Treating Chronic Insomnia

Sleeping pills stop working when you stop taking them; a short course of cognitive behavioural therapy for insomnia (CBT-I) mostly keeps working after it ends. That is why the guidelines now put CBT-I ahead of any drug.

If insomnia has lasted three months or more, start with CBT-I: at least four sessions with a trained therapist if you can get one, a well-tested digital program if you can't. Keep sleeping pills for when CBT-I has genuinely failed, for a short crisis, or as a bridge while therapy starts — and skip over-the-counter antihistamine sleep aids.

Why therapy comes before pills

The sleep-medicine guidelines recommend CBT-I first for chronic insomnia, ahead of any medication.[1] [2]

Chronic insomnia — trouble sleeping at least three nights a week for three months or more — affects about 12% of adults.[3]

Pooling 87 randomised trials, CBT-I cut insomnia severity by what statisticians call a large effect and clearly improved most sleep measures — but barely moved total sleep time: it consolidates sleep rather than lengthening it. Face-to-face courses of at least four sessions beat self-help and shorter courses.[4]

What CBT-I actually involves

CBT-I is a package of behaviour changes and thinking techniques.

  • Stimulus control. Bed is for sleep and sex only; if you are not asleep within about 20 minutes, get up.
  • Sleep restriction. Temporarily cut time in bed so sleep consolidates. It has the strongest standalone evidence: given alone by primary-care nurses over four sessions, it left people with insomnia severity scores of 10.9 at six months, against 13.9 for those given a sleep-hygiene booklet.[5] Daytime naps are usually dropped meanwhile — see Daytime naps.
  • Cognitive restructuring. Challenging thoughts like "without eight hours, tomorrow is ruined."
  • Sleep hygiene. The familiar basics. On its own it came out inert.
  • Relaxation training. For people whose insomnia runs on anxiety about sleep. Its estimate sat slightly below no component at all — not proof of harm, but a reminder that more techniques are not automatically better.[6]

Which part does the most work is debated: across 241 trials, cognitive restructuring came out on top,[7] while a separate analysis of 80 studies singled out sleep restriction.[8]

Is an app as good as a therapist?

Digital CBT-I clearly works and is a reasonable first move without a therapist, but is not proven equal to one.

Digital CBT-I programs — Sleepio, Somryst and SHUTi have the most data — kept their gains for up to 48 weeks across 11 randomised trials.[9] The largest single trial was funded by the maker of the program it tested.[10]

The one head-to-head trial, in 101 sleep-clinic patients, found digital significantly worse than face-to-face at both time points: formally inferior at nine weeks, and at thirty-three weeks inconclusive by the authors' own pre-set margin.[11] Across 52 randomised trials, no format was statistically better than another, though therapist-led formats ranked highest and unguided self-help lowest.[12] [13]

When a sleeping pill makes sense

Every drug for chronic insomnia gets only a weak, conditional guideline recommendation, because benefits are modest and harms real.[14] A pill is reasonable when CBT-I has properly failed, when chronic pain drives the insomnia, in a short crisis such as bereavement or a hospital stay, or as a bridge while CBT-I starts.

Depression is a weaker reason than it looks. Across 19 randomised trials, CBT-I roughly doubled the depression response rate compared with control — a reason to do CBT-I, not to skip it.[15]

Treat any drug ranking warily. The largest synthesis, of 170 randomised trials, found sustained benefit beyond placebo for only eszopiclone and lemborexant, found daridorexant and suvorexant, like the benzodiazepines, poorly tolerated, and judged lemborexant's long-term safety data inconclusive.[16] The order below reflects how well studied each drug is and how long it is meant to be taken — and missing long-term harm data is not the same as no harms.

The newer options: gentler, but small effects

If you need a pill for more than a few weeks, an orexin blocker or very low-dose doxepin is usually preferred — for fewer known harms, not bigger benefits.

Orexin blockers (suvorexant, lemborexant, daridorexant) damp the brain's wakefulness signal rather than sedating you. All three beat placebo, but by small margins — possibly too small to notice — in trials of one to six months.[17] Their advantage is that stopping abruptly showed no tolerance, withdrawal or rebound insomnia in those trials. Their harms:

  • Some abuse potential. All three are US controlled substances,[18] though at its prescribed dose daridorexant gave recreational sedative users less drug-liking than an above-normal dose of zolpidem (still more than placebo).[19]
  • Narcolepsy-like side effects. They roughly tripled sleep paralysis, doubled excessive daytime sleepiness and slightly raised side effects overall versus placebo; the events are uncommon, so the estimates are imprecise.[20]
  • Fractures: unresolved. One matched study found more hip fractures on suvorexant than on a Z-drug or benzodiazepine,[21] while a meta-analysis of over 46,000 people found no excess of falls or fractures.[22]

Low-dose doxepin (3–6 mg), a sedating antidepressant at a fraction of its usual dose, gave a small-to-medium benefit for staying asleep without significant next-day grogginess, in few, mostly industry-funded trials; benefit-risk and withdrawal effects remain unsettled.[23] Keep the dose low: at higher doses doxepin is one of the anticholinergics in the dementia research below.

Ramelteon, a melatonin-receptor drug that is neither a sedative nor a controlled substance, is weakly recommended for trouble falling asleep only — not for staying asleep.[24]

The older pills: short-term only

Z-drugs and benzodiazepines work at first, but long-term use brings fractures and dependence — and getting off them is more achievable than people expect.

Z-drugs (zolpidem, zaleplon, eszopiclone) are recommended only weakly and only for short periods;[25] the European guideline says four weeks at most where possible.[26]

  • Hip fractures. Pooled observational data put the relative risk 90% higher for Z-drug users and 52% higher for benzodiazepine users, and more than double for either in the first weeks of a new prescription.[27] People given sleeping pills are often frailer already, which inflates the figure, but the new-user spike is hard to explain away.
  • Complex sleep behaviours. Sleep-driving, sleep-eating and sleep-walking are rare, but carry the US Food and Drug Administration's strongest (boxed) warning on all three drugs, and a previous episode rules them out.[28]

Trazodone is widely prescribed off-label, but across 7 placebo-controlled trials (429 patients) it did not significantly improve sleep efficiency, time to fall asleep, total sleep time or time awake at night — only early awakenings and how well people felt they slept.[29] It causes meaningful next-day sedation, and makes sense mainly when depression is present.

Benzodiazepines (temazepam, triazolam, lorazepam and others) bring falls, cognitive impairment, dependence and withdrawal, plus the hip-fracture risk above — and in an older adult a hip fracture carries substantial mortality. They are not recommended for chronic use and are on the Beers Criteria list of drugs to avoid in older adults.

Coming off. Benzodiazepine withdrawal can be severe, so never stop abruptly. With information and a tapering plan given to patients themselves, roughly a fifth to two-thirds of long-term users in their late sixties and older came off benzodiazepines or Z-drugs, in studies mostly at high risk of bias.[30] The route is a slow taper with clinician support, ideally alongside CBT-I.

Why to skip over-the-counter sleep aids

Antihistamine sleep aids such as diphenhydramine (Benadryl, ZzzQuil) seem to stop sedating within days and are linked to dementia with long-term use — a poor regular sleep aid at any adult age.

In a small crossover trial of fifteen healthy men, diphenhydramine's daytime drowsiness had vanished by day four.[31] Both the American and European guidelines advise against antihistamines for insomnia.[32] [33] They also bring daytime sedation and cognitive effects.

The dementia signal. In adults aged 65 and over, the heaviest cumulative use of anticholinergic drugs carried a 54% higher dementia risk than none, rising step by step with exposure.[34] It is observational — early dementia may itself drive antihistamine use — and pooling 26 studies put the association lower;[35] the direction has held, the size for antihistamines is less certain.

From age 65, the Beers Criteria say to avoid first-generation antihistamines.[36] Below 65 there is no evidence-based cut-off, but they still work poorly.

Special cases

Treat whatever is driving the insomnia, alongside CBT-I.

Midlife women with hot flushes. CBT-I is still first-line, and when hot flushes are what breaks up sleep, treating the flushes is a better target than a sleeping pill. Discuss menopausal hormone therapy with a clinician, particularly within 10 years of menopause and without contraindications. Otherwise: low-dose paroxetine, gabapentin, venlafaxine or the newer neurokinin blockers. Elinzanetant has the best sleep data: sleep disturbance improved in two phase 3 trials, both run by the manufacturer.[37] Fezolinetant carries a boxed warning for serious liver injury: liver tests before starting, monthly for three months, then at six and nine months, and never with cirrhosis or when liver enzymes or bilirubin are already at twice the upper limit of normal.[38]

Midlife men with low energy or libido. Get screened for sleep apnea before considering testosterone; untreated apnea causes the very fatigue that sends men looking for it. But treating apnea is not a testosterone fix: across twelve studies in 388 men, CPAP did not raise total testosterone.[39] See testosterone therapy.

Insomnia and sleep apnea together. The pair is under-recognised and worse than either alone: among 5,236 adults followed for fifteen years, having both carried about 47% higher all-cause mortality than having neither.[40] CBT-I works whether or not the apnea is treated,[41] and doing it first gave about an extra hour of CPAP use per night in a randomised trial.[42] So treat the insomnia behaviourally first; a hypnotic risks masking undiagnosed apnea — see Sleep-disordered breathing.

Older adults. Where a drug is needed, an orexin blocker or low-dose doxepin is the usual choice, for fewer known harms rather than proven safety, and avoid benzodiazepines, Z-drugs and antihistamines where possible. A 2025 model projected that if Americans over 50 stopped prescription sleep medication, lifetime falls would drop 8.5% and life expectancy rise by about 40 days — but it is a model built on observational risks that leaves out sleep apnea, and some of the harm it assigns to the drugs may belong to the insomnia.[43]

Supplements have modest evidence at best and none replaces CBT-I — see Sleep supplements.

Further reading

  • Sateia MJ et al. Clinical Practice Guideline for the Pharmacologic Treatment of Chronic Insomnia in Adults: An American Academy of Sleep Medicine Clinical Practice Guideline. J Clin Sleep Med 2017.[44]
  • Edinger JD et al. Behavioral and psychological treatments for chronic insomnia disorder in adults: an American Academy of Sleep Medicine systematic review, meta-analysis, and GRADE assessment. J Clin Sleep Med 2021.[45]
  • van Straten A et al. Cognitive and behavioral therapies in the treatment of insomnia: a meta-analysis. Sleep Med Rev 2018 — 87 RCTs.[46]
  • Furukawa Y et al. Components and Delivery Formats of Cognitive Behavioral Therapy for Chronic Insomnia in Adults: A Systematic Review and Component Network Meta-Analysis. JAMA Psychiatry 2024 — 241 trials.[47]
  • Kyle SD et al. Clinical and cost-effectiveness of nurse-delivered sleep restriction therapy for insomnia in primary care (HABIT). Lancet 2023.[48]
  • Kallestad H et al. Mode of delivery of Cognitive Behavioral Therapy for Insomnia: a randomized controlled non-inferiority trial of digital and face-to-face therapy. Sleep 2021.[49]
  • Simon L et al. Comparative efficacy of onsite, digital, and other settings for cognitive behavioural therapy for insomnia: a systematic review and network meta-analysis. Sci Rep 2023, with author correction 2025.[50] [51]
  • De Crescenzo F et al. Comparative effects of pharmacological interventions for the acute and long-term management of insomnia disorder in adults: a systematic review and network meta-analysis. Lancet 2022 — 170 trials, 36 interventions.[52]
  • Kishi T et al. Comparative efficacy and safety of daridorexant, lemborexant, and suvorexant for insomnia: a systematic review and network meta-analysis. Transl Psychiatry 2025.[53]
  • Na HJ et al. Clinical safety and narcolepsy-like symptoms of dual orexin receptor antagonists in patients with insomnia: a systematic review and meta-analysis. Sleep 2024 — 11 RCTs, 7,703 patients.[54]
  • Donnelly K et al. Benzodiazepines, Z-drugs and the risk of hip fracture: a systematic review and meta-analysis. PLOS ONE 2017.[55]
  • Gray SL et al. Cumulative use of strong anticholinergics and incident dementia: a prospective cohort study. JAMA Intern Med 2015.[56]
  • Lechat B et al. Comorbid insomnia and sleep apnoea is associated with all-cause mortality. Eur Respir J 2022.[57]
  • Riemann D et al. The European Insomnia Guideline: An update on the diagnosis and treatment of insomnia 2023. J Sleep Res 2023.[58]

— § —