the shooter, conveniently shifting the reader’s attention away from the toxic mix of poor mental health care and widespread availability of guns in the USA.
    Of course it would be ridiculous to suggest that, in order to prevent future massacres, they should lock up all 16-20 year olds who are males, come from low socioeconomic groups, and have a history of prior arrest… wouldn’t it??
    Is this so different from the interventions that have been proposed by Marmot & other epidemiologists interested in the social ‘determinants’ of disease (which are really nothing more than statistical correlates)? Here’s yet another word calculated to mislead!
    Sorry Chris, I’m having trouble suspending my disbelief.

  4. Michael Busby says:

    Well done Chris. On the topic of doctors and religion … I’m constantly amazed at how smart people believe dumb things!

  5. Bruni Brewin says:

    Professor Chris Del Mar thank you for your post. If we know that ‘words’ are so powerful, why do we need to so often implant the negative to our patients? Someone once answered this question for me by saying; “But we have to be truthful!” Do we? Sometimes, silence is golden.
    I can remember one doctor (now retired) telling me that a patient had inoperable stomach cancer. His wife had asked the doctor to inform her of the results of the test out of concern for her husband. After tests showed confirmation of this, the patient’s wife said; “Doctor, please do not tell my husband. He won’t be able to cope.” And so the patient was told that he had a stomach disorder and given some medication to assist him. The cancer went away.
    If we know how powerful words are, and we know that the placebo (or should I say, belief system) is able to produce such astounding results, should we not then introduce them into our practices as a norm?
    Religion is based on ‘faith’. As healers should we not also sell faith to our clients to get over their illnesses?

  6. Robert Loblay says:

    Chris is asking me to tell him ‘what [words] should we use instead’? The first thing I would suggest is to abandon use of the term ‘risk factor’ except where a causal relationship has been established beyond reasonable doubt. There is a perfectly good term to use instead, which is ‘statistical association’.

  7. Chris Del Mar says:

    Thanks Robert, Peter and Dr Anonymous: this is useful feedback. Some responses:

    1 Robert feels that the fault of the ‘post hoc, ergo propter hoc’ fallacy lies in the way epidemiologists express it, as much as people read it, and implies a naughtiness on their (our) behalf because it assists our research. Mmmn. Let’s just assume that is right for the sake of argument. How should we say it instead? My concern here is that the fundamental problem is *educational*, not literal. People don’t understand enough about the philosophy of science. Fiddling with the words may not be enough to sort this out.
    But anyway, what should we use instead?
    2 Peter has correctly pointed to the issue of ‘luck’ (‘chance’) in our, and our patients’, fates. Anything *can* happen, of course. But what epidemiology can do is provide *some* measure of how likely that is, remembering always that rare things do happen (and I am glad you’re still with us, Peter, whatever odds you were given). Knowing the chances of any event is especially helpful when thinking about alternatives. I want the one that gives me (and my patients) the best chance of a good outcome. It is in the making of health decisions that epidemiology can be really useful to clinicians.
    Chris Del Mar

  8. Anonymous says:

    I agree with the Professor entirely and without reservations.
    I’d like to add that ‘the mind-set’ is also important along with ‘the placebo’ effect, as in ‘fear of consequences’ which most times don’t eventuate.
    Since this is my comment to this blog, I’d also like to wish all a very merry Christmas and to say to all who believe, Keep Christ in Christmas, won’t you.

  9. Peter Arnold says:

    “…we need more than a passing acquaintance with statistics and epidemiology…”
    We might indeed, Chris, but that’s no help to the individual patient.
    What did Victor Chang mean, 22 years ago, when he said I had a 70% chance of surviving 10 years post-CAGS?
    Only that 70% of his post-CAGS patients were alive 10 years later and 30% were dead.
    Which group was I in? He didn’t know – and nor did I.
    Short of our omniscience and prescience, it’s all a matter of luck, just like Mary MacKillop’s one-off ‘success’ and Ian Gawler’s survival. I even had a patient who survived 20 years after pneumonectomy for lung cancer invading the pericardium!
    “There are more things in heaven and earth, Horatio…” As we don’t know why these rare things happen, let’s simply call it ‘luck’ – being at the tail end of the normal distribution curve.
    Keep up the good work, Chris.
    ‘Skepdoc’ Arnold

  10. Robert Loblay says:

    A good start would be for epidemiologists and experts in the field of EBM to stop mis-using the term ‘risk factor’ in their publications, press releases and general discourse.
    I know they all realise (intellectually) that ‘risk factors’ are simply statistical correlations between a variable of interest and an outcome measure, but such correlations are always interpreted (emotionally) as if they had causal implications. The discussion usually morphs (often from one sentence to the next) from the identification of a ‘risk factor’ into speculation about potential interventions, e.g. what can be done to avoid or treat this ‘risk factor’ which is assumed to have caused the outcome.
    The apparently more neutral term ‘is associated with…’ is not much better because the listener usually hears ‘is caused by…’. And the problem is compounded when ‘X is associated with Y’ gets transmogrified into ‘X is a risk factor for Y’.
    We need to coin a better term which will not lead the emotions (if not the intellect) up the causal garden path. Perhaps something like ‘X is a candidate for further study…’ would get across the notion that statistical correlations are best regarded as hypothesis-generating observations rather than causal conclusions.
    If epidemiologists are unable to resist the emotional temptation to draw causal inferences from ‘risk factors’ based on their a priori prejudices, how can we expect doctors, journalists and the general public to do so?
    Is it too cynical to suggest that by continuing to use slippery ‘risk factor’ terminology, epidemiologists are serving their own narrow professional purposes when applying for research funding, submitting papers for publication, and spruiking their results in the media?

Leave a Reply

Your email address will not be published. Required fields are marked *