Name: Sleeping Pills: What They Do, What They Cost, and Why They're Last
Creator: Realised
Published: 2026-06-01

Summary

Sleeping pills — prescription Z-drugs (zolpidem) and benzodiazepines, OTC antihistamines (diphenhydramine/doxylamine), and newer orexin-receptor antagonists — all share one problem: they manage the symptom without fixing the insomnia, and the durable fix (CBT-I) is non-drug; the prescription sedatives buy a modest objective gain (Z-drugs cut sleep-onset latency by roughly 22 minutes) at the price of tolerance, dependence, rebound insomnia, and falls/cognitive impairment that hit the elderly hardest, while the OTC antihistamine route is the worst-evidenced of all — fast tolerance, next-day grog

Why Strong

Tier 1 because: the behaviour-first hierarchy rests on strong recommendations from two independent guideline bodies (AASM 2021, ACP 2016); the efficacy magnitudes and harm signals come from meta-analyses (Buscemi 2007 efficacy; Glass 2005 harm-benefit in elderly) and a regulatory safety action (FDA 2019 boxed warning); and the anticholinergic-dementia association is a large prospective cohort with dose-response plus Beers-criteria endorsement.

NOT Tier 0.5 because it carries genuine clinical nuance and a real two-sided controversy (legitimate acute use vs over-prescription), not a single undisputed axiom.
NOT Tier 2 because the core claims are replicated/guideline-level, not "a few suggestive studies" — only the orexin-antagonist long-term safety advantage sits lower (Tier 2, younger evidence), and the entry marks it as such.

Practical takeaway

The throughline: behaviour first, drugs last and brief.
• Start with CBT-I, not a pill. For chronic insomnia (3+ nights/week, 3+ months), the guideline-first move is CBT-I — it's the only approach with durable effect that outlasts treatment. See cbt_i_program_overview; its core engine, sleep_restriction_therapy, is what actually rebuilds sleep drive. For middle-of-the-night waking specifically, see sleep_maintenance_insomnia.
• If a drug is genuinely warranted, keep it acute and time-boxed. Situational insomnia (bereavement, a crisis week, jet lag, hospital) is a reasonable place for a short course — think days, not months — ideally with a plan to stop before tolerance/dependence set in. Use it as a bridge while behavioural tools come online, not as the destination.
• Avoid the OTC antihistamine route for anything routine — and especially over 65. Diphenhydramine/doxylamine "PM" aids are not a benign starter option; they tolerate out within days and carry the anticholinergic/dementia concern. If a non-prescription nudge is wanted, the better-tolerated evidence-graded options live elsewhere (magnesium_supplementation_sleep, glycine_supplementation_for_sleep) — modest, but without the anticholinergic burden.
• Never stop a regular benzodiazepine/Z-drug abruptly. Long-standing nightly use should be tapered with a prescriber — abrupt cessation can cause severe rebound insomnia and, with benzodiazepines, dangerous withdrawal (including seizures). Deprescribing works best paired with CBT-I, which fills the gap the drug leaves.
• If a pharmacologic agent is needed longer, an orexin antagonist is usually the more favourable class to discuss with a prescriber than a Z-drug or benzodiazepine — but it is still an adjunct to behavioural treatment, not a substitute for it.
• Tell any prescriber the full picture — age, falls history, other sedating/anticholinergic medications, and alcohol use (which compounds all of these and itself wrecks sleep architecture; see the diet/sleep alcohol entries).

Evidence detail

Why This Entry Exists

Insomnia is miserable and a pill is right there — on prescription after a five-minute appointment, or off the shelf at any pharmacy for a few pounds. The market built around that misery is enormous, and it is split in two: the pharma side selling branded hypnotics, and the OTC sleep-aid side selling "PM" formulations that are just an antihistamine repackaged for sleep. Both sell the same implicit promise — that the problem is a missing sedative rather than a behavioural and cognitive pattern that can be retrained.

This entry is the honest map of that shelf. It exists to do two things at once: (1) tell a person what each drug class actually does and costs, in plain magnitudes, so they're not choosing blind; and (2) hold the line that these are last-resort, short-term tools, not a chronic solution — because the guideline-endorsed durable treatment for chronic insomnia is CBT-I, which this entry is a spoke of. The goal is not "never take a sleeping pill" (there's a real acute role) — it's "don't mistake a sedative for a cure, and don't drift into nightly use that's harder to stop than it was to start."

What bad advice this protects against, both ways:
• "Just take something to sleep" → drifting into nightly hypnotic or antihistamine use that builds tolerance, breeds dependence, and produces rebound insomnia worse than the original when you stop.
• "Sleeping pills are poison, never touch them" → the opposite overcorrection that leaves someone white-knuckling acute, situational insomnia (bereavement, jet lag, a crisis week) when a few nights of a hypnotic is a reasonable bridge.
• "OTC is safer because it's not prescription" → diphenhydramine ("PM" sleep aids) is arguably the worst routine choice, especially over 65 — anticholinergic, fast-tolerant, and carrying a dementia-risk association.
• "The drug fixed my insomnia" → it suppressed the symptom; the insomnia is intact and will resurface (often as rebound) the moment the drug stops, because nothing addressed the maintaining pattern (see cbt_i_program_overview).

Evidence

1. CBT-I is first-line; medication is not (Tier 1). Both major guideline bodies put behavioural treatment first. The American Academy of Sleep Medicine (Edinger et al., 2021, J Clin Sleep Med) issued a strong recommendation for CBT-I as the treatment for chronic insomnia. The American College of Physicians (Qaseem et al., 2016, Ann Intern Med) recommends all adults receive CBT-I as the initial treatment, and that if drugs are used they be used short-term (ideally ≤4–5 weeks) as an adjunct, not a standing therapy. The pill is the fallback; the behaviour change is the treatment.

2. Z-drugs and benzodiazepines: the objective benefit is real but modest (Tier 1). A meta-analysis of non-benzodiazepine hypnotics found Z-drugs reduced polysomnographic sleep-onset latency by ~22 minutes versus placebo (Buscemi et al., 2007) — and a large share of even that gain overlaps with placebo response. They get you to sleep somewhat faster; they do not deliver the dramatic, restorative night the marketing implies, and tolerance erodes the effect over weeks.

3. The harms cluster in the elderly, and can outweigh the benefit (Tier 1). Glass et al. (2005, BMJ) meta-analysed sedative hypnotics in older adults and found the number-needed-to-harm was comparable to (and in some analyses exceeded) the number-needed-to-treat — i.e., for every older person helped to sleep, a similar number experienced an adverse event (next-day cognitive impairment, daytime fatigue, psychomotor/balance impairment). Hypnotic and Z-drug use is repeatedly associated with falls and hip fractures in older people. This is why "older adult + nightly sleeping pill" is a red flag, not a routine.

4. Tolerance, dependence, and rebound insomnia are the core trap (Tier 1). Benzodiazepines and Z-drugs act on GABA-A receptors; with regular use, receptor adaptation produces tolerance (the dose stops working), physical dependence, and on discontinuation rebound insomnia — sleep transiently worse than baseline — which the brain naturally interprets as "I still need the pill." That feedback loop is what converts a short prescription into chronic use. Guideline-endorsed short-course limits exist precisely to avoid entering it.

5. FDA boxed warning — complex sleep behaviours (Tier 1). In 2019 the FDA added its strongest warning, a Boxed Warning, to eszopiclone (Lunesta), zaleplon (Sonata), and zolpidem (Ambien) for rare but serious complex sleep behaviours — sleepwalking, sleep-driving, and other activities while not fully awake — that have caused serious injuries and deaths. The FDA also contraindicated these drugs in anyone who has previously had such an episode. This is not a theoretical risk; it is the regulator's most prominent safety flag.

6. OTC antihistamines: fast tolerance + the dementia association (Tier 1 for the harm signal). Diphenhydramine and doxylamine (the active ingredient in most "PM"/"night-time" sleep aids) are sedating first-generation antihistamines that are also strongly anticholinergic. Tolerance to the sedative effect develops within days, next-day grogginess is common, and — critically — Gray et al. (2015, JAMA Internal Medicine), a prospective cohort with ~7-year follow-up, found a dose-dependent association between cumulative anticholinergic burden and incident dementia (~54% higher dementia risk in the highest-use group), possibly not reversible on stopping. The AGS Beers Criteria list first-generation antihistamines among medications older adults should avoid. For routine insomnia, this is the worst-evidenced option on the shelf.

7. Orexin-receptor antagonists: a genuinely different mechanism, still not first-line (Tier 1–2). Suvorexant, lemborexant, and daridorexant (dual orexin-receptor antagonists, DORAs) block wake-promoting orexin signalling rather than broadly sedating via GABA. They show efficacy for sleep onset and maintenance, and early data suggest less physical dependence/withdrawal and less next-morning impairment than Z-drugs/benzodiazepines (daridorexant in particular shows minimal residual effects and no clear dependence signal in trial data). They are a more favourable pharmacologic option where one is needed — but they remain adjunct, not the primary treatment, the long-term and real-world safety record is younger and thinner, and they are expensive.

Mechanism

Why GABA hypnotics breed dependence. Benzodiazepines and Z-drugs (zolpidem etc.) potentiate GABA-A, the brain's main inhibitory system — they push the whole nervous system toward sedation. The brain homeostatically down-regulates in response (fewer/less-sensitive receptors), which is tolerance; remove the drug and the now under-damped system rebounds into hyperarousal — rebound insomnia and, with benzodiazepines especially, a withdrawal syndrome. The pill creates the physiological state that makes the next pill feel necessary.

Why antihistamines fail fast and carry an anticholinergic cost. First-generation antihistamines (diphenhydramine, doxylamine) cross the blood-brain barrier and sedate by blocking central H1 histamine — but they also block muscarinic acetylcholine receptors (anticholinergic), causing dry mouth, constipation, urinary retention, confusion, and next-day cognitive dullness. Tolerance to the H1-mediated sedation develops within a few nights. The anticholinergic action is the mechanistically plausible link to the cognitive/dementia signal: acetylcholine is central to memory and cognition, and chronic cumulative blockade in the ageing brain is the suspected pathway.

Why orexin antagonists are mechanistically cleaner. Orexin (hypocretin) is a wake-promoting neuropeptide — narcolepsy is essentially orexin deficiency. DORAs don't impose sedation; they remove a wake signal, letting the body's own sleep pressure take over. Because they don't broadly suppress the GABA system, the tolerance/dependence/rebound dynamics appear muted — though "appear muted in trials" is not "proven safe long-term in the real world."

Why none of them treat the insomnia. Chronic insomnia is maintained by conditioned arousal and unhelpful sleep-related behaviours/beliefs (the bed becomes a cue for wakefulness; effort to sleep backfires). A sedative overrides that tonight without touching the pattern — so the pattern is fully intact the moment the drug stops. CBT-I works precisely because it retrains the pattern (cbt_i_program_overview, sleep_restriction_therapy).

Risks And Contraindications

• Falls and fractures (elderly): Z-drugs, benzodiazepines, and sedating antihistamines all raise fall/hip-fracture risk in older adults — frequently the single most consequential harm. Nightly hypnotic use over 65 deserves active review.
• Complex sleep behaviours (Z-drugs): FDA boxed warning — sleepwalking/sleep-driving with injury and death; contraindicated after any prior episode. Risk rises with higher doses and with alcohol/other CNS depressants.
• Next-day impairment: psychomotor and cognitive carry-over (the "hangover") impairs driving and complex tasks the following morning — present across classes, dose- and half-life-dependent.
• Dependence, tolerance, rebound: core to benzodiazepines and Z-drugs; do not stop abruptly — taper under medical supervision.
• Anticholinergic burden (OTC antihistamines): dementia-risk association (dose-dependent), plus confusion, urinary retention, constipation, dry mouth; on the Beers list to avoid in older adults. Caution in glaucoma, BPH, and anyone already on anticholinergic medications.
• Interactions and special populations: all sedative-hypnotics compound with alcohol, opioids, and other CNS depressants (respiratory-depression risk). Use in pregnancy, significant respiratory disease (e.g., untreated sleep apnoea — sedatives can worsen it), and substance-use history requires specialist judgement, not self-medication.
• Masking, not treating: the central risk is that the pill papers over a treatable condition (or an underlying one — apnoea, depression, restless legs) that should be diagnosed and addressed.

Controversy

Nature: clinical + commercial, with overstatement at both poles — pharma/OTC industries that profit from chronic sedative use, and a wellness counter-current that demonises all sleep medication.

Position A — "Sleeping pills are an effective, appropriate treatment for insomnia." The prescribing-convenience and pharma-marketing take.
• Best evidence: hypnotics do produce a measurable objective benefit (≈22 min faster sleep onset for Z-drugs) and have a legitimate short-term, acute role; orexin antagonists are mechanistically cleaner.
• Where it's wrong: the benefit is modest and tolerates out; they don't treat chronic insomnia (CBT-I does); and routine/chronic use carries dependence, rebound, falls, and (for OTC antihistamines) a dementia signal. Guidelines explicitly relegate them to short-term adjunct.

Position B — "All sleeping pills are dangerous; never take them." The wellness/naturalistic overcorrection.
• Best evidence: chronic use really is harmful and over-prescribed, and the durable fix is non-drug.
• Where it's wrong: blanket refusal abandons people in acute, situational insomnia where a brief course is humane and reasonable, and it can scare patients into abruptly stopping a long-standing benzodiazepine — which is genuinely dangerous. "Last resort, short-term, supervised" is not "never."

The funding/bias dimension: the pharma industry's incentive is chronic prescribing (recurring revenue), and historically branded hypnotics were marketed well beyond their evidence; the OTC industry's incentive is volume — diphenhydramine repackaged as a "PM sleep aid" sells on the impression that non-prescription = safe, which the dementia/anticholinergic data contradict. On the other side, the wellness market profits from "drug-free sleep" products and fear of pharmaceuticals. The cleanest signal cuts across all of them: independent guideline bodies (AASM, ACP) and FDA safety actions converge on behaviour-first, drugs-brief.

Realised Position: CBT-I is the treatment; sleeping pills are a short-term, last-resort bridge for acute situations, not a chronic solution. Z-drugs/benzodiazepines: modest benefit, real dependence/rebound/falls risk and an FDA boxed warning — keep brief, taper never-abruptly. OTC antihistamines: avoid for routine use, especially over 65 (anticholinergic/dementia signal, Beers-listed). Orexin antagonists: the more favourable pharmacologic class if one is truly needed, still adjunct-only. We won't fearmonger anyone out of a justified short course — but we won't let a sedative masquerade as a cure.

Cross-Pillar Connections

• Sleep — the durable alternative (cbt_i_program_overview, sleep_restriction_therapy, sleep_maintenance_insomnia): this entry is a spoke of the CBT-I hub; the pill is the fallback, the behavioural program is the treatment, and sleep restriction is the engine that rebuilds sleep drive.
• Sleep — gentler non-drug nudges (magnesium_supplementation_sleep, glycine_supplementation_for_sleep): modest, better-tolerated supplement options that don't carry anticholinergic burden — the honest "if you want to take something" answer for mild cases.
• Diet/Mental — compounders: alcohol (compounds sedation, wrecks sleep architecture) and caffeine timing interact heavily with all of this; chronic stress and depression frequently underlie insomnia and should be screened rather than sedated.
• Cross-pillar — evidence literacy: the "OTC = safe" and "non-habit-forming" claims are textbook cases for the evidence-reading and cui-bono skills (marketing magnitude vs measured magnitude; who profits from chronic use).

What would change our mind

Falsifiability: explicit upgrade/downgrade criteria from source

• We'd soften the "chronic use" stance if long-term RCTs showed a hypnotic class delivering durable insomnia remission that persists after discontinuation, without tolerance — i.e., matching CBT-I's defining property. No current class does.
• We'd upgrade orexin antagonists toward a more routine role if mature, independent, long-horizon real-world data confirmed the low-dependence / low-impairment signal and showed outcomes rivalling behavioural treatment.
• We'd revise the antihistamine warning if well-controlled, confounder-robust studies overturned the cumulative-anticholinergic dementia association (current evidence is observational but dose-responsive and mechanistically coherent — strong enough to act on cautiously).
• What would NOT move us: marketing claims of a "non-habit-forming" sedative, short-term efficacy data presented as a case for chronic use, or "OTC = safe" reasoning. None address the core problem — a sedative is not a treatment for the maintaining pattern.

Industry bias note

Structural incentives the evidence base may reflect

This is a topic where industry profits at both the drug and the anti-drug poles, so independent guidelines and regulators are the anchor.
• Pharma (prescription): incentive favours chronic prescribing; branded hypnotics have a history of being marketed beyond their evidence, and the modest objective benefit is routinely overstated.
• OTC sleep-aid market: sells diphenhydramine/doxylamine as benign "night-time/PM" aids — the "non-prescription = safe" framing directly contradicts the anticholinergic/dementia and rapid-tolerance data.
• Wellness/naturalistic market: profits from "drug-free sleep" supplements and from fear of all pharmaceuticals — a counter-overstatement that can endanger people who stop benzodiazepines abruptly.
• The clean signal: AASM (Edinger 2021) and ACP (Qaseem 2016) guidelines, the FDA 2019 boxed warning, and the Glass/Buscemi/Gray meta-analytic and cohort data converge on behaviour-first, drugs-brief, antihistamines-avoid-in-elderly. Realised weights those over every commercial narrative on either side.

Sleeping Pills: What They Do, What They Cost, and Why They're Last