Insomnia

Insomnia

Insomnia is a complaint about sleep that has been asked to do the work of a disease, a risk factor, a lifestyle failure, and a drug indication. Those are not the same object. This article keeps the diagnostic construct, the physiology, the learned behaviour, and the trial endpoints in separate sentences, and it does not convert a questionnaire change into a polysomnographic cure.

Compiled by South Beach Longevity · 20 August 2026 Copyright 2026 Series SBL-41 / SP-INSOMNIA · Register A scientific article Sources peer-reviewed trials, meta-analyses, consensus guidelines, and labelled mechanistic work · verified NCBI records Constraint This document describes published research. It is not medical advice. No human use, dose, route or schedule is recommended anywhere in this document.

How to read this document Every finding is labelled, in the sentence that reports it, by the kind of study that produced it. A sleep-diary minute is not a polysomnographic minute. A fall in the Insomnia Severity Index is not a cure of hyperarousal. A hypnotic that lengthens scored sleep is not a demonstration that the learned insomnia has been unlearned. Amounts and durations appear only as reported experimental or labelling parameters, always with the population attached. Nothing here is a recommendation.

Five grades are used in the matrices: STRONG, MODERATE, LIMITED, INCONSISTENT, NOT SUPPORTED. They grade the reviewed record in hand, not a commercial future. Sleep-hygiene rows and CBT-I rows are never collapsed. Subjective and objective rows are never collapsed.

01 Definitions

Contemporary insomnia is a complaint-defined disorder. The person reports difficulty initiating sleep, difficulty maintaining sleep, or waking earlier than desired, together with daytime consequence, despite adequate opportunity and circumstances for sleep. The International Classification of Sleep Disorders, third edition, organises this as Chronic Insomnia Disorder and Short-Term Insomnia Disorder and dropped the older primary/secondary split as a required first cut (Sateia, 2014). Research diagnostic criteria assembled by Edinger, Bonnet, Bootzin, and colleagues had already required a similar complaint-plus-daytime-impairment structure and had warned that laboratory sleep alone does not make the diagnosis (Edinger et al., 2004). Lichstein, Durrence, Taylor, Bush, and Riedel’s quantitative research thresholds (sleep-onset latency or wake after sleep onset of at least 31 minutes, three or more nights per week) are operational tools for trials, not a competing clinical ontology (Lichstein et al., 2003).

The DSM-5 insomnia-disorder criteria are aligned with that complaint structure: frequency (three or more nights per week), duration (three or more months for the persistent form), opportunity, daytime impairment, and exclusion of a better-explaining sleep–wake disorder. Alignment is not identity. A clinic that treats “insomnia” as any bad night, any short actigraphy window, or any hypnotic refill is not using the construct the trials enrolled.

Four consequences follow and will be enforced for the rest of this document.

First, insomnia is not a polysomnographic disease. Poor sleep on a laboratory night can exist without the complaint, and the complaint can exist with sleep that scores inside population norms (Edinger and Fins, 1995; Rosa and Bonnet, 2000). Second, insomnia is not insufficient sleep opportunity. A person who sleeps five hours because the schedule allows five hours has sleep restriction, not insomnia disorder. Third, insomnia is not obstructive sleep apnea, restless legs, or a circadian phase disorder, though any of those can coexist. Fourth, this document is not a treatment protocol. Reported milligrams, session counts, and time-in-bed prescriptions appear only as study parameters.

02 Acute versus chronic

Short-term insomnia is common after a stressor, a schedule shock, illness, or bereavement. Most of it remits. Chronic insomnia disorder is the object of almost every treatment trial cited below: the complaint has lasted three months or longer, typically with at least three affected nights per week. Spielman’s behavioural account treated acute insomnia as a predictable response to precipitants and chronic insomnia as what remains after those precipitants have been amplified by conditioning and by extending time in bed (Spielman, Caruso, and Glovinsky, 1987). That is a model, not a stopwatch. A person can meet chronic duration criteria after a single unremitting episode, or after repeated acute bouts that never quite clear.

The trial literature is overwhelmingly a chronic-insomnia literature. Extrapolating a six-week hypnotic study to a three-year complaint, or a six-session CBT-I protocol to a two-week jet-lag week, is a category error. Acute insomnia can still be disabling. It is not the evidence base that supports first-line chronic-insomnia recommendations (Qaseem et al., 2016; Edinger et al., 2021; Riemann et al., 2017).

03 Sleep-onset versus maintenance versus mixed

Sleep-onset insomnia is a long sleep-onset latency. Sleep-maintenance insomnia is wake after sleep onset, frequent awakenings, or early-morning awakening with inability to return to sleep. Mixed insomnia is the ordinary clinical picture, not a rare third type. Drug programmes often pick one pole because the pharmacology is easier to brand: a short-acting Z-drug for onset, a maintenance claim for a longer half-life or for a low-dose sedating antidepressant, an orexin antagonist for wake-to-sleep transition across the night. Behavioural trials usually treat the syndrome.

The distinction matters for mechanism more than for the CBT-I package. Conditioned bed-arousal can present as onset difficulty. A phase-advanced circadian alerting signal can present as early awakening. Nocturia, pain, apnea, and restless legs can present as maintenance difficulty and are not insomnia until those contributors have been named. Matrices later in this document keep onset-latency minutes and wake-after-sleep-onset minutes in separate cells wherever the source did.

04 What insomnia is not

Obstructive sleep apnea produces sleep fragmentation, unrefreshing sleep, and daytime sleepiness; it is diagnosed from a sleep study plus a compatible history, not from a diary complaint of “can’t sleep.” Insomnia and apnea coexist often enough that treating one as a synonym for the other is a clinical error. The STOP-Bang questionnaire is a screening instrument for apnea risk, not an insomnia scale (Chung et al., 2008). Restless legs syndrome / Willis–Ekbom disease is an urge to move, worse at rest, worse in the evening, relieved by movement; diagnostic criteria are clinical (Allen et al., 2014). Periodic limb movements can fragment sleep without the waking urge. Circadian rhythm sleep–wake disorders, including delayed and advanced sleep–wake phase, are timing disorders: the person may sleep well at the wrong clock time (Sack et al., 2007; Auger et al., 2015). Melatonin evidence in those disorders is not melatonin evidence in insomnia disorder.

Insufficient opportunity, shift work, and environmental disruption are not insomnia disorder. They are reasons a person is not asleep. The insomnia construct begins only when opportunity is adequate and the sleep system still will not deliver.

05 Prevalence

Ohayon’s epidemiological reviews placed insomnia symptoms in a large minority of adults and insomnia disorder, under stricter operational rules, closer to one adult in ten, with higher rates in women and in older adults (Ohayon, 2002). Morin and Jarrin’s later epidemiologic synthesis kept that split: symptoms are common; the syndromal disorder is less common and more stable (Morin and Jarrin, 2022). Exact percentages move with the threshold. A one-month “trouble sleeping” item is not ICSD-3 Chronic Insomnia Disorder. Sivertsen and colleagues’ population work in Norway is a reminder that prevalence is also a function of which daytime consequences are required (Sivertsen et al., 2009).

Ford and Kamerow’s classic Epidemiologic Catchment Area analysis linked sleep complaints to subsequent psychiatric disorder, especially depression, and is an epidemiologic association paper, not a sleep-laboratory paper (Ford and Kamerow, 1989). Kessler and colleagues’ America Insomnia Survey estimated work-performance losses associated with insomnia in employed U.S. adults (Kessler et al., 2011). Those are burden estimates. They are moderately supported as population economics. They are not a licence to medicalise every short night.

06 Consequences

Daytime consequence is part of the definition: fatigue, mood disturbance, cognitive complaint, impaired occupational or social function. That circularity must be respected. One cannot then treat “daytime fatigue” as an independent discovery that insomnia is harmful.

Longitudinal epidemiology is stronger on psychiatric follow-through than on a single cardiometabolic story. Baglioni, Battagliese, Feige, Spiegelhalder, and colleagues’ meta-analysis of longitudinal studies found that insomnia conferred an elevated risk of subsequent depression (Baglioni et al., 2011). Hertenstein, Feige, Gmeiner, and colleagues extended the prediction to a wider set of incident mental disorders (Hertenstein et al., 2019). Those packages are moderately supported. They do not prove that treating insomnia prevents depression in every population, although Manber, Edinger, Gress, and colleagues showed that adding CBT-I to antidepressant treatment improved depression outcomes in patients who had both conditions (Manber et al., 2008). Wu, Appleman, Salazar, and Ong’s review of CBT-I in medical and psychiatric comorbidity found sleep improvements that often travelled with the comorbid condition rather than against it (Wu et al., 2015).

Cardiometabolic work is mixed and easily over-read. Sofi, Cesari, Casini, and colleagues reported an association between insomnia and cardiovascular disease in a meta-analysis of observational studies (Sofi et al., 2014). Laugsand, Strand, Platou, Vatten, and Janszky linked insomnia symptoms to incident heart failure in a Norwegian cohort (Laugsand et al., 2013). Those are association findings. They are not a demonstration that a hypnotic prescription is cardioprotective. Cappuccio, D’Elia, Strazzullo, and Miller’s mortality meta-analysis concerned sleep duration, short and long, not insomnia disorder (Cappuccio et al., 2010). Duration and insomnia are neighbouring variables. They are not interchangeable. Vgontzas, Fernandez-Mendoza, Liao, and Bixler have argued that insomnia with objective short sleep is the more biologically severe phenotype, with stronger medical associations than insomnia with normal-length scored sleep (Vgontzas et al., 2013). That phenotype argument is emerging. It is one reason this article refuses to treat all insomnia as one physiological object.

Léger, Guilleminault, Bader, Lévy, and Paillard, and Walsh, mapped socio-professional and health-care burden (Léger et al., 2002; Walsh, 2004). Daley, Morin, LeBlanc, Grégoire, and Savard associated insomnia with health-care utilisation (Daley et al., 2009). Burden is strongly supported. Causal arrows from a specific treatment to averted myocardial infarction are not supported.

07 Hyperarousal

The most durable physiological account of chronic insomnia is not “too little sleep drive.” It is too much wake, over 24 hours. Bonnet and Arand measured a higher 24-hour metabolic rate in people with insomnia than in matched controls and treated that as a signature of hyperarousal rather than of sleep loss alone (Bonnet and Arand, 1995; Bonnet and Arand, 2010). Vgontzas, Bixler, Lin, and colleagues reported nyctohemeral activation of the hypothalamic–pituitary–adrenal axis in chronic insomnia (Vgontzas et al., 2001). Nofzinger, Buysse, Germain, Price, Miewald, and Kupfer’s PET work showed that the usual sleep-related fall in relative metabolism in wake-promoting regions was blunted in insomnia (Nofzinger et al., 2004). Perlis, Merica, Smith, and Giles described elevated beta/gamma EEG activity around sleep onset as a cortical counterpart of the same idea (Perlis et al., 2001). Riemann, Spiegelhalder, Feige, and colleagues assembled these strands into an explicit hyperarousal model (Riemann et al., 2010; Riemann et al., 2015). Spiegelhalder, Regen, Baglioni, Nissen, Riemann, and Kyle reviewed the neuroimaging file and found it consistent with, not proof of, a single arousal circuit (Spiegelhalder et al., 2015).

Hyperarousal is strongly supported as a between-group physiological difference in chronic insomnia samples. It is not a demonstrated requirement for every person who meets diary criteria. It is also not a licence to equate insomnia with anxiety disorder. The two travel together. They are not the same diagnosis.

08 Homeostatic sleep drive

Process S, in Borbély’s two-process model, is the homeostatic sleep drive that accumulates in wake and dissipates in sleep (Borbély, 1982; Borbély, Daan, Wirz-Justice, and Deboer, 2016). Sleep restriction therapy is, in mechanistic costume, an attempt to rebuild Process S by concentrating sleep opportunity (Spielman, Saskin, and Thorpy, 1987; Kyle, Morgan, Spiegelhalder, and Espie, 2011). The model is strongly supported as a description of normal sleep regulation. The inference that chronic insomnia is simply a weak Process S is not supported. People with insomnia often spend long hours in bed producing little sleep. That looks like a diluted homeostat. It is equally compatible with a strong wake drive sitting on a normal homeostat.

Maurer, Espie, Omlin, and Kyle’s experimental work on sleep restriction treated the therapy as a probe of that physiology, not as a wellness tip (Maurer et al., 2018; Maurer, Espie, and Kyle, 2021). Miller, Espie, Epstein, Friedman, Morin, Pigeon, Spielman, and Kyle reviewed the evidence base of sleep restriction as a component and as a standalone (Miller et al., 2014). Those are behavioural-physiology papers. They are not a demonstration that everyone with insomnia is sleep-deprived in the homeostatic sense.

09 Circadian interaction

Process C is the circadian alerting signal. Insomnia is not, as a class, a circadian disorder. A delayed circadian phase can mimic sleep-onset insomnia; an advanced phase can mimic early-morning awakening. When the timing system is the primary lesion, the correct object is a circadian rhythm sleep–wake disorder, and the melatonin and light evidence belongs there (Sack et al., 2007; Auger et al., 2015). When a person with chronic insomnia has a normally timed circadian system and still cannot sleep in the permitted window, invoking “circadian hygiene” as the treatment is a slogan.

The interaction that does belong here is simpler. Hyperarousal plus a flattened homeostat plus a mistimed C produces a night that feels un-sleepable. CBT-I’s stimulus control and rise-time rules accidentally serve C as well as S. That is a side-effect of a behavioural prescription, not a circadian diagnosis.

10 Conditioning

Bootzin’s stimulus-control analysis treated the bed as a conditioned cue for arousal when it had been paired with wake, worry, and clock-watching (Bootzin and Perlis, 1992). Spielman, Caruso, and Glovinsky’s 3P model is the field’s organising cartoon: predisposing traits, precipitating events, perpetuating behaviours — especially extending time in bed, napping, and using the bed for everything except sleep (Spielman, Caruso, and Glovinsky, 1987). The model is strongly supported as a clinical account of how acute insomnia becomes chronic. It is not a pathophysiological proof. It is the reason stimulus control and sleep restriction exist as separable components rather than as “sleep hygiene plus.”

11 Cognitive models

Harvey’s cognitive model placed attention to sleep, threat interpretation, and safety behaviours at the centre of persistence (Harvey, 2002). Espie’s attention–intention–effort pathway described how trying to sleep becomes the thing that prevents sleep (Espie, Broomfield, MacMahon, Macphee, and Taylor, 2006). Perlis, Giles, Mendelson, Bootzin, and Wyatt’s neurocognitive model joined cortical arousal with conditioned cues and with the finding that people with insomnia often misperceive sleep (Perlis et al., 1997; Perlis et al., 2001). Harvey and Tang reviewed the misperception literature: many people with insomnia overestimate sleep-onset latency and underestimate total sleep relative to PSG (Harvey and Tang, 2012). Means, Edinger, Glenn, and Fins reported the same discrepancy as a measurement fact, not as malingering (Means et al., 2003).

Cognitive therapy for insomnia, tested as a component, can move beliefs and the complaint (Harvey, Sharpley, Ree, Stinson, and Clark, 2007; Edinger, Wohlgemuth, Radtke, Marsh, and Quillian, 2001). Mindfulness-based programmes, as in Ong, Manber, Segal, Xia, Shapiro, and Wyatt, are neighbouring interventions, not a replacement CBT-I evidence base (Ong et al., 2014). Grade for cognitive models as an account of persistence: MODERATE to STRONG. Grade for cognitive therapy as a standalone equivalent of the full CBT-I package: LIMITED.

12 Psychiatric and medical comorbidity

Insomnia is more often comorbid than “primary.” Depression, anxiety, PTSD, chronic pain, cancer, and neurodegenerative disease all raise the prevalence of the complaint. The older instruction to leave the insomnia untreated until the “real” disease was treated is the instruction the comorbidity trials were built to refuse. Wu and colleagues found CBT-I effective for insomnia in comorbid medical and psychiatric samples (Wu et al., 2015). Manber and colleagues found that CBT-I enhanced depression outcome when insomnia and depression were both present (Manber et al., 2008). Those are moderately supported treatment findings in mixed populations. They do not make insomnia a synonym for major depression, and they do not make an antidepressant a first-line insomnia drug.

13 Sleep history

A competent insomnia history names: the complaint pole (onset, maintenance, mixed); duration and course; weekday/weekend schedule; time in bed versus estimated sleep; naps; clock-watching; caffeine, alcohol, and nicotine timing; pain, nocturia, reflux; mood and trauma; snoring, witnessed apneas, restless legs; shift work and light; and what has already been tried. Chesson, Anderson, Littner, Davila, Hartse, and colleagues’ practice parameters for evaluating chronic insomnia, and Schutte-Rodin’s AASM clinical guideline, remain the structural references (Chesson et al., 2000; Schutte-Rodin, Broch, Buysse, Dorsey, and Sateia, 2008). A history that skips apnea and restless-legs questions is not an insomnia assessment. It is a prescription prelude.

14 Diaries

The Consensus Sleep Diary is the field’s attempt to stop every laboratory inventing its own morning form (Carney et al., 2012). A two-week diary yields sleep-onset latency, wake after sleep onset, total sleep time, time in bed, and sleep efficiency as the person estimates them. Those estimates are the CBT-I inputs. They are not PSG. They are also not optional decoration. Without a diary, sleep restriction has no denominator.

15 Actigraphy

Actigraphy estimates sleep from movement. AASM reviews treat it as useful for documenting patterns, circadian timing, and treatment response, and as weaker than PSG for staging (Littner et al., 2003; Smith et al., 2018). In insomnia it often scores more sleep than the diary admits. That discrepancy is data. It is not automatically a reason to dismiss the complaint. People with insomnia lie still while awake. The device then calls the stillness sleep.

16 Polysomnography: what it can and cannot do

PSG is required when apnea, narcolepsy, or certain parasomnias are in play. It is not required to diagnose insomnia disorder (Schutte-Rodin et al., 2008; Edinger et al., 2021). Edinger and Fins showed that the laboratory night and the complaint can disagree (Edinger and Fins, 1995). Rosa and Bonnet found that reported chronic insomnia could be independent of poor PSG sleep relative to controls (Rosa and Bonnet, 2000). Using a “normal” PSG to tell a person they do not have insomnia is a misuse of the instrument. Using a single-night PSG to grade CBT-I is also a misuse: first-night effects, the reverse first-night effect, and night-to-night variability are large relative to the minute-level changes many drugs advertise.

17 Severity scales

The Insomnia Severity Index is the usual trial primary: seven items, a documented cut for clinical insomnia, sensitivity to change (Bastien, Vallières, and Morin, 2001; Morin, Belleville, Bélanger, and Ivers, 2011). The Pittsburgh Sleep Quality Index is a broader sleep-quality instrument, not an insomnia-disorder scale (Buysse, Reynolds, Monk, Berman, and Kupfer, 1989). Response on the ISI is the most comparable number across CBT-I, digital CBT-I, and drug trials. It is still a questionnaire. A six-point ISI fall can coexist with a ten-minute PSG change.

18 Differential: apnea, restless legs, circadian disorders

STOP-Bang screens apnea risk (Chung et al., 2008). RLS is diagnosed awake, by history (Allen et al., 2014). Circadian disorders are diagnosed by timing, sleep logs, and sometimes actigraphy or melatonin markers (Sack et al., 2007; Auger et al., 2015). None of those instruments is an insomnia severity scale. A person can have two diagnoses. CBT-I still has a file in comorbid apnea once the airway is addressed; it does not treat the airway.

19 CBT-I as a package

Cognitive behavioural therapy for insomnia is a multicomponent treatment: stimulus control, sleep restriction or compression, cognitive therapy, relaxation or wind-down, and a small sleep-hygiene module that is not the treatment. The 2005 NIH State-of-the-Science conference already judged CBT-I effective (National Institutes of Health, 2005). Morgenthaler, Kramer, Alessi, Friedman, Boehlecke, Brown, and colleagues’ AASM practice parameters recommended psychological and behavioural treatments (Morgenthaler et al., 2006). Qaseem, Kansagara, Forciea, Cooke, and Denberg’s American College of Physicians guideline made CBT-I the first-line treatment for chronic insomnia disorder in adults (Qaseem et al., 2016). Edinger, Arnedt, Bertisch, Carney, Harrington, Lichstein, and colleagues’ 2021 AASM clinical practice guideline gave a strong recommendation for multicomponent CBT-I (Edinger et al., 2021). Riemann, Baglioni, Bassetti, Bjorvatn, Dolenc Groselj, Ellis, and colleagues’ European guideline, updated in 2023, is aligned: CBT-I first (Riemann et al., 2017; Riemann et al., 2023). Wilson, Anderson, Baldwin, Dijk, Espie, Espie, Gringras, and colleagues’ BAP consensus sits in the same hierarchy (Wilson et al., 2019).

Trauer, Qian, Doyle, Rajaratnam, and Cunnington’s meta-analysis of CBT-I for chronic insomnia found clinically material diary improvements: sleep-onset latency and wake after sleep onset each fell by on the order of 20 minutes, sleep efficiency rose, and total sleep time moved much less (Trauer et al., 2015). Van Straten, van der Zweerde, Kleiboer, Cuijpers, Morin, and Lancee’s broader meta-analysis of cognitive and behavioural therapies reached the same directional conclusion (van Straten et al., 2018). Mitchell, Gehrman, Perlis, and Umscheid’s comparative-effectiveness review is the matching U.S. evidence report (Mitchell et al., 2012). Brasure, Fuchs, Olson, Beilfuss, Vandam, Greer, and colleagues supplied the AHRQ psychological-and-behavioural evidence review behind the ACP guideline (Brasure et al., 2016).

Head-to-head against hypnotics, Jacobs, Pace-Schott, Stickgold, and Otto found CBT comparable or superior to pharmacotherapy on several sleep endpoints, with better durability (Jacobs et al., 2004). Morin, Vallières, Guay, Ivers, Savard, Mérette, Bastien, and Baillargeon randomised persistent insomnia to CBT, zolpidem, or both, then withdrew or continued medication; CBT was the more durable arm once the drug was gone (Morin et al., 2009). Edinger, Wohlgemuth, Radtke, Marsh, and Quillian’s earlier RCT of CBT for chronic primary insomnia is a load-bearing efficacy paper (Edinger et al., 2001). Buysse, Germain, Moul, Franzen, Brar, Fletcher, and colleagues showed that a brief behavioural treatment (BBTI) can move insomnia in older adults without a full CBT package (Buysse et al., 2011).

Grade for multicomponent CBT-I on the ISI and on diary SOL/WASO/efficiency in chronic insomnia disorder: STRONG. Grade for large increases in PSG total sleep time: LIMITED. Grade for sleep hygiene education as a stand-in for CBT-I: NOT SUPPORTED.

20 Stimulus control

Stimulus control is Bootzin’s operant prescription: bed for sleep (and sex), leave the bed if not asleep, fixed rise time, no clock-watching as a sport. It is a component with its own trials and a standard piece of the package (Bootzin and Perlis, 1992; Morgenthaler et al., 2006; Edinger et al., 2021). Grade as a component: STRONG. Grade as a complete replacement for restriction plus cognitive work in severe chronic insomnia: LIMITED.

21 Sleep restriction and compression

Spielman, Saskin, and Thorpy restricted time in bed toward estimated sleep time and then titrated upward (Spielman, Saskin, and Thorpy, 1987). Sleep compression, used more in older adults, approaches the same efficiency target from a gentler slope (Lichstein, Riedel, Wilson, Lester, and Aguillard, 2001). Kyle, Morgan, Spiegelhalder, and Espie documented that restriction is effective and that the first weeks can increase sleepiness and require monitoring (Kyle et al., 2011; Kyle et al., 2014). Miller and colleagues reviewed the component evidence (Miller et al., 2014). Maurer’s experimental series is the mechanistic file (Maurer et al., 2018). Grade for restriction/compression as a CBT-I engine: STRONG. Grade as an unsupervised consumer instruction: NOT SUPPORTED — the trials used titration and exclusion of people for whom sleepiness is dangerous.

22 Cognitive therapy

Harvey’s open trial of cognitive therapy for chronic insomnia treated beliefs, attention, and safety behaviours as the independent variable (Harvey et al., 2007). Edinger and colleagues showed that CBT alters dysfunctional beliefs about sleep alongside the diary (Edinger et al., 2001). Cognitive work is moderately supported as a component. It is the piece digital programmes most often thin out, which is one reason digital effect sizes sit a step below the best face-to-face packages.

23 Relaxation

Progressive muscle relaxation and related methods have a long, smaller file. Lichstein’s work nested relaxation with compression in older adults (Lichstein et al., 2001). Relaxation is moderately supported as an adjunct, especially when somatic hyperarousal is prominent. It is not a substitute for stimulus control and restriction. Ong’s mindfulness RCT is a neighbouring, not identical, intervention (Ong et al., 2014).

24 Digital CBT-I

Zachariae, Lyby, Ritterband, and O’Toole’s systematic review found internet-delivered CBT-I effective, with effect sizes smaller than, but in the same direction as, face-to-face therapy (Zachariae et al., 2016). Ritterband, Thorndike, Gonder-Frederick, Magee, Bailey, Saylor, and Morin’s SHUTi programme, including a later randomised trial, is a load-bearing digital efficacy file (Ritterband et al., 2009; Ritterband et al., 2017). Espie’s placebo-controlled online CBT-I trial and later digital-health papers showed ISI and diary movement against a credible control (Espie et al., 2012; Espie et al., 2019). Christensen, Batterham, Gosling, Ritterband, Griffiths, Thorndike, and colleagues tested an online programme against a control in a community sample (Christensen et al., 2016). Freeman, Sheaves, Goodwin, Yu, Nickless, Harrison, and colleagues’ OASIS trial in university students found that digital CBT-I improved insomnia and reduced psychiatric symptoms that were not the primary insomnia endpoint (Freeman et al., 2017). Vedaa, Hagatun, Kallestad, Pallesen, Smith, Thorndike, Ritterband, and Sivertsen, and Soh, Ho, Sommers, and Ho’s meta-analysis, extended the digital file (Vedaa et al., 2020; Soh et al., 2020). Luik, Kyle, and Espie reviewed the implementation problem: access versus fidelity (Luik et al., 2017). Van der Zweerde, Lancee, Ida Luik, and van Straten’s internet-delivered work sits in the same family (van der Zweerde et al., 2020).

Grade for digital CBT-I on ISI and diary endpoints versus wait-list or sleep-hygiene controls: STRONG. Grade as fully equivalent to the best in-person CBT-I in severe, comorbid, or older-adult insomnia: MODERATE. Grade for digital CBT-I as a population public-health tool: MODERATE, with the OASIS psychiatric spillover as a separate, promising finding that still needs replication in clinic populations.

25 Sleep hygiene and lifestyle

Stepanski and Wyatt’s review is the necessary cold water: sleep-hygiene education is widely given and weakly supported as monotherapy (Stepanski and Wyatt, 2003). Irish, Kline, Gunn, Buysse, and Hall’s later review of empirical sleep-hygiene evidence reached the same structural conclusion: individual behaviours (caffeine, alcohol, timing, environment) have physiology; the bundled “hygiene sheet” as treatment does not (Irish et al., 2015). Chung, Lee, Yeung, Chan, Chung, and Lin’s work on sleep-hygiene education as an active comparator is one reason many CBT-I trials look conservative: the control already received the leaflet (Chung et al., 2018).

Lifestyle findings that are real and still not CBT-I: Drake, Roehrs, Shambroom, and Roth showed that caffeine taken even six hours before bed reduced total sleep time (Drake et al., 2013). Ebrahim, Shapiro, Williams, and Fenwick reviewed alcohol’s suppression of REM and later-night fragmentation (Ebrahim et al., 2013). Kredlow, Capozzoli, Hearon, Calkins, and Otto’s meta-analysis found acute and chronic exercise effects on sleep that are small-to-moderate and not insomnia-specific (Kredlow et al., 2015). Reid, Baron, Lu, Naylor, Wolfe, and Zee, and Passos, Poyares, Santana, D’Aurea, Youngstedt, Tufik, and de Mello, tested aerobic exercise in insomnia samples and reported self-reported improvements (Reid et al., 2010; Passos et al., 2011). Grade for exercise as an adjunct: LIMITED to MODERATE. Grade for sleep hygiene alone as adequate treatment of chronic insomnia disorder: NOT SUPPORTED.

26 Melatonin and melatonin-receptor agonists

Ferracioli-Oda, Qawasmi, and Bloch’s meta-analysis of melatonin for primary sleep disorders found statistically significant but small changes in sleep-onset latency and total sleep time (Ferracioli-Oda et al., 2013). Auld, Maschauer, Morrison, Skene, and Riha, and Brzezinski, Vangel, Wurtman, Norrie, Zhdanova, Ben-Shushan, and Ford, sit in the same small-effect neighbourhood (Auld et al., 2017; Brzezinski et al., 2005). Buscemi, Vandermeer, Hooton, Pandya, Tjosvold, Hartling, and colleagues’ earlier evidence report was already cautious on magnitude and on product heterogeneity (Buscemi et al., 2005). Sateia, Buysse, Krystal, Neubauer, and Heald’s AASM pharmacologic guideline did not recommend melatonin for sleep-onset or sleep-maintenance insomnia in adults (Sateia et al., 2017). That is a recommendation about insomnia disorder, not a verdict on melatonin in delayed sleep–wake phase disorder, where the circadian file is different (Auger et al., 2015).

Ramelteon, a melatonin-receptor agonist, appears in labelled onset trials in the reviewed record store; those papers are industry-sponsored sleep-onset studies, not a CBT-I comparator. Grade for exogenous melatonin as a clinically large treatment of chronic insomnia disorder: LIMITED (statistically detectable, small). Grade as a circadian-phase tool in a diagnosed circadian disorder: a different article.

27 Dual orexin-receptor antagonists

Orexin (hypocretin) neurons stabilise wake (Scammell, Arrigoni, and Lipton, 2017). Dual orexin-receptor antagonists (DORAs) were built to turn that stabilizer down rather than to agonise GABA. Herring, Connor, Ivgy-May, Snyder, Liu, Snavely, and colleagues’ suvorexant programme, and Michelson, Snyder, Paradis, Chengan-Liu, Snavely, Hutzelmann, and colleagues’ safety-efficacy papers, showed PSG and subjective improvements versus placebo over months, with next-day somnolence as the leading adverse effect (Herring et al., 2016; Michelson et al., 2014). Rosenberg, Amato, Jochelson, Kumar, Sen, and colleagues compared lemborexant with placebo and with zolpidem extended-release (Rosenberg et al., 2019). Mignot, Mayleben, Fietze, Léger, Zammit, Bassetti, and colleagues’ daridorexant trials reported improvements in sleep and in daytime function endpoints that older hypnotic programmes often omitted (Mignot et al., 2022).

Grade for DORAs versus placebo on SOL/WASO in chronic insomnia over weeks to a few months: MODERATE to STRONG. Grade for superiority on long-term disease modification or on unlearning conditioned arousal: NOT SUPPORTED. Grade relative to CBT-I durability: the DORA file is a drug-on-board file. It has not shown that stopping the drug leaves a CBT-I-like residue.

28 Z-drugs

Zolpidem, zaleplon, zopiclone, and eszopiclone are GABA-A receptor modulators, not benzodiazepines by structure, benzodiazepine-adjacent by mechanism. Walsh, Krystal, Amato, Rubens, Caron, Wessel, Schaefer, and Canafax, and Krystal, Walsh, Laska, Caron, Amato, Wessel, and Roth, showed that eszopiclone could be taken nightly for months with maintained subjective benefit in those protocols (Walsh et al., 2007; Krystal et al., 2003). Wilt, MacDonald, Brasure, Olson, Carlyle, Fuchs, and colleagues’ ACP evidence report judged pharmacologic benefits small and harms incompletely characterised (Wilt et al., 2016). Gunja reviewed clinical and forensic toxicology of Z-drugs, including complex sleep behaviours (Gunja, 2013). In 2019 the U.S. FDA added a boxed warning for rare but serious injuries from complex sleep behaviours with zolpidem, eszopiclone, and zaleplon (U.S. Food and Drug Administration, 2019). Residual driving impairment is a class concern (Verster, Veldhuijzen, Patat, Olivier, and Volkerts, 2006).

Grade for short-term reduction of sleep-onset latency versus placebo: MODERATE to STRONG. Grade for a favourable long-term benefit–harm ratio in older adults: NOT SUPPORTED as a class conclusion (see Glass et al., 2005, in the next section). Grade as disease-modifying: NOT SUPPORTED.

29 Benzodiazepines

Benzodiazepine hypnotics reduce SOL and increase scored sleep. They also produce tolerance, dependence, withdrawal insomnia, falls, and cognitive adverse effects, especially in older adults. Glass, Lanctôt, Herrmann, Sproule, and Busto’s meta-analysis in older people with insomnia found that improvements in sleep quality came with a high rate of adverse events (Glass et al., 2005). Soyka’s review of benzodiazepine dependence and Lader’s reappraisal are the harm-mechanism papers (Soyka, 2017; Lader, 2011). The 2023 AGS Beers Criteria continue to list benzodiazepines as potentially inappropriate in older adults (American Geriatrics Society Beers Criteria Update Expert Panel, 2023). Kripke, Langer, and Kline’s observational association of hypnotics with mortality is widely cited and heavily confounded; it is not a randomised demonstration of lethality, and it is not dismissible as nothing (Kripke et al., 2012). Weich, Pearce, Croft, Singh, Crome, Bashford, and Majeed’s cohort of anxiolytic and hypnotic prescriptions and mortality is the matching U.K. observational caution (Weich et al., 2014).

Grade for short-term hypnotic effect: STRONG. Grade for chronic insomnia as a preferred long-term strategy: NOT SUPPORTED. This is an evidence grade, not a prescribing instruction.

30 Sedating antidepressants

Low-dose doxepin has a dedicated insomnia programme at milligram amounts far below antidepressant dosing, with maintenance-sleep effects in older adults (Krystal et al., 2010; Roth et al., 2007). Trazodone is widely used and thinly evidenced for insomnia; Mendelson’s review found the insomnia file small relative to the prescribing habit (Mendelson, 2005). Everitt, Baldwin, Gregory, van Marwijk, Pincus, and Hunot’s Cochrane review of antidepressants for insomnia in adults was not a licence for routine off-label use (Everitt et al., 2018). Sateia and colleagues did not recommend trazodone for sleep-onset or sleep-maintenance insomnia (Sateia et al., 2017). Grade for low-dose doxepin on maintenance endpoints: MODERATE. Grade for trazodone as an evidence-based insomnia treatment: LIMITED. Grade for using a full-dose antidepressant as a sleeping pill in a person who does not have a mood-disorder indication: NOT SUPPORTED by the insomnia-trial file.

31 Antihistamines

First-generation H1-antihistamines (diphenhydramine, doxylamine) produce sedation and next-day impairment. Richardson, Roehrs, Rosenthal, Koshorek, and Roth showed tolerance to the daytime sedative effect of diphenhydramine over a few days (Richardson et al., 2002). Church and Church’s H1-antihistamine review is the receptor-level caution: these are not selective sleep drugs (Church and Church, 2013). Sateia and colleagues recommended against diphenhydramine for insomnia (Sateia et al., 2017). Grade as a short-night over-the-counter effect: LIMITED and time-limited. Grade as treatment of chronic insomnia disorder: NOT SUPPORTED.

32 Placebo, expectancy, and overmedicalization

Yeung, Sharpe, Glozier, Hackett, and Colagiuri’s systematic review found a reliable placebo effect on perceived sleep onset when a placebo believed to be active was compared with no treatment (Yeung et al., 2018). That does not make licensed hypnotics fictions. It does mean that a two-week open improvement after a new bottle is not a specific pharmacodynamic demonstration. Moloney, Konrad, and Zimmer described the medicalization of sleeplessness: a widening of the indication, a narrowing of the behavioural first step, and a cultural preference for a pill (Moloney et al., 2011). The CBT-I guidelines are, among other things, an official attempt to reverse that funnel (Qaseem et al., 2016; Edinger et al., 2021; Riemann et al., 2023).

33 CBT-I evidence matrix

34 Drug-class comparison

Amounts in source trials are experimental parameters. They are not instructions.

35 Durability and relapse

36 Harms

37 Digital intervention matrix

38 Sleep-outcome measurement table

39 Is sleep hygiene adequate treatment?

No, not for chronic insomnia disorder. Stepanski and Wyatt (2003) and Irish and colleagues (2015) are explicit: hygiene is a set of plausible behaviours, not a validated monotherapy. CBT-I trials that use hygiene as a control often still beat it (Chung et al., 2018; Trauer et al., 2015). Caffeine timing, alcohol, and exercise have their own small files (Drake et al., 2013; Ebrahim et al., 2013; Kredlow et al., 2015). Those files do not add up to CBT-I. Grade: NOT SUPPORTED as adequate treatment.

40 Are objective sleep changes smaller than subjective improvements?

Often, yes. Diary SOL and WASO and the ISI move more, and more consistently, than PSG total sleep time, especially under CBT-I (Trauer et al., 2015; Edinger and Fins, 1995; Harvey and Tang, 2012). Rosa and Bonnet (2000) showed that the complaint can be loosely coupled to scored sleep. Drugs can look better on PSG because PSG is what many drug programmes paid to measure. That is an endpoint choice, not a demonstration that the person’s problem was never the complaint. The honest reporting rule is two columns, not one converted number.

41 How durable are CBT-I effects?

More durable than hypnotic effects after the hypnotic stops. Trauer et al. (2015), van Straten et al. (2018), Jacobs et al. (2004), and Morin et al. (2009) are the backbone: behavioural gains persist at follow-up in a way that medication gains generally do not once medication is withdrawn. Digital programmes show a similar pattern when engagement holds (Ritterband et al., 2017; Espie et al., 2019). Durability is not immortality. Skills can be abandoned. Relapse of the complaint after a new precipitant is expected under the 3P model and is a reason for booster sessions in some protocols, not a reason to declare CBT-I a failure.

42 Are medication benefits clinically meaningful relative to risks?

Sometimes, in the short term, for selected endpoints, in people who have already failed or cannot access CBT-I — which is a clinical judgement this article does not make for any reader. Wilt et al. (2016) and Qaseem et al. (2016) judged pharmacologic benefits small relative to uncertainty about harms. Glass et al. (2005) is the older-adult brake. The FDA boxed warning (2019) is the Z-drug brake. Beers (2023) is the benzodiazepine brake. DORAs improve the on-drug picture and do not erase next-day somnolence. Placebo increments are material (Yeung et al., 2018). A statistically significant SOL change of a few minutes can be real and still fail a patient’s idea of a cure. Meaningfulness is endpoint-specific. It is not a class slogan.

43 Is insomnia sometimes overmedicalized?

Yes. Moloney et al. (2011) described the widening of sleeplessness into a drug-shaped problem. Symptom prevalence far exceeds disorder prevalence (Ohayon, 2002; Morin and Jarrin, 2022). That gap is where leaflets, late-night advertising, and hypnotic refills live. Naming overmedicalization does not deny that chronic insomnia disorder is disabling, associated with depression risk (Baglioni et al., 2011; Hertenstein et al., 2019), and worth treating with the intervention that has the durability record. It denies that every bad night is a benzodiazepine deficiency.

44 How much benefit derives from expectancy and placebo?

Enough that every hypnotic claim must be placebo-adjusted, and enough that an open bottle is a biased experiment. Yeung et al. (2018) is the quantitative caution. CBT-I trials have their own expectancy: a demanding programme generates buy-in. The better digital trials used placebo or education controls (Espie et al., 2012) rather than wait-lists alone. Residual specific effects remain in those designs. They are smaller than marketing implies and larger than a pure placebo story.

The six adversarial questions above are resolved on the papers, not on any external forecast.

Standing constraint This document describes published research. It is not medical advice. It does not recommend a treatment, product, dose, route, or schedule for any person. Reported session counts, milligrams, and time-in-bed rules are study parameters attached to named populations. They are not instructions.

45 How this document was assembled

This is a South Beach Longevity science article, not a Radix peptide title. Citations were verified against NCBI records; general reference databases were queried only to locate literature and are not cited as authority, and open guideline and regulator pages were consulted directly. Binding is title-checked; no identifier was assigned from memory.

46 Evidence handling

Study type is named in the sentence. Animal and receptor papers never imply a human outcome. Conflicting evidence is left in conflict: Kripke’s mortality signal is reported as observational and confounded; Glass is reported as a randomised-harm meta-analysis in older adults; DORA daytime-function claims are reported as labelled endpoints, not as CBT-I equivalents. Subjective and objective columns are not converted into each other. Sleep duration epidemiology is not insomnia epidemiology. Circadian melatonin evidence is not insomnia-disorder melatonin evidence.

47 Unresolved science

Whether treating insomnia with CBT-I reduces incident depression at population scale is still a prediction from Baglioni and Hertenstein, only partly tested in treatment samples (Manber et al., 2008; Freeman et al., 2017). Whether the objective-short-sleep phenotype needs a different treatment algorithm remains a research programme, not a clinic rule (Vgontzas et al., 2013). Whether DORAs have a residual after discontinuation that resembles CBT-I is unshown. Whether digital CBT-I can be delivered at fidelity outside trial platforms is an implementation question (Luik et al., 2017). A consensus on what “clinically meaningful” means when ISI moves and PSG does not is still a debate, not a conversion factor.

References