Introduction

Insomnia is one of the most common complaints among patients with psychiatric disorders, and patients admitted to inpatient psychiatric units frequently complain of insomnia.^1-4 Comorbid insomnia in patients with psychiatric illness has been shown to negatively impact treatment response and exacerbate symptoms related to comorbid conditions.^1,4,5 Despite its prevalence and impact, there is limited research on the treatment of insomnia in the mixed psychiatric inpatient population.⁶ Per clinical practice guidelines, the first-line approach includes nonpharmacologic strategies such as cognitive behavioral therapy for insomnia.^7-10 However, these approaches require resources, which are often inaccessible and may not be suitable for acutely ill patients in the hospital setting. As such, pharmacological sleep aids are often used instead, though such medication options are limited and not recommended for long-term use.^7-10 Current pharmacotherapy options for insomnia that are approved by the FDA include benzodiazepines, barbiturates, non-benzodiazepine hypnotics (“Z-hypnotics”), dual-orexin receptor antagonists, ramelteon, sedating antihistamines (diphenhydramine, doxylamine), and very low-dose doxepin (3-6 mg). The majority of these options are either habit-forming, costly, or have significant side-effect profiles, so many facilities use off-label sleep aids such as trazodone or low-dose doxepin (10-25 mg).¹¹

Trazodone is a sedating antidepressant with a proposed mechanism to improve sleep through antagonism of the 5-HT2 serotonin receptors.^12,13 It is one of the most frequently used sleep aids in the inpatient psychiatric setting because of its lack of risk for dependence, low cost, and relative safety.¹⁴ Despite how common its use is, the literature regarding the efficacy of trazodone for insomnia is conflicting. A Cochrane review on the use of trazodone for insomnia found an improvement in subjective sleep outcomes and sleep efficiency, though the evidence was only of very low to moderate quality.¹¹ Most of the current insomnia treatment guidelines focus on the management of chronic insomnia and do not recommend trazodone for long-term use.^7,8 The guidelines that comment on the short-term treatment of insomnia state that sedating antidepressants such as trazodone may be effective, especially in the case of comorbid mood disorders.^9,10

Doxepin is another sedating antidepressant with antihistaminic properties that is often used as an alternative to trazodone because of a similar low risk of misuse and safe side-effect profile.¹⁴ It is a tricyclic antidepressant that is primarily studied for insomnia at doses of 3 or 6 mg per day, where it has been shown to improve subjective sleep quality, total sleep time (TST), sleep efficiency, and wakefulness after sleep onset, with safety profiles comparable to placebo.^15-17 Current insomnia treatment guidelines generally recommend these doses of doxepin for treatment of either acute or chronic insomnia over no treatment.^7-9 However, owing to cost concerns, many institutions do not carry these doses of doxepin on formulary; rather, higher doses of 10 or 25 mg are often used. While higher doses are not as well studied for insomnia, they appear to be comparably safe to the FDA-approved doses, with 1 study demonstrating a similar incidence of treatment-emergent adverse events with 6 and 50 mg.¹⁸

This study aimed to evaluate the efficacy and tolerability of trazodone 100 mg versus doxepin 25 mg for psychiatric inpatients who, after initial failure with the commonly ordered trazodone 50 mg, require a change in medication.

Materials and Methods

Endpoints

The primary endpoint was treatment failure of trazodone 100 mg or doxepin 25 mg, which was defined as (1) a dose increase, indicating poor efficacy, (2) a dose decrease, indicating poor tolerability, (3) a change to an alternative medication, indicating either poor efficacy or tolerability, or (4) discontinuation, similarly indicating either poor efficacy or tolerability. Discontinuation of the sleep aid at discharge would only be considered a treatment failure if the patient reported subjectively poor sleep the night before discharge, which was documented daily by nursing staff. The reason for treatment failure was determined using clinician progress notes, which specified whether the dose adjustment or change in sleep aid was due to poor efficacy or poor medication tolerance.

Secondary endpoints included subjective sleep quality, subjective TST, estimated objective TST, estimated time to sleep onset, wakefulness after sleep onset, and nighttime awakenings (Table 1). Each secondary endpoint was recorded as an average between all nights the study medication was taken. All patients were weighted equally for these endpoints regardless of the number of study nights incorporated into the average. If documentation were missing from a study night, the entire night would be excluded such that the average for each patient would only include nights with complete data; however, this was very uncommon because of standardized documentation procedures.

TABLE 1Secondary endpoint definitions

View Fullscreen

Results

All patients admitted to the adult voluntary inpatient psychiatric unit who received at least 1 dose of trazodone 50 mg between July 2020 and June 2022 were screened for eligibility. Of those, 122 patients fit the eligibility criteria, 54 subsequently received trazodone 100 mg, and 68 received doxepin 25 mg. One patient was excluded because of an excluded treatment sequence of trazodone 50 mg, doxepin 25 mg, trazodone 100 mg, followed by doxepin 25 mg once again. No patients were excluded for actively using trazodone or doxepin for sleep before admission. The baseline characteristics and concomitantly used sedating medications were similar between the 2 groups (Table 2).

TABLE 2Patient baseline characteristics

View Fullscreen

The primary outcome of treatment failure was noted in 35.2% of patients in the trazodone group and 41.2% in the doxepin group (P = 0.58) (Table 3). Most treatment failures were related to poor efficacy of the treatment medication, with only a few patients requiring treatment adjustments due to adverse events. Adverse events leading to discontinuation were next-day somnolence, which occurred in 1 patient in the trazodone group and 2 patients in the doxepin group, and constipation, which occurred in 1 patient in the doxepin group (Table 3).

TABLE 3Efficacy and safety endpoints

View Fullscreen

The secondary endpoints were also similar between the 2 groups (Table 3). Patients in the trazodone group had a nonstatistically significant higher fraction of subjectively good sleep days. Patients in the doxepin group had nonstatistically significantly more sleep time according to average subjective TST and average estimated objective TST, faster average time to fall asleep, less average wakefulness after sleep onset, and lower frequency of nighttime awakenings.

Discussion

To our knowledge, this was the first study to compare trazodone and doxepin as sleep aids. After treatment failure with trazodone 50 mg, there was no statistically significant difference between trazodone 100 mg and doxepin 25 mg as a pharmacologic sleep aid in the psychiatric inpatient population. As neither option was superior to the other, it is reasonable to consider either option as an acceptable step-up therapy to trazodone 50 mg.

Of note, the doses used in this study were not concordant with insomnia clinical practice guideline recommendations for the off-label use of trazodone or doxepin. For trazodone, the initial dose of 50 mg is the most commonly used.¹¹ Trazodone 100 mg is often used as a step-up dose after poor efficacy with trazodone 50 mg, as it may be more effective for insomnia.¹⁹ For doxepin, most of the guidelines base their recommendations on doxepin doses of 3 or 6 mg, as these doses have the most data.^7-10 One small study of 47 patients used doxepin doses of 25 to 50 mg and found an acute improvement in sleep efficiency and better sleep quality with doxepin, which was the basis for the use of 25 mg at the study facility.²⁰ Doxepin doses of 10 mg are used as the starting dose for sleep at some other facilities, which may more closely approximate the better-studied 6 mg dose; however, this dose has not been formally studied at the time of writing.

This study had multiple limitations. Group assignments to either trazodone 100 mg or doxepin 25 mg could not be randomized, given the retrospective nature of the study. Potential confounding variables cannot be ruled out, though baseline characteristics were similar between the groups, and the primary prescribing physician on the voluntary psychiatric inpatient unit did not cite a particular reason for picking one option over the other. Additionally, the sleep-parameter endpoints were estimated based on staff member documentation rather than confirmatory sleep studies. It cannot be certain whether patients were truly asleep when they were noted as “eyes closed, lying down” (eg, a patient who is facing away from the door and appears to be asleep might be documented as “eyes closed, lying down” even if they are not asleep). This contributes to a probable overestimation of objective TST and a possible underestimation of time to sleep onset. The subjective TST is also unlikely to be an accurate estimate of actual TST, as patients with insomnia tend to overestimate their sleep problem.²¹ Last, power could not be calculated for this study due to a lack of existing estimates for treatment differences between trazodone and doxepin.

Further data are needed to determine if there is a significant difference between trazodone and doxepin for the treatment of insomnia in an inpatient setting. A comparison between trazodone and doxepin as the initial sleep aid of choice upon admission could support change in the common practice of using trazodone 50 mg in admission order sets. Ultimately, there is insufficient evidence to preferentially recommend either trazodone or doxepin as a pharmacological sleep aid for psychiatric inpatients, and a randomized controlled trial directly comparing these 2 agents could be beneficial to guide treatment in this setting.