7

On Seth Robert's blog, Seth published an email discussing sleeping pills from Daniel Kripke saying:

Imagine my surprise when I observed that sleeping pill use was associated with a comparable mortality hazard ratio as cigarette smoking!

Are sleeping pills that dangerous?

Christian
  • 33,271
  • 15
  • 112
  • 266
  • 4
    Or does sleeping pill use indicate underlying medical issues that are causing the mortality? – Loren Pechtel Nov 09 '15 at 22:57
  • Anecdotal evidence: I have heard that sleeping pills are claimed to be an easy and painless way for Japanese to commit suicide, although this idea is mostly used in fictions i.e. Japanese TV drama series. On the contrary, Japanese in real life resort to much efficient way than sleeping pills (which I shall not mention here). So, sleeping pills may not be that dangerous after all. –  Nov 10 '15 at 18:23

1 Answers1

2

There are no large randomized controlled trials of comparing mortality between cigarettes usage and sleeping pills usage to conclude that sleeping pill use was associated with a comparable mortality hazard ratio as cigarette smoking which can be inferred when one goes through Daniel F. Kripke's review here. Also the available evidence of an association between hypnotics and mortality is not sufficiently strong and more independent studies are needed assessing the possible risks of hypnotics in order to give health care recommendations referring to Victor Vallejo-Garcia et.al. who declare no conflict of interest.

For absolute proof, we would need large randomized controlled trials of cigarettes or sleeping pills, but nobody is going to do such trials. How about the sleeping pill companies? Of course, now that we know that particular sleeping pills are associated with excess mortality, it would probably be unethical to do such a controlled trial, so for those particular sleeping pills, we will probably never have absolute proof whether they cause mortality or not.

The study on which Kripke had based his findings for sleeping pill use having a comparable mortality hazard ratio as cigarette smoking also had several other drawbacks listed below.

  1. Concurrent psychiatric diagnoses were not addressed and so mortality may have been related more to suicide or substance abuse such as alcohol use.

  2. Researchers were not able to conclude if the sleeping pills specifically contributed to the increase in mortality since in addition to risk of early death, participants taking sleeping pills had higher rates of asthma, cardiovascular disease, obesity and high blood pressure.

  3. Associations seen in the study by Kripke et al. were heavily influenced by erroneous selection of the unexposed control cohorts referring to Anton Pottegård et.al. in 2014. Morten Andersen and Jesper Hallas, co-researchers in this mentioned point have participated in research projects funded by Nycomed, the manufacturer of nitrazepam, and Pfizer, the manufacturer of Halcion (triazolam) and Tafil (alprazolam), with grants paid to institutions where they have been employed. Dr. Kripke published the first controlled clinical trial of bright light treatment in 1981 and for the past 30 years has been doing research on light treatment and treating patients with light.

  4. Limitations such as selection bias, overdiagnosis of cancer exist in the study and this makes it difficult to determine the specific contribution of hypnotics to mortality.

pericles316
  • 22,676
  • 2
  • 84
  • 161
  • In our medical system absolute proof is not required for health care recommendations about side effects. Anti-depressants have warnings that they mght increase suicide chances that are not based on controlled studies. We don't believe in the smoking-cancer link either because of controlled studies and even when the tabacco industry campainged for a long time that nobody should stop smoking because of lack of controlled studies, I don't history takes their side. – Christian Nov 11 '15 at 11:54
  • @Christian-Point taken for side effects but what about mortality, doesn't it need absolute proof in the form of large randomized controlled trials or meta analysis for confirming rather than dealing with selection bias? – pericles316 Nov 11 '15 at 12:06
  • That was the position of the cigarette lobby for decades. I'm not aware that large randomized controlled trials where done at any time for cigarettes. At the same time we generally believe that society should have sooner accepted that cigarettes cause cancer and raises mortality. – Christian Nov 11 '15 at 12:21
  • Several researchers have objected or have raised concerns regarding the findings of Kripke et.al. The main concern is that this finding should not generalized for all sleeping pill users but is addressed towards a select population using Benzodiazepines such as older people with insomnia ! – pericles316 Nov 11 '15 at 12:28
  • 3
    I have read his argument, and I simply don't understand it. Can anyone summarise it more simply? Surely there doesn't need to be a trial directly comparing smokers to sleeping pill consumers? Surely a large scale of the individual treatments is feasible - in fact, doesn't he claim to have performed some with 30,000 subjects? What am I missing? – Oddthinking Nov 11 '15 at 12:53
  • The study compared 10,000 patients who took sleeping pills and over 20,000 matched patients who did not take sleeping pills. The patients who took normal doses died 4.6 times as often and patients who took higher doses died 5.3 times as often. Those patients who took fewer than 18 pills per year had a very significant elevated mortality of 3.6 times over that of patients who took no hypnotics. However, there were drawbacks to the study such as absence of criteria for psychiatric diagnosis for risks of suicides due to substance absence and the patients having prior obesity, asthma and BP. – pericles316 Nov 11 '15 at 13:10
  • As far as the Anton Pottegård et al (2014) paper goes, it's worth noting that two of the people have a conflict of interest and received money from sleeping pill producers. That means that the fact that the paper exists to object to the Kripke is expected whether or not the objection itself has merits. Tabacco companies also had their scientists put forward criticism of studies that found the smoking cancer link. – Christian Nov 11 '15 at 15:37
  • 2
    OK, so I understand there may be a problem going from correlation to causation here, but apparently smoking has a similar around 3 times mortality. So it seems the answer is, YES, sleeping pill usage appears remarkably similarly dangerous as smoking (to the best of our current measurements). – Jonathon Nov 11 '15 at 20:18
  • @Jonathon Wisnoski-The study on which the answer is to be based as yes has several drawbacks such as absence of criteria for psychiatric diagnosis for calculating risks of suicides due to substance abuse which might have an impact on overall mortality and the inclusion of patients having diseases such as obesity, asthma, cardiovascular disease and high blood pressure. We might never know that whether the effect on causing 3 times mortality similar to smoking was only due to the sleeping pill usage or due to several other prexisting factors ! – pericles316 Nov 12 '15 at 12:23
  • @Christian-Researchers such as Anton Pottegård and Søren Friis who have no conflicts of interest as the other researchers mentioned have said that "Kripke et al may inadvertently have introduced bias by some of the choices they have made in their study design and analysis." However, they too agree that "Prescription of hypnotics should generally be avoided, and these drugs should almost exclusively be reserved for short- term use in selected patients." Also Kenneth G Terkelsen raises a valid question on whether psychiatric illness caused the mortality rather than the hypnotics ! – pericles316 Nov 12 '15 at 12:33
  • 1
    Referring to Daniel F. Kripke himself here... that kind of introduction is a good way to deflect the attention of the reader because it is simply a poor introductory statement for a text. – bandybabboon Jan 12 '16 at 07:08
  • @ufomorace-Changes done! – pericles316 Jan 12 '16 at 07:11
  • Example of conflict of interest being called out in the BMJ: http://bmjopen.bmj.com/content/2/1/e000850.short/reply#bmjopen_el_336 – Sklivvz Jan 12 '16 at 10:26