Wallace B. Mendelson MD


All prescription sleeping pills, like other medicines, have to be shown to be more effective than placebo (‘sugar pills’) in order to be approved by the US Food and Drug Administration. Large analyses combining many studies, and using elaborate rules of evidence have found that benzodiazepines and zolpidem help sleep in chronic insomnia, at least in short term use (usually one week). One limitation of such an approach, though, is that it gives information about a group of medicines in large groups of people, but doesn’t provide as much insight as to how a particular drug (and dose) will do in a given individual. A clearer answer to whether sleeping pills help also involves looking at a number of smaller questions: How is improvement measured? How large was the improvement? How long does it last?


How is improvement measured? Usually sleeping pills are tested in two different ways: with physiologic sleep studies, and by asking patients how they feel. Each has certain advantages and disadvantages. The sleep recording (polysomnography) brings a measure of objectivity and a look at physiology when assessing efficacy. Patient reports are valuable in that in a sense that’s what we’re especially interested in—how people feel about their sleep, and also can give judgments about overall sleep quality. So a combination of both types of information are important. As we will see in the next section, in general the effects of sleeping pills on polysomnographic measures are more modest that those on patient reports.


How large is the improvement? One large analysis which combined the results of many individual studies of benzodiazepines and related sleeping pills found that in terms of the sleep EEG, on average they increased sleep by about one hour, while decreasing the time to sleep onset by only about four minutes. As always, how patients felt about their night’s sleep differs from what brain waves show: in this case, they felt that they had fallen asleep about 14 minutes more quickly than when taking placebo. Another such analysis of studies of the elderly, which included benzodiazepines as well as zolpidem and zaleplonshowed more modest gains, with an increase in total sleep of about 25 minutes. It found no differences in sleep quality between the benzodiazepines and the Z drugs.


To summarize: All prescription sleeping pills have been shown to be effective in both sleep recordings and reports of how patients feel about their night’s sleep. The results of patient’s reports are usually more robust than the sleep recordings. In general, though, it is important to remember that these are not cure-alls. They can, and usually do, help sleep to some degree, but it’s best to have modest expectations. Often, it’s wise to use them as part of a broader way of improving sleep, which can include improving sleep habits, non-pharmacologic ‘talking’ therapy, and looking into other sources of difficulties such as other medicines which may be disturbing sleep.


How long do they last? Most studies of nightly use of sleeping pills have been limited in duration to one or two months, and often less. There are a few exceptions including zolpidem, eszopiclone and zaleplon tested for one year, in ‘open label’ studies in which the investigators were aware of what medicine was being taken and hence were potentially less precise. There have also been studies using the more stringent ‘double-blind’ design, which showed continued effectiveness of suvorexant as measured by patient reports for one year, and of ramelteon as measured in sleep studies for six months. In general, though, we are in a situation in which most drugs are assessed and shown effective in the short term, which is not reflective of the long term usage in the real world. The answer, then, is that ultimately we are really not sure, though the evidence we do have available suggests that they do not lose effectiveness. This is more secure in talking about the Z drugs than the benzodiazepines, only because the studies of the latter have never really been done, and now that Z drugs and newer agents are here, it seems unlikely that they will be. There is also another reason to be less secure in using benzodiazepines in the long term, as studies have shown impairment in a variety of thinking processes in persons taking them for long periods of time. We also do not know how long effectiveness lasts when taking sleeping pills in non-nightly use, though the limited information we have is that the Z drugs do not show evidence of tolerance. At this time, the only drugs which do not have formal FDA indications limited to short-term use are eszopiclone, ramelteon, zolpidem ER, and suvorexant. The broader problem of sleeping pills being tested in the short-term, but often taken in the long-term, is also true of other types of medicines—it comes up, for instance, with antidepressants. Many researchers, including your author, believe that we need to determine who should do these kinds of studies—for instance, the drug companies (which at least so far have not done very many), a government agency, or perhaps the drug companies by government requirement.




Wallace B. Mendelson MD is a psychiatrist, sleep specialist and author.  His recent books include Understanding Sleeping Pills (available on Amazon), from which this article is excerpted.