How bad research fuels dodgy claims

f-files-small.gif
 

0

He was red in the face and the veins on his neck were standing out. He involuntarily spat as he shouted while furiously jabbing a scientific journal paper high into the air. “This paper proves that palliative sedation doesn’t hasten death!,” he bellowed from the audience.

The recipient of the Catholic priest’s[§] ire was Dr Rodney Syme, who had just delivered a paper at a Catholic bioethics conference in Melbourne, Australia.

The presentation didn't win over an already skeptical audience given that his address was to argue, by referring to the scientific literature, that palliative care can’t always help alleviate refractory end-of-life symptoms and that continuous deep sedation (CDS) may hasten a patient’s death.

By the end of the address the audience was bristling with hostility and our priest bolted to his feet to commence the attack.

I hadn’t seen the journal paper the priest was brandishing, so immediately after the conference I retrieved a copy and studied it.1 You can obtain your own copy here.

But first, a little background on the doctrine of double effect.

 

Doctrine of double effect

There has been ongoing debate—both scholarly and otherwise—about whether CDS (particularly the administration of higher doses of opioids) for dying patients may hasten their death. CDS may be administered where the dying patient is suffering from intolerable and otherwise unrelievable symptoms, for the purpose of relief from those symptoms through the cessation of consciousness, even if it might hasten the patient's death.

To some, this is an uncontroversial medical intervention supported by the doctrine of double effect. The doctrine is widely attributed to thirteenth century Catholic Saint Thomas Aquinas, though he probably developed it from ancient Roman principles.

The doctrine says that it is acceptable to take this kind of action if it is proportionate to the suffering of the patient, that there are no other

 

less problematic alternatives, and that the doctor's intention is to relieve the suffering and not to hasten the patient’s death.

Skeptics of the doctrine point out that it can’t be reliably determined what the doctor was actually intending when they administered CDS prior to death. They also identify that the notion that it’s OK for a doctor to kill their patient (hasten her death) as long as the doctor “doesn’t really mean to,” is ethically questionable.

So, on the one hand the doctrine of double effect is relied upon to defend doctor actions, and on the other, it is claimed that CDS doesn’t hasten death in any case.

That’s having your cake and eating it too. If CDS didn’t hasten death, doctors wouldn’t need the doctrine of double effect as a defense.

Poor methodology

Back to the brandished paper. Having read it, I don’t understand how this paper passed peer review, assuming it was indeed peer reviewed. It’s a fact of life that on occasion, even prestigious journals like Lancet and the Journal of the American Medical Association publish manuscripts that shouldn’t see the light of day. I argue that this article (from Annals of Oncology) is one of them.

The research suffers from multiple, serious methodological problems.

The research suffers from multiple, serious methodological problems. Sure, the researchers attempted to match test patients (who received CDS) with control patients (who didn’t receive CDS), by age, class, gender, reason for admission and for functional status. But there the good work ends.

Here are some of the serious methodological flaws in the study.

  1. Primary measure invalid. The authors report that they measured patient survival time from admission to hospice until death from any cause. This alone invalidates the entire study. If you say you are measuring the effects of CDS, then you have to measure from the commencement of CDS, not from some other event that is not the substance of your test.
  2. Insufficient inclusion control. There is so much variability amongst the test and control subjects that the possibility of sufficient sensitivity to detect differences is highly unlikely. For example, around half of patients (53%) were admitted for uncontrolled symptoms of many and varied kinds but were not imminently dying, while a large minority (41%) were imminently dying. There was a wide range of cancer types from slow to aggressive, and of metastatic sites. The sheer breadth of patient contexts gave rise to a survival standard deviation (68% of observed variance, 6 days) that was 50% larger than the mean survival period (4 days). Translating that into 95% confidence (two standard deviations), the variance was three times larger (12 days) than the mean (4 days). Put another way, at 95% confidence, patients survived from minus 8 days to plus 16 days (from admission, not administration of CDS). Hardly the stuff to drive assertive conclusions.
  3. Insufficient intervention assessment. The researchers considered only the administration of sedatives, not of any other medical or palliative interventions in their assessment of patient survival. What other interventions were administered, when and in what amount, and how successful were they? The authors also assumed that any necessary nutrition and hydration was given to patients (whether natural or by tube) equally across hospice units because “the best palliative care should be provided.” There was no mention of whether the various units had identical (or any) written palliative care practice standards or whether those standards were known and actually practiced by attending physicians and nurses in patient cases for this study.
  4. Insufficient intervention control. A wide range of sedative types (not just opioids) were administered in a wide range of doses for widely ranging periods of time; from sudden, light and intermittent, to heavy and continuous. Around half the patients receiving sedation received multiple different sedatives.

It is hardly surprising then, given this sorry state of affairs, that the study failed to find a significant difference in ‘survival’ between test and control groups: the design was certain to result in insufficient sensitivity. (Curiously, the CDS patients survived on average longer (12 days) than the control patients (9 days), though the difference was not statistically significant.)

Unscientific conclusion

In scientific studies—particularly those which have yet to be successfully replicated by other researchers—best practice dictates that the minimum ‘significance’ is cautiously attributed to the results. The minimum significance in this case would be that “our study did not find evidence to support the contention that administration of sedation hastens patient death”.

The conclusion asserted by the researchers
is not established by the study.

The next level of attribution is still cautious, but a little more assertive: “the results suggest that the administration of sedation doesn’t hasten patient death”. It is my belief, especially given the poor design and conduct of the study, that this level of attribution would be too high.

But the authors went all out. They categorically concluded that “PST [palliative sedation therapy] does not shorten life” and trumpeted it in the article’s title: “Palliative sedation therapy does not hasten death.” On the basis of this unjustifiable interpretation they further categorically concluded that the doctrine of double effect is therefore unnecessary.

The authors in my view committed a serious scientific blunder in translating an exploratory and poorly-designed study into unqualified and assertive, unsupported conclusions. How this passed peer and editorial review is a mystery.

The upshot

The upshot of this is that our priest, perhaps with an understandable deficit of expertise in scientific endeavour, grabbed a paper that supported his hypothesis and fervently believed it to provide unarguable evidence, given its unequivocal title and conclusion.

In reality, each and every paper published in scientific journals has to be read and judged on its own merits. It is not appropriate to simply accept a study’s conclusions merely by virtue of its publication in a journal, even a prestigious one.

In reality, each and every paper published in scientific journals has to be read and
judged on its own merits.

Critical appraisal (including cross-checking other sources) is the scientific standard of practice I employ as I follow trains of research and commentary on various facets of end-of-life decisions and assisted dying, across my literature collection now well in excess of five thousand journal articles, theses, books, professional policy papers and other sources.

It is a standard I commend to everyone in order to hone in on the truth and to avoid dissemination of flapdoodle, fudge and fiction.

References

1    Maltoni, M, Pittureri, C, Scarpi, E, Piccinini, L, Martini, F, Turci, P, Montanari, L, Nanni, O & Amadori, D 2009, 'Palliative sedation therapy does not hasten death: results from a prospective multicenter study', Ann Oncol, 20(7), pp. 1163-9.


[§] The meeting was conducted under Chatham House rules, so what was said can be reported, but who said it cannot.


Share This Post: