CBT-I vs sleeping pills: what the research says

In short-term trials, CBT-I and prescription sleep medication improve sleep by similar amounts. Where studies followed people afterwards, the gains from CBT-I tended to last once treatment ended, while gains from medication tended not to. The number of head-to-head trials is small, so the evidence is moderate at best.

This article compares research outcomes only. It is not a recommendation to start, stop or change any medication. Please speak with your doctor or pharmacist before changing any medicine.

What the research says

Similar at first, different later. A systematic review of 5 randomised trials found, on low to moderate quality evidence, that CBT-I beat benzodiazepine and non-benzodiazepine sleep medicines in the long term, while very low quality evidence suggested benzodiazepines work better in the short term. ^[1] A meta-analysis of 21 studies and 470 people found short-term improvements moderate to large for both and similar in size, with behaviour therapy reducing time to fall asleep more than medication did. ^[2] In 78 older adults over 8 weeks, behavioural therapy, medication and the two together all beat placebo, and gains held at follow-up for people given behavioural therapy but not for those given medication alone. ^[3] In a placebo-controlled trial of 46 older adults, CBT beat zopiclone on 3 of 4 sleep measures in both the short and long term, and on most measures zopiclone did not differ from placebo. ^[9] In plain terms, both help about equally during treatment, and CBT-I's benefit tends to stay after it stops.

The studies behind this

• Mitchell systematic review (2012): 5 randomised trials comparing CBT-I with sleep medication. Low to moderate quality evidence found CBT-I more effective than benzodiazepine and non-benzodiazepine sleep medicines in the long term. Very low quality evidence suggested benzodiazepines work better in the short term. ^[1]
• Smith meta-analysis (2002): 21 studies, 470 people. Short-term improvements were moderate to large for both approaches and similar in size. Behaviour therapy reduced time to fall asleep more than medication did. ^[2]
• Morin trial (1999): 78 older adults, 8 weeks. Behavioural therapy, medication and the two together all beat placebo. Gains held at follow-up in people given behavioural therapy but not in those given medication alone. ^[3]
• Sivertsen trial (2006): 46 older adults, placebo-controlled. CBT gave better short- and long-term results than zopiclone on 3 of 4 sleep measures, and for most measures zopiclone did not differ from placebo. ^[9]

Why might the long-term picture differ?

This is our reading of the findings, not a proven mechanism. Medication acts while it is being taken. CBT-I is a set of skills and habit changes aimed at the factors that maintain insomnia, so it can keep working after the sessions end.

What do the guidelines say?

• American College of Physicians (2016): CBT-I as the initial treatment for all adults with chronic insomnia. For people whose insomnia persists after CBT-I, the guideline advises a shared decision with a clinician about adding short-term medication, weighing benefits, harms and costs. That second recommendation is graded weak, on low-quality evidence. ^[4]
• European Insomnia Guideline (2023): CBT-I first-line, and a medication can be offered when CBT-I is not sufficiently effective. The guideline discusses duration of use for each type. ^[5]
• American Academy of Sleep Medicine (2017): weak recommendations for several individual medications compared with no treatment. The authors note that evidence for each drug is limited by few trials, differences between trials and funding sources. The AASM guideline gave weak recommendations for zolpidem, eszopiclone, zaleplon, suvorexant, temazepam, triazolam, ramelteon and doxepin, and weak recommendations against trazodone, diphenhydramine (an ingredient in many over-the-counter sleep aids), melatonin, valerian, tryptophan and tiagabine. In the guideline's own words, a recommendation against, particularly on low-quality evidence, "is not equivalent to a demonstration of ineffectiveness." ^[6] Zopiclone was not among the medicines the AASM guideline reviewed. What the evidence says about melatonin has its own article.

Why is medication still so common?

In a Canadian telephone survey of 2,000 adults, 13.4% met criteria for insomnia, and prescribed medication was the most widely used treatment: about 10% had used a prescribed sleep medicine in the previous year. Only about 13% had ever consulted a health professional about sleep. ^[7] Access to trained CBT-I providers is limited in many places, which is one reason digital versions have been studied.

What the studies cannot tell you

• Very few trials compare the two head to head, and several are small.
• Trials used different medications, doses and durations, so no single result fits every drug.
• Most studies recruited people with chronic insomnia and without major other conditions. Results may not carry over to everyone.
• CBT-I has its own demands. Some parts, such as limiting time in bed, can make people sleepier for a few weeks. The American Academy of Sleep Medicine notes sleep restriction may be contraindicated, meaning unsuitable, for people in high-risk occupations such as driving or operating heavy machinery, people predisposed to mania or hypomania, and people with poorly controlled seizure disorders. ^[8] Talk with a doctor first if any of these apply, or if you have another medical or psychiatric condition.

So which is better?

Neither always. Guidelines put CBT-I first, and medication is a reasonable conversation to have with your own doctor, especially when CBT-I alone has not been enough or is unavailable. If you already take a sleep medicine and are wondering about stopping, coming off sleeping pills covers what the research says, and that conversation belongs with your doctor or pharmacist. Do not stop or reduce a medicine on your own because of anything you read here.

If you want to try CBT-I, Small Hours builds a plan from your own sleep diary. You can sign up at trysmallhours.com.

General education, not medical advice.