Australian psychologist, Dr. James Alexander is author of the 2012 book, The Hidden Psychology of Pain, and the 2017 book, ‘Getting the Z’s You Want’. He became a psychologist after being nearly killed in a car accident as an 18 year old when his car was hit by a drunk driver. In addition to being very physically damaged, he was emotionally traumatized by the experience- resulting in many years of problems with sleep. When he began to physically recover form the accident, he found a collection of pop-psychology self help books in his fathers book shelf, and his psychological recovery began. Within the year, he decided he wanted to become a psychologist so as to help other people that had been similarly traumatized. This was 35 years ago, and for the last 30 years, he has been providing psychological services to a broad range of people in hospital settings, pain management clinics, rehabilitation services, and for the last 15 years, in private practice in NSW, Australia. he has a PhD in clinical health psychology, and whilst open to a range of approaches, he thinks of himself primarily as an EMDR practitioner. His new book is based on his own efforts to create better sleep.
Get A 10% discount on all my courses below through the Zur Institute using the discount code: DRDAVE88
Get 10% discount on all lectures at The JungPlatform using our discount code: DRDAVE
A psychology podcast by David Van Nuys, Ph.D.
Copyright 2018: David Van Nuys, Ph.D.
Podcast: Play in new window | Download
Subscribe: Apple Podcasts | RSS
At about 47:34, Dr. Alexander makes reference to “particularly disturbing” research that shows that those using sleeping pills (several types) “had up to 5 time greater chance of dying due to all causes … and there was a dose-dependent relationship …”
The research that I believe Dr. Alexander is referring to is “Hypnotics Association with mortality or cancer: a matched cohort study” by Dr. Daniel Kripke et al, which can be found here:
http://bmjopen.bmj.com/content/2/1/e000850
The study wasn’t quoted explicitly in the podcast, but this one does match numbers Dr. Alexander referred to.
I have researched this subject and subsequent follow-on studies, and am quite convinced that this study cannot make the causal conclusion that is so often implied when it is quoted. It is in my opinion a case of the classic error of “association is not causation”.
The causes of death in this study were not reported, nor were they in a very much larger follow-up study with British data (with similar but less significant association). But they were finally for about half of the subjects of a 3rd even larger study. Because it is the largest, has causes reported (from British data, not from French data it also uses), and also explores a better mathematical model that avoids certain assumptions of the first 2 studies, this is the study that should be quoted:
https://www.researchgate.net/publication/280865575
But even this study’s abstract (which concludes an association) doesn’t do justice to the details. This is a short-term study (only 12 months), and what was the major cause of death (Table 3 in the above) by more than a factor of 2 above all other causes? Cancer.
So the presumption had to be, that the sleeping pills both caused cancer and killed patients, in a relatively short period of time (this study was not as long as the first study).
How about this alternative explanation: cancer patients have trouble sleeping, especially in late stages, and they tend to be prescribed sleeping pills. Then some of them die. The sleeping pill is indeed associated, but was not a cause.
The rule in statistics, is that you cannot call a link in an observational study causal without a sound mechanism for the cause, and even then, that would be on very thin ice – you have to follow up to try to prove it another way. Without “cause of death” data in the first study, a sensible causal link cannot be made.
Studies that make flashy appearances in the media (as this first one did) are more often wrong than right according to one study (which yes, may itself also be wrong). Followup studies that say differently don’t get the same attention, so an incorrect conclusion, implication, or interpretation is often not corrected.
thanks Bob- fair comment. My recollection of the research is that they statistically controlled for health status. Is this not correct? You appear to have looked more into the study than i have. Agreed- correlation does not mean causation. However, i suspect it does not exclude causation either (but this does not argue against the need for more exact language).
The authors state, “Results were robust within groups suffering each comorbidity, indicating that the death and cancer hazards associated with hypnotic drugs were not attributable to pre-existing disease.”