Intended for healthcare professionals


Treating insomnia

BMJ 2004; 329 doi: (Published 18 November 2004) Cite this as: BMJ 2004;329:1198
  1. Anne M Holbrook, director (holbrook{at}
  1. Division of Clinical Pharmacology, McMaster University, Hamilton, ON, Canada L8N 3Z5

    Use of drugs is rising despite evidence of harm and little meaningful benefit

    “By the way, doctor, I don't sleep at night. Can you give me something for that?” or “You can take away any of my other pills, but I have to have something for sleep!” are sentiments heard daily in any general medical environment. Patients with persistent complaints of insomnia—often elderly, frail, with multiple morbidities, multiple medications, already on or previously on a hypnotic medication—become problematic. Once past invoking sleep hygiene guidelines; looking for primary causes of insomnia; discussing medication risks such as falls, impaired cognition, driving crashes, and dependence; or discontinuing sedatives (and grumbling to the nurses for promoting them and to the house staff and referring doctors for prescribing them), what is the doctor to do? Why is this one of the least satisfying symptoms to treat and to educate medical professionals about?

    Perhaps because the definition of “normal sleep” remains elusive, as do the determinants of normal sleep, the correlation of psychopathology (which many doctors have neither time nor training to deal with appropriately) with poor sleep satisfaction is strong, the independent prognostic importance of insomnia itself is unknown, and the well known drug treatments for insomnia all have uncertain but worrisome ratios of harm to benefit.1 2 Once the time consuming sedative prevention or withdrawal exercise is completed, one has a sense of resigned acknowledgment that most patients persist with their complaint until some doctor finally prescribes and represcribes a sedative.

    Despite evidence of major harm and little evidence of clinically meaningful benefit, prescriptions for benzodiazepines continue to grow and are part of a “top 20” list of prescribed drugs in many jurisdictions.3 Furthermore, although their use is associated with poor functional status, cognitive impairment, daytime sleepiness, falls, and depressed mood, patients' satisfaction with effectiveness is high.46

    The conclusions of the recent guidance from the National Institute for Clinical Excellence (NICE) on newer hypnotic drugs will be a disappointment to those clinicians who were holding out hope that the newer “Z” drugs (zaleplon, zolpidem, and zopiclone) are superior to benzodiazepines in effectiveness or safety.7 The guidance, while limited to comparisons between the Z drugs and benzodiazepines, is the usual, high quality NICE product, with a comprehensive systematic review including industry submissions, extensive stakeholder consultation, and a highly readable summary of findings.

    Although initially promoted as superior to benzodiazepines in terms of daytime sedation, dependence, and withdrawal, the Z drugs have not delivered on several fronts. On the quality of evidence, of the 17 randomised trials with a total of 1284 patients, all were industry funded, outcomes were poorly and often selectively reported in favour of positive findings, comparators were suboptimal, durations were very short (maximum six weeks), and surrogate markers (generally sleep variables) were highlighted.7 On the risk-benefit front, no consistent difference was found between the Z drugs and benzodiazepines for either effectiveness or safety.7 On the economic front, since the Z drugs are each several times the cost of older benzodiazepines, without evidence of superior effectiveness, they cannot be considered cost effective.

    With the NICE guidance added to the weighty pile of negative evidence and expert opinion on hypnotic agents, do reasonable alternatives exist? Drugs are still important considerations in two main areas. Firstly, some commonly used drugs are well known to disrupt normal sleep—notably alcohol, selective serotonin reuptake inhibitors, drugs for Parkinson's disease, methyxanthines, and diuretics—and should be considered as potential contributors to the insomnia. Secondly, drugs can be effective for causes of secondary insomnia, especially pain and depression. Antidepressant medications are increasingly being prescribed for insomnia8; whether they are safe and effective outside of depression-associated insomnia is still controversial. The use of antidepressants, antipsychotics, or anticholinesterase inhibitors for insomnia related to delirium or dementia is also unproved.

    No drug has yet been shown to be more effective and safer than placebo in primary insomnia for the type of outcomes that matter, such as quality of life, daytime function, cognition, falls and fractures, or dependency. Placebo itself, as in every domain of therapeutics, has been found to be effective—in this case in improving sleep.9 The current aim of treatment is less focused on reduction of arousal from sleep and more focused on changing beliefs and attitudes about sleep.10 Thus NICE's recommendation that long term, non-pharmacological interventions for insomnia should be a primary target for evaluation of cost effectiveness. Although physical exercise shows merit and can be recommended by primary care providers,11 the current front runner for non-pharmacological treatment is cognitive behaviour therapy,12 w1 w2 w3 a technique not familiar to providers outside of psychiatry. Trials of different methods of training family doctors in cognitive behaviour therapy are currently under way.

    Cognitive behaviour therapy has many elements—stimulus control (your bed is only for sleep), sleep restriction (restrict your time in bed to your usual sleeping time), sleep hygiene, and relaxation therapy. Which of these elements are effective is unclear, but this is worth finding out as more passive interventions, such as audit and feedback to primary care doctors, are not effective.w4 In the meantime, the newer hypnotics remain with the older hypnotics as prime examples of iatrogenesis imperfecta.


    • Embedded Image Additional references w1-w4 are on

    • Competing interests None declared.


    1. 1.
    2. 2.
    3. 3.
    4. 4.
    5. 5.
    6. 6.
    7. 7.
    8. 8.
    9. 9.
    10. 10.
    11. 11.
    12. 12.
    View Abstract