Trazodone

Trazodone is a triazolopyridine antidepressant, the first of the serotonin antagonist and reuptake inhibitors (SARIs) [2]. It was approved in the United States in 1981 and is licensed only for major depression in adults [1], yet by 2012 insomnia was the most common reason it was prescribed there, and it was the second most prescribed treatment for insomnia [2].

Insomnia. The trials are small and short. A 2024 meta-analysis of 44 randomized trials against placebo that reported sleep outcomes (3,935 participants) found that trazodone did not change how long people thought they slept but improved rated sleep quality (standardized mean difference −0.58), cut awakenings and time awake after sleep onset by about 13 minutes, and added about 28 minutes of sleep and 3 percentage points of sleep efficiency on polysomnography, with minor effects on sleep latency [4]. An earlier meta-analysis of seven placebo-controlled trials found better perceived sleep quality and fewer awakenings but no change in sleep efficiency, latency or total sleep time [8], and a Cochrane review judged that there may be a small short-term improvement, from low-quality evidence [9]. The American Academy of Sleep Medicine found only one eligible trial: 50 mg for 14 nights shortened self-reported sleep latency by about 10 minutes and added about 22 minutes of sleep, neither clinically significant, and it suggested that clinicians not use trazodone for insomnia [5]. A 2022 network meta-analysis of 154 trials counted trazodone among drugs that can work in the short term but have poor tolerability or no long-term data; it found no long-term trazodone trials [10]. In 16 people with primary insomnia, 50 mg a night reduced awakenings and light sleep and raised deep sleep after a week, without shortening sleep onset [11].

Particular groups. In 30 people with Alzheimer's disease and sleep problems, 50 mg at 10 pm added about 42 minutes of night-time sleep over two weeks without daytime napping or any change in cognition [12]; the Cochrane review of sleep drugs in dementia rated this low-certainty evidence from a single trial [13]. After alcohol detoxification, 50–150 mg improved sleep quality over three months but people taking it improved less in days abstinent and drank more per drinking day once it was stopped [14]. In people on methadone it did not improve sleep [15], whereas in men on buprenorphine 82% reached a normal sleep-quality score against 16% on placebo [16]. In depression with insomnia, a 2026 meta-analysis of ten trials found large improvements in sleep quality and smaller ones in depressive symptoms, with much more somnolence and blurred vision than placebo [17].

Depression. In 412 adults with major depression, the once-daily form at a mean 310 mg lowered Hamilton scores by 11.4 points against 9.3 on placebo, a difference present from the first week [6]. Against venlafaxine XR in 324 adults it acted faster at day 7 but was slightly less effective at week 8 (1.6 points); the difference was not significant in those who completed the protocol [18]. Both trials were run by the makers of the once-daily form [6][18]. In the 522-trial network meta-analysis of 21 antidepressants every drug beat placebo, but in head-to-head trials trazodone was among the least effective and had among the highest dropout rates [7].

Cognition and dementia. Trazodone came to dementia research from a screen of 1,040 licensed drugs: it reversed the shutdown of protein synthesis caused by the eIF2α-P branch of the unfolded protein response, and at 40 mg/kg a day it prevented neurological disease in 12 of 15 prion-infected mice and reduced hippocampal atrophy and tau pathology in a tauopathy model [19]. In people the evidence is observational. Among 50 older adults with sleep complaints matched for diagnosis and baseline scores, 25 regular users (median 50 mg at bedtime) lost 0.27 Mini-Mental State points a year against 0.70 in non-users over about four years; the difference was not significant among people also taking cholinesterase inhibitors, and the secondary cognitive tests were not significant after correction [20]. A 2021 systematic review of 16 human studies found no effect on cognition in seven, benefit in five and impairment in four, with acute use tending to impair and long-term use possibly protecting, and no clinical evidence that it treats neurodegenerative disease [21]. A crossover trial of 50 mg in 100 older adults with amnestic mild cognitive impairment (REST) published its design in 2024 [22].

Other uses. In 26 people with frontotemporal dementia, trazodone up to 300 mg a day lowered Neuropsychiatric Inventory scores, with 10 improving by more than half, and did not change cognition [21][23]; a Cochrane review combining that trial with one in Alzheimer's disease found no statistically significant benefit on behavior [24]. For erectile dysfunction, six small trials showed a non-significant advantage over placebo (37% against 20% responding), possibly larger at 150–200 mg and in psychogenic cases [25]. In obstructive sleep apnea, a single 100 mg dose let people tolerate more carbon dioxide before waking [26], and it did not change the apnea–hypopnea index in pooled trials [27].

A dose-dependent mix of receptors. Trazodone binds most tightly to 5-HT2A serotonin receptors and to alpha-1 adrenergic receptors, with Ki values of 14–27 nM at 5-HT2A, α1B and α1D, and much more weakly to the histamine H1 receptor (190 nM) and the serotonin transporter (280 nM) [3]. The prescribing information gives a similar order: 5-HT2A 35.6 nM, 5-HT2B 78.4 nM, 5-HT1A 118 nM (as a partial agonist), α1A 153 nM, α2C 155 nM, 5-HT2C 224 nM and the transporter 367 nM [1]. Because the gap between targets is wide, which ones are engaged depends on dose. A 2018 occupancy model concluded that about 50 mg a day should block the receptors behind the hypnotic effect, while antidepressant effects need higher doses [28]; a 2022 physiologically based model predicted that a single 30 mg dose occupies over 90% of 5-HT2A and α1B receptors but only 43% of H1 receptors and 34% of transporters, rising to 75% and 68% at 300 mg a day of the once-daily form [3]. These are model estimates, not imaging measurements.

Sleep. Reviewers attribute the hypnotic effect to blocking 5-HT2A, H1 and alpha-1 receptors together [2][10]. It deepens sleep: it increased slow-wave sleep in eight healthy men, whereas imipramine suppressed REM sleep [29], 50 mg increased slow-wave sleep in people with insomnia after a week [11], and pooled polysomnography trials showed more stage N3 sleep and less stage N1 sleep, with no change in REM [27].

Depression. The prescribing information describes trazodone as both a serotonin reuptake inhibitor and a 5-HT2 antagonist and states that how these produce its antidepressant effect is unknown [1]. The usual rationale is that blocking 5-HT2A and 5-HT2C while inhibiting reuptake avoids the insomnia, anxiety and sexual side effects of SSRIs [2].

Unfolded protein response. In cells and mice, trazodone acts downstream of eIF2α phosphorylation: it lowered ATF4 and restored protein synthesis in diseased hippocampus without changing eIF2α-P, and it did not reduce the misfolded prion protein itself, so the protection seen was against the cell's stress response rather than the disease protein [19]. Whether this happens at human doses has not been tested.

Metabolite. CYP3A4 converts trazodone to m-chlorophenylpiperazine (mCPP), which is pharmacologically active [1][30].

Immediate release. Tablets peak about an hour after a fasting dose, or two hours with food, and food increases the amount absorbed [1]. Their high, early peak may contribute to drowsiness and hypotension in the first week, which led to the development of prolonged-release forms [2].

Once-daily extended release. At steady state, 300 mg once daily gave the same total exposure as 100 mg three times a day with a 43% lower peak; a single 300 mg dose gave about 20% less exposure and a 60% lower peak than three 100 mg doses eight hours apart; and a high-fat meal raised the peak by 86% without changing total exposure [32]. The manufacturer's data cited in the 412-patient trial gave a similar 42% reduction in peak level [6].

Body size and age. Its volume of distribution rises with body weight, so its half-life in obese adults was more than twice that in normal-weight adults, and older men cleared it more slowly than younger men [33].

Common effects. In the placebo-controlled trials summarized in the prescribing information, drowsiness affected 41% of outpatients on trazodone against 20% on placebo, dizziness or light-headedness 28% against 15%, and dry mouth 34% against 20% [1]. In the 412-patient depression trial 12.4% stopped because of side effects against 2.9% on placebo, most often for dizziness or sedation [6]. In sleep trials, dropouts for side effects were 2.3 times as common as on placebo [4], and in the AASM's single trial 30% on 50 mg had headache (19% on placebo) and 23% somnolence (8%) [5].

Next-day function and falls. After a week of 50 mg, people with insomnia showed small but measurable deficits in short-term memory, verbal learning, balance and arm endurance the next morning, though not in simulated driving; only the short-term memory deficit passed the trial's correction for multiple outcomes, and the balance effect was about a third of that of a 0.05% breath alcohol level [11]. In Ontario nursing homes, residents newly given low-dose trazodone had the same rate of fall-related injury over 90 days as matched residents newly given a benzodiazepine (5.7% against 6.0%) [35].

Heart and blood pressure. Trazodone blocks the hERG potassium channel (IC50 0.69 µM) and lengthened the QT interval in isolated rabbit hearts at clinically relevant concentrations [36]. A single 100 mg dose lowered blood pressure and prolonged the corrected QT interval in healthy volunteers [31]. The prescribing information warns of arrhythmias, including torsade de pointes reported at 100 mg or less, advises avoiding it in people with QT prolongation or recent heart attack, and warns of orthostatic hypotension and fainting [1].

Priapism and other warnings. Prolonged painful erections have been reported and can cause permanent damage; an erection lasting more than four hours needs emergency treatment [1]. Like other antidepressants it carries a boxed warning about suicidal thoughts in people under 25, and the label also lists serotonin syndrome, bleeding with anticoagulants, mania in bipolar disorder, low sodium, angle-closure glaucoma and discontinuation symptoms [1]. Pregnancy studies have not linked it to birth defects, but they are small [1].

Overdose. Among 206 overdose exposures reported to US poison centers and drug-abuse surveillance by 1986 there were no deaths, against 16 deaths among 2,263 tricyclic and tetracyclic overdoses; the nine deaths reported to the manufacturer all involved other drugs or alcohol [37]. The prescribing information lists priapism, respiratory arrest, seizures and QT prolongation as the most severe effects of overdose alone [1].

Alcohol recovery. In the post-detoxification trial, trazodone improved sleep while it was taken but was associated with fewer days abstinent and, after it was stopped, more drinks per drinking day [14].

Trazodone must not be combined with MAO inhibitors, including linezolid and intravenous methylene blue, or started within 14 days of one, because of serotonin syndrome; triptans, other antidepressants, lithium, tramadol, fentanyl, tryptophan, buspirone and St John's wort add to that risk [1]. Strong CYP3A4 inhibitors such as ketoconazole, itraconazole, clarithromycin and ritonavir raise its levels, and inducers such as rifampin, carbamazepine and phenytoin lower them [1][30]. It adds to the effects of alcohol and other sedatives, can raise digoxin and phenytoin levels, adds to bleeding risk with anticoagulants and antiplatelet drugs, should not be combined with other QT-prolonging drugs, and may require a lower dose of blood-pressure medicines [1].