Sleeping Pills are Very Dangerous

Do you know how dangerous prescription sleeping pills are? I didn’t, and I do sleep research.

I came across Dr. Daniel Kripke’s book yesterday’s post on undisclosed risks of medical treatments. I had written an almost-complete draft a year ago. One line in the draft said “undisclosed risks of sleeping pills” with no additional information. I couldn’t remember why I’d written that so I googled “dangers of sleeping pills” and found Dr. Kripke’s book. I was unaware the evidence was so strong. I asked Dr. Kripke to tell the story of how he came to write it. He replied:

It is almost a life-long story.

As a young psychiatrist, I learned that the American Cancer Society had done a questionnaire survey of a million people which showed mortality related to long and short sleep. [People who sleep less or more than average have higher death rates.] In 1975, I asked if they would collaborate with me on a more complete analysis of the data on sleep length and insomnia.  As a control variable, we included analysis of their one question about sleeping pill use.  To my surprise, it looked like sleeping pill use was a strong predictor of early death, while insomnia was not (if you controlled for sleeping pill use by insomniacs).

There were many reasons why these results needed further study, so I asked if I could refine the questions for the new Cancer Prevention Study II (CPSII) which the American Cancer Society commenced in 1982 with 1.1 million participants. Imagine my surprise when I observed that sleeping pill use was associated with a comparable mortality hazard ratio as cigarette smoking! These studies, and about 20 more done all over the world with similar results, had two important limitations: in general, the studies did not identify the sleeping pills used and did not measure whether those taking sleeping pills at the start of the study continued the drugs, or whether those who were not taking sleeping pills (the comparison group) started taking them.  So another study was needed.

Meanwhile, sleeping pills were never my main scientific concern.  I was mainly interested in bright light treatment of depression and trying to understand how light worked. When I saw that patients needed information about light treatment, I wrote a very short book called “Brighten Your Life”, but it wasn’t long enough to publish, so I added information about sleeping pills to make it longer. When we found no publisher for the book, I made the information available at two web sites: and  I found that the web site about sleeping pills was more popular than the advice about light treatment–indeed, one of the most popular sources about sleeping pills at Google.  Therefore, over the years, I have worked to revise and update both web sites to try to help patients.  It costs some money to program and maintain the web sites, but people write me to tell me how they have benefited.  I see so much misinformation coming from the drug companies that I want people to have an alternative source.

Five or six years ago, my friend Dr. Bob Langer was working at the Geisinger Health Research Center, which had access to electronic health records about sleeping pill use from a large number of people. It took us five years to plan a study, obtain approval from ethics committees, retrieve the complex data from computer files in anonymized form, and analyze the very complex results.  When these were published by the medical journal BMJ Open, the new information became available at  It is an interesting web site which includes more data in a supplement to the main article and some comments and debate about the article.  The interest in the article was world-wide, with stories on the BBC, at Agence France Press, in major newspapers in Japan, India, and China, and even mentions in far-off places like Myanmar and Ruanda.  The new data showed that people taking drugs such as zolpidem and temazepam had about 4.6 times the mortality rate of people of the same age and sex who took no sleeping pills. The new data confirm that sleeping pills might cause as much death as cigarettes, and also some cancer, so I feel a big responsibility to make the information available.  There may be hundreds of thousands of lives at stake.  People need to know that sleeping pills are too risky to use, and I wish I had more help in telling people.

Recently we updated the Dark Side Of Sleeping Pills and Brighten Your Life and made them available together in a Kindle book, which is easy to purchase at Amazon and read off-line.  The books have some new information which we have not yet had a chance to put in the web sites.

Even with, now, more than 20 scientific papers showing that taking sleeping pills is associated with more death and more cancer, many people don’t believe it. They imagine there is some other explanation, though nobody has been able to demonstrate an alternative explanation.  Of course, statistical association is not quite the same thing as proof of causality, but if it is good enough for the American Cancer Society to advise avoiding cigarettes, it is enough evidence of risk to stay away from sleeping pills, in my opinion. The problem is that the drug companies have never done a controlled trial study large enough to prove one way or another whether the sleeping pills cause death and cancer, and I think they never will.  The cigarette companies have never tried to prove that cigarettes are safe, and they know better than to try.  It is the same.  Whereas the FDA requires the very large studies for heart and diabetes drugs and so forth, the FDA has dropped the ball with sleeping pills.  For more information about that, please see the Kindle book.  There is, however, a new alternative to large, expensive, and dangerous controlled trials called a Mendelian randomization study, which uses the new genetic methods to determine causality when a genetic variation causes a risk factor such as sleeping pill usage.  Since the genetic data already exist to do the Mendelian randomization studies, it is a matter of doing the difficult statistical analyses.  I hope scientific colleagues will join in this task, because I can’t do it by myself.  It is crucial to determine for sure the risks of sleeping pills.  Too many lives are at stake.

Sleeping pills are astonishingly dangerous for something that is treated as more or less safe. In some cases, they are associated with a five-fold increase in death rate after only a few years of use. Cigarette smoking is associated with only a two- or three-fold increase in death rate after long use. And doctors don’t prescribe cigarettes. Is there anything else treated as safe that is associated with such a large increase in death rate? I can’t think of anything.

9 Replies to “Sleeping Pills are Very Dangerous”

  1. Thanks for this. I am incapable of sleeping more than six hours per night. It’s just the way I am. I used to worry about it, but then I decided to embrace it and use those extra hours in the morning for self improvement. I take on-line courses, sometimes watching the lectures on a treadmill at the gym. My only problem is that my physician thinks I (and everyone else) need 8 hours per night. I am continually being offered sleeping pills. I prefer not to take them, and it’s good to know my instincts are basically correct.

  2. I’m wondering about the big risks of hypnotic drugs study– ” Data were adjusted for age, gender, smoking, body mass index, ethnicity, marital status, alcohol use and prior cancer”. The data wasn’t matched for sleep problems, so it’s possible that the drugs aren’t more dangerous than insomnia.

    Seth: I agree. What’s also possible, and plausible, is that sleeping pills increase mortality by a large amount. I don’t know of any study showing that insomnia is associated with anything close to the same increase in mortality.

  3. How much years of life are we talking about for a five-fold increase in death rate?

    Seth: Let’s say you’re 40 years old and your expected total lifespan is 80 years. Then you have a roughly 1/40 chance of dying each year. If you have a 5/40 chance of dying each year, your expected total lifespan goes down to 48 years — a loss of 32 years of life.

  4. Paul Jaminet’s new edition of The Perfect Health Diet discusses the health effects of disrupted circadian rhythms. I wonder if, particularly in the case of cancer deaths, what’s showing up in the data as an association between sleeping pills and death is that people with disturbed circadian rhythms are more likely to use sleeping pills.

    Seth: Yes, that is possible. In this study, however, “sleeping pill use was associated with significantly increased mortality after control for reported sleep durations and insomnia.”

  5. Let’s say you’re 40 years old and your expected total lifespan is 80 years. Then you have a roughly 1/40 chance of dying each year. If you have a 5/40 chance of dying each year, your expected total lifespan goes down to 48 years — a loss of 32 years of life.

    But this 40-year-old doesn’t face a 1/40 chance of dying in Year 1. His risk of death starts out far below 1/40 in his early 40s and end up far above 1/40 in his late 70s. I doubt that quintupling the death rate would really push his life expectancy below 50, although I don’t have the patience (or, really, the skill) to work it out mathematically.

  6. uh seth, your arithmetic is completely wrong. if you have a life expectancy of 40 years from now, you don’t have a constant 2.5% chance of dying each year. you have a small chance of dying in your 40s, a somewhat larger chance of dying in your 50s, and so on. if i’m 40 and healthy my chance of dying in my 40s is pretty low – i don’t know exactly but maybe 3%? say that goes to 15%. that means an 85% chance of making it to 50 – already a 48.5 life expectancy, and that’s if everyone kicked at 50, so it’s quite quite a bit larger.

    Seth: Right. It was an approximation to give a rough idea of what the answer would be.

  7. Is melatonin classified as a sleeping pill? It can be used as a supplement and causes drowsiness.

    Seth: No. This study was about prescription sleeping pills.

Comments are closed.