Robert Hughes.
Aheadline in late-February stated ‘‘Anti-depressants don’t work’’ and the story reported current controversies about antidepressants featured at a Mental Health in Crisis conference held in Christchurch a few days before. The article provoked a strong response, with many individuals saying, in effect, ‘‘I took antidepressants and was helped, therefore they work!’’
Such personal testimonies are commonly reported for new or different health treatments, so why can’t they be relied on as the basis for determining if a treatment works?
One problem is that every treatment ever devised in the history of medicine has likely had positive testimonials from those who claimed benefit from the treatment, including treatments such as blood-letting, which was popular for hundreds of years with doctors and their patients.
In 1849 Charles Darwin, one of the greatest scientists of all time, began the ‘‘water cure’’ (hydrotherapy) for his persistent gastric illness (now diagnosed as probably Crohn’s disease, exacerbated by lactose intolerance).
Over the next 15 years he repeatedly subjected himself for weeks at a time to daily treatments that included being roughly rubbed down with wet towels, being drenched with cold water, sitting for hours wrapped in wet towels, cold foot baths, and at some spas, homeopathic treatments and hypnotherapy.
Why did he endure this horrible treatment? What was the evidence that it worked? He did so because newspapers in Victorian times carried many personal testimonies about hydrotherapy cures and it was recommended to him by folk who personally testified to its enormous benefits.
Since Darwin’s time we have learned much more about how to evaluate health treatments, and we now discount personal testimony.
Personal testimonials are not reliable when deciding if a health treatment works, write doctors
and
There are two main reasons.
The first reason was discovered by Darwin’s cousin, Francis Galton. During his investigations of heredity, Galton discovered the phenomenon known as ‘‘regression to the mean’’. In the context of health treatments for chronic or relapsing conditions, this involves symptoms which cycle over time from non-existent to mild to severe.
This symptom cycling may happen within a single day (as is common with depression) or it may involve cycles over longer times, such as weeks or months (as also happens with depression). In such cases it is when symptoms are approaching their most severe and intolerable that we are most likely to seek help. The treatment that we then begin may be accompanied by a reduction in symptom severity.
But the ‘‘regression’’ phenomenon signals that this may well have happened anyway – after reaching a peak, the symptoms would have reduced to less severe levels no matter what we did. This produces the illusion of a treatment effect.
To guard against this, investigations of novel treatments need sufficiently long assessment periods to allow for the possibility that regression to the mean will reduce symptoms in the period