Doxepin

Doxepin is a dibenzoxepin tricyclic antidepressant, sold as a mixture of E and Z isomers in a ratio of about 85:15 [2][6]. It was first approved in the United States in 1969, and its capsule label now covers major depression in adults at 75–150 mg a day, up to 300 mg [1]. A low-dose tablet (Silenor) is licensed there for insomnia characterized by difficulty staying asleep, at 6 mg (3 mg to start at 65 and over) [2]; a 2018 Cochrane review noted that it was not licensed for insomnia in Europe [3]. A 5% cream is licensed for up to 8 days of itch in atopic dermatitis and lichen simplex chronicus [7].

Insomnia at 1–6 mg. The trials behind the low-dose product were funded by its developer, Somaxon [8][9]. In 240 adults aged 65 and over, 3 mg nightly for 12 weeks cut time awake after sleep onset from about 109 minutes on placebo to 75 on the first night and 76 at week 12, added 30–43 minutes of sleep, and improved sleep in the last quarter of the night; it did not shorten measured time to fall asleep, though people reported falling asleep faster [9]. In 221 adults under 65 given 3 or 6 mg for 35 nights, average wake time after sleep onset was 62.5 minutes on placebo, 44.4 on 3 mg and 39.5 on 6 mg (effect sizes 0.58 and 0.72), and measured sleep onset was faster on the first night only [10]. Two-night crossover trials of 1, 3 and 6 mg in 67 younger and 76 older adults showed the same pattern [8][11], a 4-week outpatient trial of 6 mg in 254 older adults improved reported sleep time but not reported time to fall asleep [12], and in 565 healthy adults put through a model of transient insomnia, a single 6 mg dose shortened sleep onset by 13 minutes and added 51 minutes of sleep [13]. A 2025 pooled reanalysis of the two longer 3 mg trials found sleep onset 6.4 minutes faster on the first night and no faster after 2 or 4 weeks, below the 10-minute threshold the AASM treats as clinically significant [14].

Pooled evidence. The AASM guideline pooled four trials of 3 mg and three of 6 mg: time awake after sleep onset fell by about 22 and 23 minutes and sleep efficiency rose by about 7 percentage points, while gains in sleep onset fell below clinical significance; it rated the overall evidence very low because of industry sponsorship, imprecision and possible publication bias, and made a weak recommendation for doxepin in sleep-maintenance insomnia [4]. A Cochrane review pooling four doxepin trials found a moderate gain in self-rated sleep quality (standardized mean difference −0.39), about 23 more minutes of sleep and little or no change in sleep onset, and no evidence on long-term use [3]. A 2015 review of nine trials reached the same split between sleep maintenance and sleep onset and noted that all the low-dose trials were industry-sponsored [15]. A 2017 meta-analysis of nine polysomnography trials of tricyclics for insomnia, most using very low doses of doxepin, found more total sleep (standardized mean difference 0.61) and more somnolence (6.1% against 3.2%) than placebo, with no more dropouts [16]. In a network meta-analysis of hypnotics in older adults, low-dose doxepin ranked first for reported total sleep time (about 28 minutes over placebo) and measured sleep efficiency [17]; a larger 2022 network meta-analysis of 154 trials in adults found it well tolerated but judged its efficacy data too scarce for firm conclusions [18].

Higher sleep doses and particular groups. Before the low-dose product, 25–50 mg for 4 weeks improved sleep efficiency, rated sleep quality and working ability in 47 adults with primary insomnia, but severe rebound insomnia was more common in the doxepin group in the first nights after stopping [19]. In a three-arm pilot of 18 people with Parkinson's disease, 10 mg a day for 6 weeks improved insomnia severity, fatigue and Montreal Cognitive Assessment scores against placebo [20]. Against zolpidem in 120 adults over 8 weeks, 6 mg gave more sleep and less time awake, zolpidem gave faster sleep onset, and adverse events were more common on doxepin (23% against 13%) [21].

Depression and anxiety. In 579 outpatients with major depression, 150 mg nightly for 10 weeks beat placebo on depression, associated anxiety and sleep, as did dothiepin, which caused less drowsiness, weight gain and appetite increase [22]. A 1977 review judged its mood effect similar to amitriptyline's and probably smaller than imipramine's, and its sedation an advantage in depression with insomnia or anxiety [23]. In 35 mildly depressed people on methadone maintenance, doxepin improved depression more than placebo [24].

Itch, hives and mouth pain. The 5% cream relieved itch in 85% of 270 people with atopic dermatitis against 57% on vehicle cream within 7 days, mostly within the first 24 hours [25], and in a second trial of 309 people with other eczemas, 84% of those on doxepin had relief by the end of the week [26]. Taken by mouth at 25 mg three times a day, it reduced lesions, itch and angioedema in 16 adults with chronic idiopathic urticaria [27], and in a 24-patient crossover it matched the antihistamine mequitazine, both beating placebo [28]. As a 25 mg mouth rinse it reduced radiotherapy-induced mucositis pain over 4 hours in two phase 3 trials, by an amount the second trial found smaller than its threshold for a clinically important difference [29][30].

Cognition. In the low-dose trials no memory impairment or cognitive disorder was reported [9][10]. The zolpidem comparison found larger improvements on the Wisconsin Card Sorting Test with doxepin [21]. In 1-year-old rats, 1 and 5 mg/kg a day for 21 days lengthened retention in a passive-avoidance memory test while 10 mg/kg did not [31]. At antidepressant doses its anticholinergic action points the other way (see Safety).

Histamine H1 blockade. Doxepin is among the most potent H1 antagonists known. In post-mortem human brain it bound the H1 receptor with a KD of 0.24 nM, the tightest of 25 antidepressants tested [32], no antihistamine in a later panel of ten bound more tightly [33], and the low-dose label gives a Ki below 1 nM [2]. Its affinity is roughly seven times that of mepyramine, the reference H1 antagonist [8]. In 2011 the first crystal structure of the human H1 receptor was solved with doxepin bound: it sits deep in a largely hydrophobic pocket, its amine anchored to Asp107 and in direct contact with Trp428, a conserved residue involved in receptor activation. Antihistamines act as inverse agonists that reduce the receptor's basal activity, and the conserved pocket explains why first-generation antihistamines such as doxepin also bind other amine receptors [34].

Dose decides which targets are hit. At antidepressant doses of 75–300 mg a day doxepin inhibits reuptake of norepinephrine and serotonin, which is probably how it treats depression; at 1–6 mg it blocks histamine receptors with very little action elsewhere [3]. Its side effects are attributed to blocking H2, 5-HT2, alpha-1 adrenergic and muscarinic receptors as well as to reuptake inhibition [34]; its alpha-1 KD of 24 nM was the most potent of the antidepressants tested [32]. Stahl argued that the next-day sedation, tolerance and weight gain blamed on antihistamine hypnotics may come from these other actions, and that H1-selective dosing could avoid them [35]. The labels are cautious: the low-dose label calls the mechanism in sleep maintenance unclear but possibly H1 antagonism [2], and the capsule label says its antidepressant mechanism is not well understood [1].

Effects on sleep. Low-dose doxepin shortens time awake after an arousal without reducing the number of arousals, and its effect persists into the last hours of the night [8][9]; the AASM analysis likewise found no fall in the number of awakenings [4]. The authors of the 12-week trial suggest it blocks one of several parallel wake-promoting systems rather than damping the brain globally as GABAergic hypnotics do [9], and Roth and colleagues that the morning rise in histamine release overcomes the blockade, which might explain the lack of measured next-day sedation [8]. Sleep architecture changed little: more stage 2 sleep, slightly less REM as a percentage at 3–6 mg, and no change in deep sleep [8][10]. In 10 people with insomnia, 25 mg lowered evening and nocturnal cortisol along with improving sleep, which the authors read as a normalization of the stress axis [36].

Isomers and metabolite. The Z isomer and Z-nordoxepin are considered the more active antidepressants; plasma doxepin stays near the 85:15 E:Z ratio of the drug, but nordoxepin levels shift toward the Z form [6][37]. Nordoxepin is pharmacologically active and outlasts the parent drug [1].

Skin and mouth. The cream label attributes doxepin's antipruritic effect to H1 and H2 blockade without settling the mechanism, and notes that drowsiness may itself reduce awareness of itch [7]. Taken by mouth it suppressed histamine- and codeine-induced skin wheals [27]. Its anesthetic effect as a rinse may come from sodium channel blockade, and the drowsiness it caused suggests absorption through damaged mucosa [29].

Low-dose tablets. The 3 and 6 mg tablets are immediate release [2]. A high-fat meal raised total exposure by 41% and peak by 15% and delayed the peak by about 3 hours, which is why the label advises against taking it within 3 hours of a meal [2].

Capsules. Capsules of 10–100 mg are the antidepressant form; a single 75 mg dose peaks at 2–4 hours [1]. The 10–50 mg doses used in some insomnia studies fall in this range rather than in that of the licensed sleep product [19][20].

Cream. Doxepin crosses the skin: plasma levels from the 5% cream reached 47 ng/mL in eczema patients, levels that can match oral therapy, and drowsiness was more common when more than 10% of the body was treated; occlusive dressings are not to be used [7]. Serum doxepin was detectable in 9 of 24 people with atopic dermatitis after a week of four-times-daily use [38].

Mouth rinse. The rinse is a compounded preparation, swished for a minute and spat out [29][30].

At 1–6 mg. In the label's long-term trials, somnolence or sedation affected 4% on placebo, 6% on 3 mg and 9% on 6 mg, and hypertension 3% on 3 mg against none on placebo; 0.4–1.0% stopped for adverse reactions against 0.6% on placebo [2]. The published trials reported no anticholinergic effects, memory impairment, complex sleep behaviors or weight gain [9][10], and found no difference from placebo on next-morning psychomotor tests [4][8][9][10]. The label is less clean: it reports modest next-day decrements on symbol copying and sleepiness ratings after 6 mg in the transient insomnia study, and small decreases in digit-symbol and symbol-copying scores at 6 mg in the 35-day trial [2]. After 35 nights, stopping caused no rebound insomnia (1%, 1% and 4% met the criteria on placebo, 3 mg and 6 mg) or withdrawal syndrome [10], though 5% on 6 mg had nausea or vomiting in the discontinuation nights [2]. Like other hypnotics, its label warns about sleep-driving and other complex behaviors, worsening depression and suicidal thoughts, and alcohol, and advises against it in severe sleep apnea, where it has not been studied [2].

At antidepressant doses. Among 1,635 patients with depression, somnolence affected 17%, dry mouth 15%, dizziness 6%, constipation and fatigue 5% each, and blurred vision, tachycardia and hypotension 3% each [1]. The capsule label carries a boxed warning about suicidal thoughts in people under 25, warns of serotonin syndrome, angle-closure glaucoma, lowered seizure threshold and mania, contraindicates it in glaucoma and current or past urinary retention, and reports withdrawal symptoms after abrupt stopping and, after marketing, blood-cell disorders, arrhythmias and jaundice [1]. At 25–50 mg for insomnia, two of 47 patients left the trial with raised liver enzymes or low white cell and platelet counts [19].

Heart. In a thorough QT study of 206 healthy adults, neither 6 mg nor 50 mg a day for a week lengthened the QTc interval, while the positive control did [39]. In a 2024 review of observational studies, doxepin's risk of sudden death or ventricular arrhythmia in normal use was no greater than paroxetine's in one study, while some studies linked it with stroke or transient ischemic attack [5].

Overdose. The same review found the fatal toxicity of doxepin in overdose broadly similar to that of amitriptyline, clomipramine, dosulepin, imipramine and trimipramine; of the tricyclics, only lofepramine was clearly less dangerous [5]. In an animal overdose model doxepin was less arrhythmogenic than amitriptyline at similar blood levels [45], but an early review already warned that its cardiotoxicity in overdose is that of any tricyclic [23]. Overdose causes arrhythmias, severe hypotension, seizures and coma; a QRS duration of 0.10 seconds or more signals severity, and dialysis does not help [1].

Anticholinergic burden. Two large UK case-control studies linked long-term prescription of strongly anticholinergic antidepressants with a higher later risk of dementia; in the first, amitriptyline and dosulepin were the most prescribed strongly anticholinergic drugs, and its authors noted that the drugs might have been prescribed for early symptoms of dementia [46][47]. Neither reports doxepin separately, and the 1–6 mg doses caused no anticholinergic effects in trials [9][10].

Genetics. CYP2D6 poor metabolizers clear E-doxepin about three times more slowly than extensive metabolizers and carry four times as much nordoxepin [41]; the capsule label advises lower doses in CYP2D6 and CYP2C19 poor metabolizers [1], and genotype-based dosing guidance for tricyclics is considered applicable to doxepin, though with fewer data than for amitriptyline [48].

Pregnancy and breastfeeding. Epidemiological studies have not established a link with birth defects or miscarriage, but newborns exposed late in pregnancy have needed respiratory support and tube feeding [1]. Doxepin and nordoxepin reach breast milk at about plasma levels, and nordoxepin was measurable in a breastfed infant [44]; a 9-day-old breastfed by a mother taking 35 mg a day was admitted drowsy, floppy and feeding poorly, and recovered within 48 hours of stopping breastfeeding [49]. Both labels advise against breastfeeding [1][2].

Cream. In controlled trials 22% of cream users were drowsy against 2% on vehicle, 23% had burning or stinging, and 10% dry mouth; contact allergy has been confirmed by patch test, and a 2.5-year-old who had 27 g applied over three days had a seizure and fell into a coma before recovering [7].

Doxepin must not be combined with MAO inhibitors, including linezolid and intravenous methylene blue, or started within 14 days of one, because of serotonin syndrome; SSRIs, SNRIs, other tricyclics, triptans, opioids, lithium, tryptophan, buspirone and St John's wort add to that risk [1][2]. Cimetidine doubled exposure to a 6 mg dose, so the low-dose label caps the dose at 3 mg with it; sertraline raised exposure by 21% and peak by 32% and worsened psychomotor performance 2–4 hours after dosing [2]. Strong CYP2D6 inhibitors call for a lower dose, and carbamazepine cut doxepin and nordoxepin exposure by 55% [1]. Alcohol, sedating antihistamines and other central nervous system depressants add to its sedation, and one patient on tolazamide developed severe hypoglycemia after doxepin 75 mg a day was added [2].