Choosing Antidepressants When Depression Meets Insomnia: Comparative Efficacy, Sleep Effects, and Sexual-Function Trade-offs—A Narrative Review

1. Introduction: When Depression and Insomnia Coexist

Insomnia and depression are inseparable in clinical practice. Approximately 60% of adults meeting criteria for MDD complain of insomnia, and between 10% and 20% of adults endorsing current insomnia display evidence of MDD [2]. In clinical samples the figure is higher still: up to 90% of patients with MDD present with comorbid insomnia [12], and insomnia has been identified as the most common complaint associated with depression, reported by more than 90% of depressive patients in large cohort studies [13]. Sleep disturbance is among the core symptomatic domains of MDD alongside anhedonia and anxiety [14], and it is the most prominent symptom in depressive patients [1].

The epidemiological relationship is not merely associative. Longitudinal studies have established insomnia as an independent risk factor for the development of emerging or recurrent depression among young, middle-aged, and older adults, giving rise to the now widely accepted view of a bidirectional association between sleep disturbance and depression [1,15]. More recent meta-analytic work confirms that insomnia disorder increases the risk of subsequent depression and conceptualizes insomnia as a transdiagnostic process in psychopathology [16]. The clinical corollary is sobering: sleep problems do not reliably disappear when depression improves. In the STAR*D effectiveness trial, among remitters to citalopram, 29.5%, 54.9%, and 16.6% retained at least mild initial, mid-nocturnal, or early-morning insomnia, respectively, and 9.7%, 40.5%, and 6.8% had these symptoms at at least a moderate level [3]. Overall rates of residual insomnia in remitters to pharmacotherapy range from 44% to 53% across studies [3]. Residual symptoms in general — and sleep disturbance in particular — increase the probability of depressive relapse and are associated with poorer functional and psychosocial outcomes [17,18]. Whether residual insomnia specifically predicts relapse is less settled than for residual anxiety, with both positive and negative associations reported [19].

Two clinical realities follow. First, antidepressant selection in the depressed patient who cannot sleep is a decision about the whole symptom picture, not just mood: activating agents may worsen the presenting complaint, while sedating agents may resolve it. Second, residual insomnia is frequently undertreated even when it is recognized — in STAR*D, although participants were permitted adjunctive hypnotics or low-dose trazodone, only 21 of 943 remitters (2.6%) took hypnotics and 24 of 943 (2.6%) took adjunctive trazodone [3].

The decision matters against a backdrop of broadly similar efficacy. The most comprehensive network meta-analysis of antidepressants ever undertaken — 522 trials and more than 116,000 patients, comparing 21 drugs and placebo — found that while all agents were more efficacious than placebo, differences between individual antidepressants were modest [4,20]. When average efficacy is nearly interchangeable, the clinically meaningful axes of differentiation become tolerability and symptom-specific effects — and no two domains illustrate this better than sleep and sexual function.

2. Materials and Methods: Data Sources and Study Selection

This is a narrative, clinically oriented review rather than a systematic review or meta-analysis; no PRISMA protocol was registered. PubMed/MEDLINE and Google Scholar were searched from database inception through August 2026 using combinations of the terms antidepressant, insomnia, sleep, sleep architecture, polysomnography, REM sleep, sexual dysfunction, major depressive disorder, and STAR*D, cross-referenced with the names of individual agents (e.g., mirtazapine, trazodone, agomelatine, bupropion, selective serotonin reuptake inhibitor, selective serotonin-norepinephrine reuptake inhibitor). The reference lists of retrieved articles, prior systematic reviews and meta-analyses, and current clinical practice guidelines — including the 2023 European Insomnia Guideline [28] — were hand-searched for additional relevant citations. Priority was given to randomized controlled trials, network meta-analyses, systematic reviews, large effectiveness cohorts such as STAR*D [3], and major guideline documents that reported comparative antidepressant efficacy or quantified sleep or sexual-function outcomes in adults with major depressive disorder; case reports, non-English-language sources, and studies not addressing these three domains were excluded. Forty-three sources satisfying these criteria were retained and are synthesized qualitatively by drug class and individual agent below. As a narrative synthesis, study selection and weighting reflect clinical and methodological judgment rather than a fixed, reproducible search algorithm; this is discussed as a limitation in the Conclusions.

3. Sleep Architecture in Depression and How Antidepressants Modulate It

Depression is characterized by a characteristic polysomnographic signature: disturbances of sleep continuity, diminished slow-wave sleep, and altered rapid eye movement parameters, including increased REM density and reduced SWS in the first sleep cycle, which have long served as biological markers in antidepressant drug development [5]. An ideal antidepressant would reverse these abnormalities; in practice, many worsen them [5].

The most consequential distinction among antidepressants is between activating and sedating profiles. Polysomnographic studies show that SSRIs, SNRIs, and activating tricyclics increase REM latency, suppress REM sleep, and may impair sleep continuity, whereas sedating antidepressants decrease sleep latency, improve sleep efficiency, increase SWS, and usually have little or no effect on REM sleep [6]. These effects are strongest in the first few weeks of treatment but may persist in some patients, aggravating insomnia complaints or causing daytime somnolence [6]. Drugs such as bupropion and nefazodone, by contrast, lack REM-suppressant effects altogether, supporting the view that the neurochemical mechanisms regulating discrete sleep stages can be dissociated pharmacologically [21]. Understanding these signatures is clinically essential because the acute sleep effects of an antidepressant can determine adherence in the first weeks — exactly when dropout risk is highest [22].

4. Individual Agents and Their Sleep Signatures

4.1. Activating Antidepressants: SSRIs and SNRIs

For SSRIs, polysomnographic studies in both healthy volunteers and depressed patients demonstrate a decrease in sleep efficiency and total sleep time, a lengthening of sleep latency, and a deterioration in sleep continuity, including an increase in the number of awakenings and wake time during the total sleep period; sleep architecture typically shifts toward more stage 1 and stage 2 sleep, with decreases in stage 3, stage 4, and REM sleep, and a lengthening of REM latency [23]. The objective picture is partially offset subjectively: while healthy volunteers report unchanged or worsened sleep, depressed patients often report improvement, a discrepancy that complicates clinical interpretation [23]. Activating drugs — fluoxetine, paroxetine, venlafaxine, bupropion, and others — frequently require co-prescription of sleep-promoting agents, and even in maintenance treatment with activating antidepressants, as many as 30–40% of patients may still suffer from insomnia [5,6].

4.2. Mirtazapine

Mirtazapine, a noradrenergic and specific serotonergic antidepressant, has the most consistent evidence for sleep improvement among the newer agents. Polysomnographic studies in depressed patients show that during acute administration mirtazapine significantly increases total sleep time, sleep efficiency, stage 2, REM, and slow-wave sleep percentages, and decreases sleep latency and wake time — effects that persist after five weeks of treatment [24]. These objective changes are accompanied by improved subjective sleep quality, reversal of sleep markers of depression, and reduction of depressive symptoms [25]. The cost is somnolence: in clinical studies of sedating antidepressants, treatment-emergent somnolence was reported by 54% of patients treated with mirtazapine compared with 18% receiving placebo, although treatment-emergent insomnia complaints were very low (below 2%) [6].

4.3. Trazodone

Trazodone is unique among antidepressants in being used as both a full-dose antidepressant and a low-dose hypnotic. Its sleep-promoting action is mediated principally through antagonism at 5-HT2 serotonin receptors rather than H1 histamine blockade, which distinguishes it from most sedating antidepressants and is associated with a low risk of weight gain [26]. Relative to hypnotics, trazodone is less effective for sleep-onset insomnia and must be administered at least one hour before bedtime, but it is very effective for sleep-maintenance insomnia, particularly in patients with comorbid mental disorders or in patients treated with activating antidepressants [26]. Its hallmark is an increase in deep sleep duration, the opposite of the reduction in slow-wave activity produced by benzodiazepine receptor agonists [26]. Comparative data in MDD are encouraging: in a 12-week study, response rates with trazodone were similar to those with SSRIs, but trazodone was significantly more effective at reducing insomnia severity on the Athens Insomnia Scale (−12.5, 95% CI −15.4 to −9.5, versus −3.7, 95% CI −5.7 to −1.8; p < 0.001) [27]. Treatment-emergent somnolence with trazodone is 46% versus 19% with placebo [6]. The 2023 European Insomnia Guideline, however, notes that evidence for trazodone in insomnia disorder itself shows only small effects on various sleep parameters [28].

4.4. Agomelatine: A Non-Sedating Sleep-Promoting Option

Agomelatine is a first-in-class antidepressant: an agonist at MT1 and MT2 melatonin receptors and an antagonist at 5-HT2b and 5-HT2c serotonin receptors [29,30]. It promotes sleep not through sedation but through resynchronization of the circadian rhythm [5]. Four studies have shown positive effects of agomelatine on sleep continuity and quality and shortening of sleep latency [30], and its chronobiotic properties, mediated primarily through melatonin-receptor agonism in the suprachiasmatic nucleus, translate into clinical advantages: agomelatine improves sleep and circadian rest-activity rhythms compared with venlafaxine in depressed patients [7], and significantly improves the amplitude of the circadian rest-activity/sleep–wake cycle while decreasing depression and anxiety symptoms compared with sertraline [31]. Its antidepressant efficacy has been demonstrated in three short-term, pivotal, randomized, placebo-controlled studies [30], though its effect size relative to placebo has been questioned by some meta-analyses [29].

4.5. Low-Dose Sedating Antidepressants as Hypnotics

Sedating antidepressants are among the most commonly prescribed treatments for insomnia in clinical practice, yet the evidence base is strikingly thin. The one exception is doxepin, which has been approved for the treatment of insomnia characterized by difficulties in maintaining sleep [5]. For low-dose amitriptyline and mirtazapine — widely prescribed off-label for over a decade — placebo-controlled evidence is still lacking, and the DREAMING randomized trial is currently testing low-dose amitriptyline (10–20 mg/day) and mirtazapine (7.5–15 mg/day) against placebo in general practice [32]. A Cochrane review of antidepressants for insomnia identified 23 randomized trials (2,806 participants) but found only three small studies (135 participants) comparing SSRIs with placebo, and combining results was not possible [33]. The European Insomnia Guideline similarly finds that tricyclics have significant but small effects on sleep quality, total sleep time, and sleep efficiency, but not on sleep-onset latency [28]. The gap between prescribing practice and evidence is an important caveat for clinicians who reach for sedating antidepressants primarily as sleep agents in depressed patients.

5. Comparative Efficacy in Depression with Insomnia

5.1. Monotherapy Choice

For the depressed patient whose presenting complaint is dominated by insomnia, the comparative data favor agents with sleep-promoting profiles. Trazodone achieves antidepressant response comparable to SSRIs while producing substantially greater insomnia reduction [27]; mirtazapine improves sleep rapidly and objectively [24,25]; and agomelatine improves sleep continuity and restores circadian rhythmicity without the somnolence burden of sedating agents [7,30,31]. Against the broader comparative-efficacy literature, these are essentially equivalent antidepressants that differ in symptom targeting: the network meta-analysis of 21 drugs found the differences between individual agents to be modest relative to their separation from placebo [4,20]. Sleep effects, therefore, are a legitimate basis for choice precisely because efficacy is not.

5.2. Adjunctive Sleep Medication: Eszopiclone and the Z-Drugs

When the antidepressant of choice is activating — or when sleep fails to improve despite adequate treatment — adjunctive sleep medication is an evidence-supported strategy. A randomized controlled trial of eszopiclone added to fluoxetine produced statistically significant improvements in subjective sleep quality, wakefulness after sleep onset, total sleep time, and sleep efficiency [34]. In a post hoc analysis of two trials in patients with insomnia and anxious depression, eight weeks of eszopiclone co-administered with an SSRI resulted in significantly greater improvement in insomnia, significantly greater reductions in HDRS-17 total score, and significantly greater HDRS-17 response rates than placebo co-administration, although differences in remission rates and in response rates excluding insomnia items were not significant [35].

The strongest quantitative evidence comes from a 2025 systematic review and meta-analysis of combination therapy: compared with antidepressant monotherapy, adding a Z-drug improved remission rates (risk ratio 1.25, 95% CI 1.08–1.45, p = 0.003), depressive symptoms (SMD 0.17, 95% CI 0.01–0.33, p = 0.04), and insomnia symptoms (SMD 0.43, 95% CI 0.28–0.59, p < 0.001) within 12 weeks, with no differences in safety outcomes except dizziness [8]. The authors caution that long-term adjunctive Z-drug therapy has not been evaluated [8]. Low-dose trazodone is a widely used alternative adjunct, and it was among the rescue options permitted in STAR*D [3].

5.3. The Emerging Role of Orexin Receptor Antagonists

A newer avenue directly targets the hyperarousal believed to link insomnia and depression. Seltorexant, a highly selective antagonist of the orexin-2 receptor, has shown both sleep-promoting and antidepressant effects in patients with MDD and comorbid insomnia, improving core depressive symptoms over ten days relative to placebo [36]. While the dual orexin receptor antagonists approved for insomnia — suvorexant, lemborexant, and daridorexant — appear safe and useful for concomitant insomnia in antidepressant-treated patients, they have not been approved for that indication; seltorexant, by contrast, has shown significant antidepressant effects in Phase 2 and Phase 3 trials as an adjunctive treatment for patients with inadequate response to standard antidepressants [37]. At least one more unequivocally positive pivotal study will be required for regulatory approval, but the orexin system represents a mechanistically novel approach to treating the sleep–mood interface [37].

6. The Sexual-Function Trade-Off

6.1. The Magnitude of the Problem

Sexual dysfunction is the adverse effect most likely to be silently endured: rates of treatment-emergent sexual dysfunction in depressed patients from randomized clinical trials range from 15% to 80%, and a European cross-sectional study estimated the prevalence among patients prescribed SSRIs or SNRIs at between 37.1% and 61.5% [10]. Population surveys suggest that over one-third of participants report antidepressant-associated sexual dysfunction [38]. These figures substantially understate the problem, because sexual dysfunction is underreported in efficacy studies — particularly when no targeted or structured method is used to obtain information on sexual functioning at baseline and throughout treatment — and physicians underestimate its prevalence [10,39]. The consequences are not trivial: onset or worsening of sexual dysfunction can result in premature discontinuation of antidepressant treatment, relapse of depression, and worsened health outcomes and quality of life [10].

6.2. Differential Rates Across Agents

Sexual dysfunction is not uniform across antidepressants. The pivotal meta-analysis by Serretti and Chiesa found significantly higher rates of total and phase-specific treatment-emergent sexual dysfunction compared with placebo for sertraline, venlafaxine, citalopram, paroxetine, fluoxetine, imipramine, phenelzine, duloxetine, escitalopram, and fluvoxamine — in decreasing order of impact — whereas no significant difference from placebo was found for agomelatine, amineptine, bupropion, moclobemide, mirtazapine, and nefazodone [11]. Consistent with this, the first study to assess sexual dysfunction across newer antidepressants with a validated rating scale found that SSRIs and venlafaxine XR were associated with higher rates of sexual dysfunction than bupropion or nefazodone [39]. The observational data compiled by Reichenpfader and colleagues illustrate the scale of the differences [10] (Table 1).

6.3. Managing the Trade-Off

For patients who develop sexual dysfunction, the evidence supports switching as a first-line strategy. In randomized trials, nefazodone and bupropion have been associated with less sexual dysfunction than the SSRI sertraline, and reboxetine with greater sexual satisfaction than the SSRI fluoxetine [38]. The meta-analytic data identify bupropion, agomelatine, mirtazapine, and nefazodone as the agents least likely to cause treatment-emergent sexual dysfunction [11]. The practical implication is that the choice of antidepressant can be tailored to a patient’s priorities: for a patient with an active sexual relationship, an agent with a favorable sexual profile may be preferable even if the sleep profile is less ideal, and vice versa.

7. A Framework for Drug Selection

Synthesizing the evidence yields a pragmatic, symptom-priority framework.

When insomnia dominates the presentation. Prefer a sedating antidepressant (mirtazapine, trazodone) or agomelatine, whose sleep-promoting action avoids the somnolence and potential weight effects of histaminergic agents [5,6,26]. Mirtazapine produces rapid, objective sleep improvement but carries a 54% rate of somnolence and greater appetite/weight effects; trazodone is effective for sleep-maintenance insomnia and increases deep sleep but is less potent for sleep-onset difficulties [24,26]. If the antidepressant selected is activating, adjunctive Z-drug therapy is supported by meta-analytic evidence of improved remission, mood, and sleep outcomes [8], and eszopiclone plus an SSRI improves both insomnia and depressive severity [34,35].

When sexual function is the priority. Choose bupropion, agomelatine, or mirtazapine, the agents with no significant difference from placebo in treatment-emergent sexual dysfunction [11,39]; avoid or carefully monitor SSRIs and venlafaxine, which carry rates approaching or exceeding two-thirds of patients in prospective assessments [10,11]. Structured, direct inquiry about sexual function is essential at baseline and follow-up, because spontaneous reporting grossly underestimates the problem [10,39].

Nonpharmacological adjuncts. Cognitive behavioral therapy for insomnia is the first-line treatment for insomnia in current guidelines, and it works in the context of depression: in a meta-analysis of patients with mental disorders and comorbid insomnia, CBT-I reduced insomnia severity with a moderate effect size of 0.5 (95% CI 0.3–0.8) in depressed patients and reduced comorbid depressive symptom severity (0.5, 95% CI 0.1–0.8) [9]. Therapist-delivered CBT-I produces significant reductions in both insomnia and depression severity, and has been recommended as an adjunct for patients whose depression has not remitted with antidepressants [40]. Digitally delivered CBT-I achieves greater reductions in both insomnia and depression severity than sleep education, with significantly higher remission rates, and is comparably efficacious across demographic groups [41]. Notably, CBT-I is superior to traditional depression treatment in improving insomnia symptoms among patients with comorbid psychiatric disorders, though results are mixed on whether it improves overall depression outcomes [42].

A caveat on precision. The ambition to match patients to treatments by total symptom scores faces a statistical obstacle: in a study of 91 double-blind placebo-controlled trials comprising 18,965 participants, variability in response did not differ significantly between antidepressants and placebo, and the authors concluded that the data provide no empirical support for personalizing antidepressant treatment based on total depression scores [43]. Symptom-specific targets — like insomnia and sexual function — may nonetheless be the domains where differential drug effects are real and clinically usable.

8. Conclusions and Future Directions

When depression meets insomnia, the treating clinician faces a two-dimensional problem that most efficacy trials ignore: the same drugs that best treat the mood disorder may worsen the sleep disorder, and the same drugs that improve sleep may carry the heaviest burden of sexual dysfunction [11,23]. The evidence assembled in this review supports four conclusions.

First, sleep effects are a legitimate basis for antidepressant selection because average efficacy differences among drugs are small: activating agents (fluoxetine, paroxetine, venlafaxine, bupropion) can perpetuate insomnia, while mirtazapine, trazodone, and agomelatine actively improve sleep [5,6,7,24,27]. Second, residual insomnia is common, consequential, and undertreated — persisting in up to half of remitters and rarely addressed even when adjunctive options are available [3,17,18]. Third, when the primary antidepressant does not fix sleep, adjunctive Z-drugs are supported by meta-analytic evidence (25% improvement in remission rates) [8], and CBT-I — therapist-delivered or digital — should be considered a first-line adjunct [9,40,41]. Fourth, the sexual-function trade-off is quantifiable and actionable: bupropion, agomelatine, mirtazapine, and nefazodone differ significantly from placebo in sexual dysfunction rates, whereas SSRIs and venlafaxine are associated with rates of 25–73% in prospective assessments [10,11].

Looking forward, the orexin receptor antagonist seltorexant may offer a genuinely novel mechanism at the sleep–mood interface, with Phase 2 and Phase 3 evidence of antidepressant efficacy in patients with inadequate response to standard treatment [36,37]. The DREAMING trial will finally provide placebo-controlled data on low-dose amitriptyline and mirtazapine for insomnia [32]. What is most needed, however, is a change in trial design: comparative antidepressant studies that measure sleep and sexual function prospectively with structured instruments, in addition to mood outcomes, and that test symptom-priority algorithms head-to-head. As a narrative rather than systematic review, this synthesis is subject to selection and interpretation judgments that a protocol-driven review would constrain; the conclusions should be read as a clinically oriented synthesis rather than a quantitative pooled estimate. Until then, the practical answer to “which antidepressant when depression meets insomnia” remains: choose an agent whose sleep and sexual profile matches the patient’s most disabling symptoms, monitor both domains explicitly, and treat residual insomnia — pharmacologically or behaviorally — rather than accepting it as an unavoidable cost of remission [3,10].