Effect of graduated drug therapy for moderate-to-severe chronic insomnia on the severity of disease: an observational study in Germany

Abstract

Study Objectives:

Severe chronic insomnia is a common sleep disorder that is mostly persistent and needs to be treated. Pharmacologic treatment options and guidelines are sparse, particularly for long-term treatment. Our study aimed to investigate a graduated therapy scheme for moderate-to-severe chronic insomnia in practice, considering the effects on self-reported sleep quality and quality of life.

Methods:

Patients with moderate-to-severe chronic insomnia were given appropriate medication according to a graduated therapy scheme, ranging from l-tryptophan (as the first choice, least potent) to Z-drugs and combination therapies (as the last option, most potent). Each step of the graduated therapy scheme was tested for at least 4 weeks. Data related to sleep and quality of life were collected in questionnaire form (Insomnia Severity Index, Pittsburgh Sleep Quality Index, Beck Depression Inventory, second edition, and Short Form 36 Health Survey) at baseline and during the course of the treatment after 1, 3, 6, 9, and 12 months.

Results:

Of 86 eligible patients, 60.5% started treatment with l-tryptophan and 8.1% with melatonin. After 3 months, 12.5% were still taking l-tryptophan and 12.5% were taking melatonin. There was a significant decrease in mean Insomnia Severity Index, Pittsburgh Sleep Quality Index, Beck Depression Inventory, second edition, and Short Form 36 Health Survey scores after 3 months of treatment for all patients in the study (n = 64). After 6 months, 22.2% were still taking l-tryptophan, melatonin, or agomelatine, and the remainder had switched to more potent drugs such as antidepressants, hypnotics, daridorexant, or combination therapies.

Conclusions:

A significant number of patients already responded favorably to mild sleep medications, whereas others demonstrated a need for more potent treatments. Ongoing monitoring will evaluate the long-term effectiveness of both approaches.

Clinical Trial Registration:

Registry: German Clinical Trials Register; Name: Schlafqualität und Lebensqualität mit einer medikamentösen Langzeittherapie bei moderater bis schwerer Insomnie; URL: https://drks.de/search/de/trial/DRKS00033175; Identifier: DRKS00033175.

Citation:

Boer J, Toncar T, Stange A, Rosenblum L, Fietze I. Effect of graduated drug therapy for moderate-to-severe chronic insomnia on the severity of disease: an observational study in Germany. J Clin Sleep Med. 2025;21(1):33–45.

BRIEF SUMMARY

Current Knowledge/Study Rationale: Chronic insomnia is a persistent challenge with limited long-term treatment options. Current pharmacological guidelines lack specificity, especially concerning long-term management.

Study Impact: This study investigated a graduated therapy scheme for moderate-to-severe chronic insomnia, focusing on its effects on self-reported sleep quality and quality of life. Initial findings suggest that a significant number of patients respond positively to milder medications, such as l-tryptophan, melatonin, and agomelatine, whereas others require more potent treatments. This study highlights the need for continuous monitoring to assess the long-term efficacy of different medications of the graduated therapy scheme in the treatment of chronic insomnia and to determine which patients (in terms of age and sex) respond to which medications.

INTRODUCTION

Based on the International Classification of Sleep Disorders, third edition, chronic insomnia is a disorder in which patients experience disturbances in nighttime sleep associated with daytime impairments at least 3 times per week for a period of at least 3 months. Disturbances in nighttime sleep include difficulties initiating sleep (sleep-onset insomnia), difficulties maintaining sleep (sleep-maintenance insomnia), and early morning awakening (early-awakening insomnia). Both sleep and daytime impairments may not be caused solely by inadequate opportunities (eg, insufficient time for sleep) or inadequate environments (eg, noisy, bright surroundings).¹ Chronic insomnia as an independent disorder is a disorder that is mostly persistent and needs to be treated.^2,3

The European (and German) medical guideline for insomnia management recommends cognitive behavioral therapy for insomnia (CBT-I) as the first-line treatment for chronic insomnia.^4,5 CBT-I includes sleep hygiene, sleep restriction, relaxation techniques, and cognitive therapy (psychoeducation).^4,6 It has been shown to improve self-reported sleep with fewer side effects than occur with pharmacotherapy.⁶ It is, however, controversial whether CBT-I also leads to an objective improvement in sleep⁷ and whether or to what extent CBT-I is helpful for severe chronic insomnia. Medication including benzodiazepines, nonbenzodiazepines (Z-drugs), and sedating antidepressants may be added to CBT-I for synergistic effects⁸ or if CBT-I is not effective or available.^2,5 Although the administration of sleep medication is only recommended for a short time,⁴ in reality medication is often used long-term (>1 year instead of 4 weeks).⁹ This is understandable, considering that insomnia usually tends to return after the medication is discontinued.²

Current pharmacologic treatment options for patients with chronic insomnia are sparse, particularly for long-term treatment. Treatment options available in Germany include herbal medicines, l-tryptophan, melatonin, agomelatine, antihistamines, sedating antidepressants, neuroleptics, anxiolytics, benzodiazepines, Z-drugs, and daridorexant.¹⁰ Ramelteon, suvorexant, and lemborexant, which are available in the United States, are not approved in the European Union. Conversely, agomelatine and zopiclone are not available in the United States.^10,11

With a lack of national and international recommendations, the treatment of moderate-to-severe chronic insomnia is a major challenge. Sleep medication, especially taken long-term, should be safe and improve both sleep and quality of life (QoL). Considering that benzodiazepines do not meet the safety criteria, these ordinarily should not be given for chronic insomnia.¹² This approach is followed at the sleep medicine outpatient clinic of the Charité – Universitätsmedizin Berlin, where medication is prescribed following a graduated therapy scheme (GTS).¹³ Our study aimed to investigate the described GTS and the individual therapy stages in practice, taking into account the effects on self-reported sleep quality and QoL. The study follows the intent of the current European insomnia guideline, which emphasizes the urgent need to investigate and evaluate insomnia management in clinical practice.⁵

METHODS

Study design

The study was a prospective, longitudinal, observational study aiming to investigate the effects of long-term drug treatment for moderate-to-severe chronic insomnia over a period of 12 months.

In line with the clinical routine of treatment, patients with moderate-to-severe chronic insomnia were given appropriate medication according to the GTS developed by the sleep medicine outpatient clinic of the Charité – Universitätsmedizin Berlin for insomnia, which has been used in clinical practice for around 8 years.¹³ The GTS ranges from l-tryptophan (least potent) to Z-drugs and combination therapies, which since the publication of the aforementioned therapy scheme also includes eszopiclone (approved in Germany in 2020) and the dual orexin antagonist daridorexant (approved in Germany in 2022).¹⁰ Therapy is usually started with daily l-tryptophan obtained from a pharmacy, a dietary supplement that is not recommended by the American Academy of Sleep Medicine for the treatment of chronic insomnia due to the lack of evidence of efficacy and limited evidence of harm¹⁴ but is often favorably accepted by patients of the Charité – Universitätsmedizin Berlin at the outset of pharmacological treatment due to its potential sleep-enhancing properties and lack of associated risks.^15–17 If there is no self-reported improvement, a dose escalation is recommended after 2 weeks. This procedure is repeated for each therapy stage of the GTS (Figure 1). Although it is generally not recommended to start the GTS with herbal medicines, patients may choose to begin with this stage, especially if they have never tried this approach before. Prolonged-release (PR) melatonin was prescribed for melatonin supplementation,¹⁸ including for people under 55 years of age. This decision was based on the positive experience in the United States with ramelteon for sleep-onset insomnia in adults including those under 55 years of age, a melatonin receptor agonist that is not approved in Germany but is used in the United States and whose efficacy and safety profile have been demonstrated in a broader adult population.^14,18,19 Studies on PR melatonin similarly show a low rate of side effects.^14,20 Agomelatine is prescribed for patients who are willing to take a sleep-inducing antidepressant.^4,21 Other sedative antidepressants or an antihistamine (hydroxyzine) are prescribed if the patient wishes (in the case of concomitant depressive mood or anxiety), with the treating physician selecting 3 different sedative antidepressants or the aforementioned antihistamine.¹⁹ Daridorexant is administered at a dosage of 50 mg. This dosage has been shown to provide the best efficacy in both older and younger patients regarding sleep and daytime functioning.^22,23 Daridorexant was tried for 4–8 weeks based on our clinical practice experience regarding the onset of its effect unless the patient wished to continue treatment for a longer period. Z-drugs and combination therapies are the last options in the GTS for sleep-onset and/or -maintenance insomnia. Eszopiclone is given in doses of 1 mg, 2 mg, and 3 mg, with each dose being taken for an initial 2 weeks. Patients were informed about the addictive potential and the potential for tolerance of Z-drugs during quarterly consultations before starting treatment with these medications.

Figure 1. Graduated therapy scheme (GTS) for insomnia developed by the sleep medicine outpatient clinic of the Charité – Universitätsmedizin Berlin.

Each step of the GTS symbolizes a new level of therapy. Prolonged release (PR) melatonin was used for melatonin supplementation.

If therapeutic success for any medication at whatever therapy stage and with whatever dosage is evident, patients should continue to take this medication without increasing the dosage or changing the treatment of the GTS. Only if there is no recognizable therapeutic success, if it is insufficient, if it diminishes, or if intolerable side effects occur should the patient step up to the next level in the GTS. For all medications, and in particular for Z-drugs, patients were advised not to increase the dose by more than the amount described in the GTS. Instead, we recommended switching medications if they experienced reduced efficacy. Patients who have already received medication from the GTS for their chronic insomnia in their medical history and for whom no success was apparent within approximately 4 weeks of intake can skip this stage of medication.

Sleep- and QoL-related data were collected in questionnaire form at baseline and during the course of the treatment after 1, 3, 6, 9, and 12 months. A clinic-specific medical history or progress questionnaire was also filled out at each time point. In addition to completing the questionnaires patients were asked to return to the outpatient clinic every 3 months for a follow-up visit to discuss any queries and the course of therapy. All collected data were pseudonymized and entered and managed using REDCap (Research Electronic Data Capture) tools hosted at Charité – Universitätsmedizin Berlin. REDCap is a secure, online software platform for data collection in research studies.²⁴

The study was conducted in accordance with the Declaration of Helsinki and approved by the ethics committee of the Charité – Universitätsmedizin Berlin on July 25, 2022 (reference number of the research ethics committee: EA1/047/22). The study was registered at the German Clinical Trials Register (DRKS ID: DRKS00033175).

Participants

Patients were recruited from the outpatient department of the Interdisciplinary Center of Sleep Medicine at Charité – Universitätsmedizin Berlin. Eligible patients met the following inclusion criteria: (1) any sex and ages between 18 and 80 years, (2) diagnosis of chronic insomnia according to the International Classification of Sleep Disorders, third edition¹ definition, (3) a moderate-to-severe insomnia according to Insomnia Severity Index (ISI) with a score of ≥ 15 points. Patients were excluded if they met at least 1 of the following criteria: (1) receipt of long-term therapy for moderate-to-severe chronic insomnia in the past 3 months; (2) severe, life-limiting illnesses and symptoms that significantly affect sleep, independent of insomnia, such as malignancies, pain disorders, and unstable or exacerbated illnesses; (3) alcohol or drug abuse; and (4) no form of behavioral therapy had been carried out previously.

After verbal and written information on the study, eligible patients had to agree in writing to participate in the study and thus agree to the use of their pseudonymized data for research purposes. The patient’s signature also certified that the patient wished to undergo drug treatment and did not wish to repeat or continue any type of behavioral therapy.

End points

Sleep parameters

The primary end point, change in self-reported insomnia severity between month 1 (M1) and M3, was assessed using the ISI questionnaire. The patients were handed out the German translation of the questionnaire, which has been shown to be valid for both sexes and different ages.²⁵

The Pittsburgh Sleep Quality Index (PSQI) was used as a second tool to measure the quality of sleep. The total score (between 0 and 21) is the sum of the individual scores obtained in the 7 categories, with (as with the ISI score) higher scores correlating with poorer sleep quality.²⁶

QoL parameters

The Short Form 36 Health Survey (SF-36) was used to measure the patient’s health-related QoL. The SF-36 score is divided into a physical and a mental health score, the so-called Physical Component Summary (PCS) and Mental Component Summary (MCS). Scores above 50 indicate better and scores below 50 worse mental or physical health, respectively, compared to the average of the reference population.²⁷ The calculation of MCS and PCS scores was performed using a preprogrammed SPSS (IBM, Armonk, New York) syntax. The calculation of MCS and PCS values is based on a US reference population but can also be applied to Germany.²⁷ Sleep is not addressed in the SF-36.

Depression severity was measured using the Beck Depression Inventory, second edition (BDI-II), which has been shown suitable for detecting clinically significant (moderate to severe) depression in patients with insomnia.²⁸ For patients with insomnia, raising the cutoff to ≥ 17 points for mild depression reduces the possibility of overclassifying depression.²⁸ To avoid the risk of overclassification, the latter cutoff was used in this study.

Therapy satisfaction and treatment response

During the course of therapy, the patient’s satisfaction with the treatment was ascertained in the form of a progress questionnaire. The patients decided for themselves whether they wanted to stay with the current therapy option or whether they wanted to go on to the next treatment stage due to a self-reported lack of effectiveness or side effects. In accordance with the GTS, which stipulates a change of medication after 4 weeks (or in the case of eszopiclone and daridorexant, 6 and 4–8 weeks, respectively) if treatment is subjectively not successful, treatment response was defined in this study as taking medication for at least 5 consecutive weeks (or in the case of eszopiclone and dariodorexant, 7 and 9 weeks, respectively) up to the time of last data collection. Treatment response was also defined for those patients who chose not to take any medication for at least 5 consecutive weeks up to the time of the last data collection and preferred to maintain this medication-free approach.

Safety

Patients were asked to record treatment-emergent adverse events (TEAEs) as well as therapy discontinuations due to TEAEs in the progress questionnaire. These were also queried separately at the doctor’s consultation, which was held every 3 months.

Sample size

The power calculation for the cohort is based on a paired t test with a 1-sided significance level of α = 2.5% and a power of 90% to reject the null hypothesis. The null hypothesis can be rejected if the patient’s ISI score after 3 months of treatment is equivalent to the ISI score after 1 month of treatment, with an equivalence limit of 0.5 points. A standard deviation of the difference of 1 is assumed. With a dropout estimation of 58%, the power calculation results in a case number of n = 105 patients to ensure a total of 44 evaluable data sets.

Statistics

Baseline data were presented using means and standard deviations for continuous variables, and frequencies were expressed as counts and percentages. Percentages were calculated based on the number of responses to the questions.

To analyze the mean differences in sleep parameters and QoL parameters, Student’s t tests were used. The suitability of these tests was determined through specific assessments for each assumption. The Levene test was used to evaluate homoscedasticity, ensuring that variances across groups were equal. For assessing the approximation to normal distribution, the Shapiro–Wilk test was used, complemented by graphical analyses such as QQ plots and histograms. If the normal distribution approximation was not possible, a log transformation (right-skewed distribution) or the square root transformation (for a slightly right-skewed distribution) was examined. In cases where the assumption of homoscedasticity was not met, the analysis was performed with the Welch t test, which is more robust regarding unequal variances. When the data did not sufficiently meet the assumption of approximate normal distribution, a nonparametric approach was performed using the Wilcoxon rank-sum test. Results were reported as means with 95% confidence intervals, and significance was determined at a P value of less than .05. For scenarios where the conditions for a robust t test or the nonparametric Wilcoxon rank-sum test were not satisfied, particularly in analyzing mean differences in sleep parameters across different medication groups with limited sample sizes, no significance testing was conducted. Instead, descriptive statistics, including mean, quartiles, and standard deviation, were provided to accurately characterize the data.

All data analyses were carried out using SPSS (IBM SPSS Statistics 28.0.1.0) and R Statistical Software (v4.3.1; R Foundation for Statistical Computing, Vienna, Austria).²⁹

RESULTS

Study participants

Of 89 eligible patients, 86 patients started the GTS as recommended by the treating physician (Figure 2). A total of 48 patients had data for the ISI score at both M1 and M3 and were considered for the primary end point analysis. Reasons for discontinuation included voluntary withdrawal and loss to follow-up.

Figure 2. Patient flow chart.

Reasons for withdrawal: ^a1 patient does not want to fill out questionnaires, because it is too exhausting, 1 patient does not want to take sleep medication after all; ^b1 patient left the country, 1 patient is pregnant and does not want to continue using sleep medication, 1 patient is concerned about dependency with sleep medication, 2 patients do not want to take sleep medication after all, 4 unknown; ^cunknown; ^d2 patients experienced a bettering of insomnia, 1 patient did not want to take any more sleep medication; ^epatient does not want to take any more potent sleep medication. *1 patient lost to follow-up after having an allergic reaction with l-tryptophan and switching to melatonin.

The majority of patients were female (75.6%) and under the age of 65 (90.7%; mean age: 46.2 years, age range: 18–73) (Table 1). Three quarters (76.7%) of patients reported being employed or self-employed and 26% of patients were still living in a household with children. Only 8 patients (9.3%) were experiencing only sleep-onset insomnia and 27 patients (31.4%) were experiencing only sleep-maintenance insomnia. A total of 59.3% of patients had a combined sleep-onset and sleep-maintenance insomnia (mixed insomnia). Before participation in the study, daytime sleep was impossible for most patients (63.4%). On average, patients had been experiencing insomnia for more than 12 years (12.3 ± 11.2 years) before making an appointment in our sleep clinic, and half of them were already sensitive sleepers before the onset of insomnia. Over half of the patients (55.6%) reported sleep disorders in the family. A total of 79% of patients had tried sleep medication before visiting our sleep clinic, the most common being over-the-counter melatonin preparations (43.0%), herbal medicines (41.9%), and antihistamines (38.4%).

Table 1.

Baseline characteristics.

	Patients at Baseline Assessment (n = 86)
Female, n (%)	65 (75.6)
Age, years, mean (SD)	46.2 (13.9)
Body mass index, kg/m2, mean (SD)	23.5 (3.5)
Patients with ≥ 1 concomitant condition, n (%)	45 (52.3)
Most common concomitant condition (> 5% of at least 1 group), n (%)
Hypothyroidism	12 (14.0)
Hypertension	8 (9.3)
Depression	6 (7.0)
Previous nonpharmacological treatment attempts, n (%)
Sleep hygiene	30 (96.8)
Relaxation technique	67 (77.9)
Sleep restriction	5 (16.1)
Cognitive therapy (psychoeducation)	13 (41.9)
Previous pharmacological treatment attempts, n (%)
Herbal medicines	36 (41.9)
Melatonin	37 (43.0)
Antihistamines	33 (38.4)
Antidepressants	13 (15.1)
Nonbenzodiazepines (Z-drugs)	13 (15.1)
Benzodiazepines	7 (8.1)
Other hypnotics and sedatives	15 (17.4)
Baseline ISI score, mean (SD)	20.0 (3.5)
ISI score ≥ 22 points, n (%)	29 (33.7)
Baseline PSQI score, mean (SD)	13.4 (3.0)
Baseline BDI-II score, mean (SD)	13.9 (9.9)
BDI-II score 0–17 points, n (%)	68 (85.0)
BDI-II score 18–28 points, n (%)	12 (15.1)
BDI-II score ≥ 29 points, n (%)	0 (0.0)
Baseline SF-36 scores, mean (SD)
Mental Component Summary	37.2 (11.7)
Physical Component Summary	49.1 (9.0)

BDI-II = Beck Depression Inventory, ISI = Insomnia Severity Index, PSQI = Pittsburgh Sleep Quality Index, SD = standard deviation, SF-36 = Short Form 36 Health Survey.

GTS in practice

Of the 86 patients, 52 (60.5%) started treatment with l-tryptophan, 7 (8.1%) with melatonin, and 4 (4.7%) with antidepressants (Figure 3). At M3, 8 patients (12.5%) were still taking l-tryptophan and melatonin, respectively. After M3, 17 patients (41.4%) continued using either herbal medicines, l-tryptophan, melatonin, or agomelatine, 7 patients (17.1%) continued treatment with antidepressants, 12 (29.3%) with Z-drugs, and 2 (4.9%) with daridorexant. After M6, 4 patients (22.2%) were still using l-tryptophan, PR melatonin, or agomelatine, 3 (16.7%) were using antidepressants, 8 (44.4%) were using Z-drugs, and 3 (16.7%) were using daridorexant.

Figure 3. Absolute number of medications taken per calendar week.

Color codes indicate the different medications. The numbers below the graph reflect the absolute numbers of medications per calendar week and drug.

For 61.5% of 65 patients (patients that had reached the fifth calendar week) a treatment response was detected to any medication of the GTS ranging from herbal medicines to Z-drugs and daridorexant (Figure 4): for 9 patients (13.8%) using agomelatine, 7 patients (10.8%) using l-tryptophan, 7 (10.8%) using eszopiclone, and 4 patients (6.2%) using PR melatonin. The average age of patients with treatment response to PR melatonin was 48.3 years, slightly higher than the mean age of 47.2 years observed in the melatonin-nonresponder group. For all responder groups, no significant difference was found between the time since the onset of insomnia (P > .05).

Figure 4. Number of self-reported treatment responses per medication group.

Criterion of self-reported treatment response: taking a medication (combination) for at least 5 consecutive weeks (or 7 and 9 weeks for eszopiclone and daridorexant, respectively) up to the time of the last data collection. Two patients had not yet reached the fifth calendar week and were therefore excluded from the analysis. Color codes indicate the respective medication groups.

Sleep quality

There was no significant difference in baseline ISI for patients who showed a self-reported treatment response during the course of the study compared with those who did not (20.19 ± 3.76 vs 19.86 ± 2.98 points; t value = −0.349, degrees of freedom = 63, P = .728) (Table S1 in the supplemental material). l-tryptophan responders had the lowest mean ISI score at baseline (17.43 ± 2.99 points) compared with all medication groups.

Between M1 and M3, there was a marginal reduction in self-reported insomnia severity, reflected by a mean ISI difference of −0.92 points (95% confidence interval [CI], −2.46 to 0.63). However, this change did not reach statistical significance (Table 2). The mean differences in the ISI scores for all patients from M1, M3, and M6 compared to baseline were statistically significant. The scores decreased from −3.19 points (95% CI, −4.44 to −1.95; P < .001) at M1 to −4.11 points (95% CI, −5.54 to −2.69; P < .001) at M3 and further to −6.27 points (95% CI, −7.87 to −4.67; P < .001) at M6. When selecting only those patients who had a self-reported response to treatment as defined by us during the study (66.2% of patients), the mean differences in ISI score were even greater: −4.09 points (95% CI, −5.83 to −2.35; P < .001) at M1, −4.61 points (95% CI, −6.39 to −2.83; P < .001) at M3, and −6.77 points (95% CI, −8.56 to −4.99; P < .001) at M6.

Table 2.

Mean change in ISI and PSQI between M3 and M1, M1 and BL, M3 and BL, and M6 and BL.

		All Patients		Patients Showing Treatment Response
		Mean Difference [95% CI]	n	Mean Difference [95% CI]	n
M3 – M1	ISI	−0.92 [−2.46; 0.63]	48	−0.55 [−2.35; 1.26]	31
	PSQI	−0.67 [−1.55; 0.22]	48	−0.71 [−1.87; 0.45]	31
M1 – BL	ISI	−3.19 [−4.44; −1.95]***	52	−4.09 [−5.83; −2.35]***	33
	PSQI	−0.10 [−0.95; 0.75]	49	−0.84 [−1.94; 0.26]	31
M3 – BL	ISI	−4.11 [−5.54; −2.69]***	62	−4.61 [−6.39; −2.83]***	41
	PSQI	−0.92 [−1.77; −0.06]*	59	−1.49 [−2.55; −0.42]**	39
M6 – BL	ISI	−6.27 [−7.87; −4.67] ***,a	37	−6.77 [−8.56; −4.99]***,a	31
	PSQI	−2.09 [−3.4; −0.78]**	35	−2.66 [−4.14; −1.17]***	29

The differences are shown for all patients and for those who responded to treatment. *P < .05. **P < .01. ***P < .001. ^aMean ISI difference ≥ 6 points. BL = baseline, CI = confidence interval, ISI = Insomnia Severity Index, M = month, PSQI = Pittsburgh Sleep Quality Index.

At M1, 21.6% of the patients (11 in total) achieved an improvement of 7 or more points in their ISI scores. This figure rose to 30.6% (19 patients) by M3 and further to 45.9% (17 patients) by M6. At M9, the improvement was sustained by 28.6% of the patients (4 in total), and by M12, 57.1% (also 4 patients) continued to show improvement in their ISI scores. At M3 and M6, 10 (15.6%) and 6 patients (16.2%) still reported an ISI of ≥ 22 points.

The mean differences in the PSQI scores were statistically significant for all patients for M3 and M6 compared to baseline. The scores decreased by −0.92 points (95% CI, −1.77 to −0.06; P < .05) at M3 and −2.09 points (95% CI, −3.40 to −0.78; P < .01) at M6. When selecting only those patients who had a self-reported response to treatment during the course of the study (66.2% of patients), the mean differences in PSQI score were even greater: −1.49 (95% CI, −2.55 to −0.42; P < .01) at M3 and −2.66 (95% CI, −4.14 to −1.17; P < .001) at M6.

l-tryptophan responders showed the greatest ISI reductions compared to treatment responders of other medication groups between M1 and baseline (−8.00 ± 3.92 points), M3 and baseline (−8.71 ± 5.28 points), and M6 and baseline (−11.00 ± 1.73 points) (Figure 5). Agomelatine responders showed a decrease in mean ISI score between M1 and baseline (−5.57 ± 3.78 points), with all agomelatine responders taking l-tryptophan until this point. The mean decrease in ISI score was maintained at M3 and M6. Both treatment responders with antidepressants/antihistamines and treatment responders with Z-drugs/daridorexant showed a mean ISI difference of ≥ 6 points between M6 and baseline: −9.43 ± 5.16 points and −6.27 ± 3.98 points, respectively (Table S2 and Table S3 in the supplemental material).

Figure 5. Change in ISI scores among patient groups responding to treatment with various medication groups.

The differences are shown for M1 and BL, M3 and BL, M6 and BL, and M3 and M1. Median (black, solid line), mean (white, dotted line), first and third quartile (box borders), and outliers (black dots) are shown. BL = baseline, ISI = Insomnia Severity Index, M = month.

QoL

There is no significant difference in the mean baseline BDI-II for patients who showed a self-reported treatment response during the course of the study compared to those who did not (12.74 ± 8.47 vs 15.86 ± 11.59 points; t value = 1.232, degrees of freedom = 62, P = .223) (Table S1). l-tryptophan responders had the lowest mean baseline BDI-II score (8.86 ± 7.82 points), and patients showing no treatment response had the highest mean baseline BDI-II score (15.86 ± 11.59 points). l-tryptophan responders also had the highest mean baseline PCS (51.4 ± 7.61) and second-highest mean baseline MCS (43.8 ± 9.23) scores compared to other treatment responders. Patients responding to antidepressants/antihistamines showed the lowest mean baseline MCS scores (31.82 ± 10.01) while having the second-highest PCS scores (50.3 ± 10.58).

There was a significant improvement in mean BDI-II scores at M1, M3, and M6 (Table 3). The scores decreased from −1.76 points (95% CI, −3.39 to −0.14; P < .05) at M1 to −3.43 points (95% CI, −5.39 to −1.46; P < .001) at M3 and further to −3.67 points (95% CI, −6.06 to −1.27; P < .01) at M6. The effect was less pronounced after selecting only those patients who had a self-reported response to treatment, with mean BDI-II score decreases from −1.41 points (95% CI, −3.54 to 0.72) at M1 to −2 points (95% CI, −4.19 to 0.19) at M3 and further to −2.6 points (95% CI, −5.06 to −0.14; P < .05) at M6.

Table 3.

Mean change in BDI-II, MCS, and PCS between M3 and M1, M1 and BL, M3 and BL and M6 and BL.

		All Patients		Patients Showing Treatment Response
		Mean Difference [95% CI]	n	Mean Difference [95% CI]	n
M3 – M1	BDI-II	−1.4 [−2.79; 0]*	48	−0.32 [−1.54; 0.9]	31
	MCS	0.29 [−1.5; 2.07]	45	−0.38 [−2.4; 1.63]	28
	PCS	0.74 [−1.62; 3.11]	45	0.58 [−1.94; 3.1]	28
M1 – BL	BDI-II	−1.76 [−3.39; −0.14]*	51	−1.41 [−3.54; 0.72]	32
	MCS	1.68 [−0.44; 3.8]	41	2.74 [−0.13; 5.61]	25
	PCS	1.48 [−1.09; 4.05]	41	1.68 [−1.91; 5.27]	25
M3 – BL	BDI-II	−3.43 [−5.39; −1.46]***	61	−2 [−4.19; 0.19]	40
	MCS	1.73 [−0.34; 3.79]	51	2.42 [−0.34; 5.17]	33
	PCS	1.81 [−1.11; 4.74]	51	1.41 [−1.65; 4.47]	33
M6 – BL	BDI-II	−3.67 [−6.06; −1.27]**	36	−2.6 [−5.06; −0.14]*	30
	MCS	2.37 [−0.7; 5.44]	29	2.52 [−0.99; 6.02]	24
	PCS	1.66 [−3.08; 6.4]	29	1.16 [−4.45; 6.78]	24

The differences are shown for all patients and for those who respond to treatment. *P < .05. **P < .01. ***P < .001. BDI-II = Beck Depression Inventory, BL = baseline, CI = confidence interval, M = month, MCS = Mental Component Summary, PCS = Physical Component Summary.

There was a steady improvement in the MCS and PCS from M1 to M6. Increases in MCS and PCS did not reach statistical significance at any time point (Table 3).

l-tryptophan and agomelatine responders showed the greatest mean MCS score improvements compared to treatment responders of other medication groups at M6 compared to baseline (Table S2). PR melatonin responders showed the greatest mean PCS score improvement by M6 (5.03 ± 0.25 points). Patients who showed no treatment response had the greatest mean BDI-II score decrease of −9 ± 7.67 points (Table S3).

Safety

No serious adverse events occurred during the course of the study. In total, 6 patients experienced TEAEs leading to discontinuation of medication intake. Three patients encountered TEAEs that necessitated the discontinuation of l-tryptophan (TEAEs: headache, dry oral mucosa, dizziness, nightmares, heartburn, allergic reaction). Additionally, 1 patient on eszopiclone (TEAEs: eczema, skin rash, redness, dry skin, bad breath, limited coordination ability), another on a combination of eszopiclone and mirtazapine (TEAEs: dry eyes, tinnitus intensification, depression intensification), and 1 on daridorexant (TEAEs: nausea, diarrhea) experienced TEAEs, leading to the cessation of these medications. There was also 1 case where a patient faced adverse events without being on any medication (adverse events: headache, muscle tremor in the chest).

DISCUSSION

Chronic insomnia, especially when it is severe, tends to persist.³ Persistent insomnia is correlated with various other diseases including cardiovascular and neurodegenerative diseases, anxiety disorders, and depression, which can result in suicide.^30–34 Because spontaneous remissions cannot be assumed, drug treatment for chronic insomnia should be initiated following diagnosis.³ We pursued this approach as part of a graduated therapy and found that even mild sleep-promoting medication can subjectively help with moderate-to-severe insomnia, but only in a minority of those affected.

The patients in our study are similar in phenotype to those in previously described studies, related to sex distribution, age, and body mass index.^32,35,36 Combined sleep-onset and sleep-maintenance insomnia occurred more frequently than sleep-onset or sleep-maintenance insomnia alone.³⁵ At baseline, only about 15% of patients showed signs of moderate-to-severe depression with a BDI-II of over 17 points,²⁸ and both PCS and MCS scores of the SF-36 were below 50, indicating below-average physical and mental health compared to the reference population. Although the variance in physical health is marginal, there is a notable discrepancy in mental health, highlighting substantial impairments of patients with untreated insomnia. These findings are consistent with the baseline results on mental health observed in a larger study involving over 800 patients with insomnia.³⁷

Most of our patients started treatment with l-tryptophan. By the third month of treatment, there was an observable amelioration in insomnia severity overall, characterized by reduced mean ISI and mean PSQI scores. Notably, around one third of patients achieved a substantial improvement of 7 or more points in their ISI scores by M3, enabling them to transition to a lower level of insomnia severity as defined by Morin et al.³⁸ Moreover, on average, patients reported fewer depression symptoms and an enhanced QoL after 3 months on the GTS. There was a marginal reduction in self-reported insomnia severity between the first and third month of treatment that was, however, not significant, leading to the primary end point’s remaining unmet.

At M3, one eighth of patients continued to respond favorably to 500 mg and 1,000 mg of l-tryptophan. For these patients, notable improvements were observed in ISI, PSQI, BDI-II, and SF-36 scores. The American Academy of Sleep Medicine did not find clinically significant efficacy for a dosage of 250 mg of l-tryptophan.¹⁴ This difference in response may be due to varying dosages of l-tryptophan used in these studies or differences in the measuring instruments used to determine insomnia severity. Sutanto et al suggest that higher doses of tryptophan (≥ 1 g) show improved efficacy compared to lower doses (< 1 g).³⁹

At M3, another one eighth of patients exhibited a favorable response to PR melatonin. These patients showed improvements in ISI, BDI-II, and SF-36 scores within the first month of treatment with most patients taking l-tryptophan until this point. The ISI continued to decrease even more by the third month, coinciding with the patients transitioning to the second phase of the GTS, PR melatonin. However, over the course of 1–6 months, the differences in BDI-II scores and mental health scores compared to baseline gradually became smaller. Yet, one eighth of the patients still decided that the treatment was sufficient. Similarly, a study focusing on melatonin supplementation in patients with insomnia revealed no significant decrease in PSQI scores. Although there was a reduction in ISI scores, it was not statistically significant.⁴⁰

Agomelatine responders showed an improvement in mean ISI, BDI-II, and SF-36 scores between M1 and baseline. Notably, all agomelatine responders had been taking l-tryptophan until this point, indicating that agomelatine responders also positively responded to l-tryptophan. The improvements in ISI, BDI-II, and SF-36 scores observed at M1 initially decreased slightly by M3 before showing signs of improvement again at M6. A study on agomelatine intake over 8 weeks also demonstrated ISI and PSQI score reductions as well as simultaneous bettering of depressive symptoms, and the improvements were already apparent after 1 week.²¹

After 6 months, only a few patients chose to persist with mild insomnia medications. Most decided that a more potent therapy with antidepressants/antihistamines, hypnotics, or daridorexant was necessary to treat their insomnia. By M6, both treatment responders with antidepressants/antihistamines or Z-drugs/daridorexant on average achieved an improvement of 7 or more points in their ISI scores such that they could be downgraded at least 1 level of insomnia severity. The general effectiveness of Z-drugs and daridorexant has been demonstrated in various studies and meta-analyses for different sleep parameters.^21,23,41,42

Treatment response was detected for around two thirds of patients. Approximately 10% each responded to l-tryptophan, agomelatine, and eszopiclone. There was no significant disparity in baseline ISI scores between patients who exhibited self-reported treatment responses during the study and those who did not. However, the data suggested that individuals responding to milder sleep medications generally presented lower baseline ISI and PSQI scores.

No serious TEAEs occurred with any medications during the course of the study. This agrees with other studies showing that herbal medicines, PR melatonin, agomelatine, doxepin, mirtazapine, eszopiclone, daridorexant, zopiclone, and zolpidem are safe treatment options for insomnia.^{14,21,40,41,43–45} Sufficient and systematic evaluation of safety data on l-tryptophan, trimipramine, trazodone, and hydroxyzine intake for insomnia is still missing.^2,14 After being temporarily banned by the German Federal Health Agency and the United States Food and Drug Administration in 1989, most likely due to contamination issues in certain l-tryptophan preparations causing eosinophilia-myalgia syndrome, l-tryptophan was reintroduced with strict purity standards in 1996 in Germany and in 2001 in the United States.^46,47 In our study, l-tryptophan was prescribed to ensure it was obtained from a pharmacy, a reputable source for medications in Germany.

Limitations of the study include the rather small sample size for the first analysis. The limited duration of observation and the specific criterion for treatment response should be taken into account when interpreting the results in terms of treatment efficacy. These findings warrant further investigation in a larger cohort and over an extended period. Additionally, the sleep quality data in this study are solely based on self-reported data. The absence of polysomnographic data precludes any conclusions about objective improvements in sleep quality.⁴⁸ Whether the efficacy of the milder drugs, especially l-tryptophan, is primarily due to a placebo effect remains uncertain. This is particularly relevant because it is known that placebo effects can be significant in the treatment of insomnia.⁴⁹ Existing studies suggest that the placebo effect of drug treatment is stronger than the placebo effect induced by CBT.⁵⁰ The GTS thus offers a strategic opportunity to identify those patients who respond positively to the placebo effect and enables effective treatment with mild agents.

CONCLUSIONS

In summary, a significant proportion of patients with moderate-to-severe chronic insomnia displayed favorable responses to mild sleep medications, which included the first steps of the GTS, l-tryptophan, PR melatonin, and agomelatine. At the same time, a notable portion of patients demonstrated their necessity for more potent medications due to their lack of response to the mild treatments. This study does not provide conclusive evidence that the GTS, and with that, l-tryptophan, should be included in guidelines for insomnia therapy. However, given the limited availability of over-the-counter insomnia treatments in Germany, the study suggests that l-tryptophan could potentially be of benefit to insomnia patients, even those with severe insomnia.

Ongoing monitoring of these patients will continue, extending observations up to M12, to evaluate the long-term effectiveness of both mild and more potent sleep medications and to determine which patients (for instance in terms of age and sex) are responding to which medications.

DISCLOSURE STATEMENT

Work for this study was conducted at Charité – Universitätsmedizin Berlin, Interdisziplinäres Schlafmedizinisches Zentrum, Berlin, Germany. Only internal funds from Charité – Universitätsmedizin Berlin were used to finance this project. T.T. and I.F. prescribed off-label prolonged-release melatonin to patients under 55 years of age and sedating antidepressants for insomnia. The rationale for this approach is based on a combination of experience, recommendations, studies, and available resources (see Study design). I.F. has received a speaker honorarium from Hennig Arzneimittel GmbH & Co. KG, Idorsia Ltd., and Vanda and has been involved as a consultant for Idorsia Ltd., P&G, and STADA. J.B., T.T., A.S., and L.R. report no conflicts of interest.

ABBREVIATIONS

BDI-II

Beck Depression Inventory

CBT-I

cognitive behavioral therapy for insomnia

CI

confidence interval

GTS

graduated therapy scheme

ISI

Insomnia Severity Index

M1/M3/M6/M9/M12

month 1/3/6/9/12

MCS

mental component summary

PCS

physical component summary

PR

prolonged release

PSQI

Pittsburgh Sleep Quality Index

QoL

quality of life

REDCap

Research Electronic Data Capture

SF-36

Short Form 36 Health Survey

SOL

sleep onset latency

TEAE

treatment-emergent adverse event