Summary
Background
Behavioural, cognitive, and pharmacological interventions can all be effective for insomnia. However, because of inadequate resources, medications are more frequently used worldwide. We aimed to estimate the comparative effectiveness of pharmacological treatments for the acute and long-term treatment of adults with insomnia disorder.
Methods
Findings
We included 170 trials (36 interventions and 47 950 participants) in the systematic review and 154 double-blind, randomised controlled trials (30 interventions and 44 089 participants) were eligible for the network meta-analysis. In terms of acute treatment, benzodiazepines, doxylamine, eszopiclone, lemborexant, seltorexant, zolpidem, and zopiclone were more efficacious than placebo (SMD range: 0·36–0·83 [CINeMA estimates of certainty: high to moderate]). Benzodiazepines, eszopiclone, zolpidem, and zopiclone were more efficacious than melatonin, ramelteon, and zaleplon (SMD 0·27–0·71 [moderate to very low]). Intermediate-acting benzodiazepines, long-acting benzodiazepines, and eszopiclone had fewer discontinuations due to any cause than ramelteon (OR 0·72 [95% CI 0·52–0·99; moderate], 0·70 [0·51–0·95; moderate] and 0·71 [0·52–0·98; moderate], respectively). Zopiclone and zolpidem caused more dropouts due to adverse events than did placebo (zopiclone: OR 2·00 [95% CI 1·28–3·13; very low]; zolpidem: 1·79 [1·25–2·50; moderate]); and zopiclone caused more dropouts than did eszopiclone (OR 1·82 [95% CI 1·01–3·33; low]), daridorexant (3·45 [1·41–8·33; low), and suvorexant (3·13 [1·47–6·67; low]). For the number of individuals with side-effects at study endpoint, benzodiazepines, eszopiclone, zolpidem, and zopiclone were worse than placebo, doxepin, seltorexant, and zaleplon (OR range 1·27–2·78 [high to very low]). For long-term treatment, eszopiclone and lemborexant were more effective than placebo (eszopiclone: SMD 0·63 [95% CI 0·36–0·90; very low]; lemborexant: 0·41 [0·04–0·78; very low]) and eszopiclone was more effective than ramelteon (0.63 [0·16–1·10; very low]) and zolpidem (0·60 [0·00–1·20; very low]). Compared with ramelteon, eszopiclone and zolpidem had a lower rate of all-cause discontinuations (eszopiclone: OR 0·43 [95% CI 0·20–0·93; very low]; zolpidem: 0·43 [0·19–0·95; very low]); however, zolpidem was associated with a higher number of dropouts due to side-effects than placebo (OR 2·00 [95% CI 1·11–3·70; very low]).
Interpretation
Overall, eszopiclone and lemborexant had a favorable profile, but eszopiclone might cause substantial adverse events and safety data on lemborexant were inconclusive. Doxepin, seltorexant, and zaleplon were well tolerated, but data on efficacy and other important outcomes were scarce and do not allow firm conclusions. Many licensed drugs (including benzodiazepines, daridorexant, suvorexant, and trazodone) can be effective in the acute treatment of insomnia but are associated with poor tolerability, or information about long-term effects is not available. Melatonin, ramelteon, and non-licensed drugs did not show overall material benefits. These results should serve evidence-based clinical practice.
Funding
UK National Institute for Health Research Oxford Health Biomedical Research Centre.
Introduction
The prevalence of insomnia in the general population ranges from 12% to 20%.
Insomnia disorder has a chronic course, with persisting symptoms in 86% of individuals after 1 year and 59% after 5 years of a formal diagnosis.
Functional consequences of insomnia include reduced productivity, increased absenteeism, increased use of health care, and increased accident risk, with costs exceeding US$100 billion per year in the USA alone.
Insomnia is also a risk factor for mental health disorders such as depression, anxiety, and alcohol dependence;
,
metabolic syndrome;
hypertension and coronary heart disease;
worsened quality of life;
and increased mortality.
International guidelines recommend cognitive behavioural interventions and medications as effective specific treatments for insomnia disorder,
,
,
,
but the lack of training and availability of clinical staff limit the use of non-pharmacological strategies worldwide. Digital cognitive behavioural therapy has shown some promising results, but more research is needed because this approach is associated with high rates of early dropout or disengagement.
Regulatory agencies have historically approved medications for the treatment of insomnia on the basis of evidence from short-term and placebo-controlled trials, and only recently have asked pharmaceutical companies to submit long-term data for licensing purposes.
As a result, pharmacological treatment is now recommended only for the acute management of insomnia disorder
,
,
,
and little evidence is available about the comparative effectiveness of active treatments.
Therefore, in this study, we did a systematic review and network meta-analysis to inform clinical practice by comparing different pharmacological treatments for the acute and long-term treatment of adults with insomnia.
Results
Figure 1Study selection process
Overall, 154 double-blind, randomised controlled trials correspond to 30 interventions. For the acute treatment analysis, 86 trials (27 interventions) were included for efficacy, 100 trials (28 interventions) for acceptability, and 76 trials (25 interventions) for tolerability. For the long-term analysis, five trials (five interventions) were included for efficacy, eight trials (seven interventions) for acceptability, and eight trials (seven interventions) for tolerability. For safety, 86 trials (27 interventions) were included. RCT=randomised controlled trial. *Industry websites, websites of regulatory agencies, contact with authors, and hand-searched reviews. †The total number of unpublished records is the total number of results for each drug and on each unpublished database source.
For the acute treatment analysis, 86 trials (27 interventions; 21 213 participants) were included for efficacy, 100 trials (28 interventions; 27 991 participants) for acceptability, and 76 trials (25 interventions; 22 811 participants) for tolerability. For the long-term analysis, five trials (five interventions; 2560 participants) were included for efficacy, and eight trials (seven interventions; 5152 participants) for acceptability and tolerability. For safety, 86 trials (27 interventions; 26 543 participants) were included.
Figure 2Network of eligible comparisons for efficacy, acceptability, and tolerability at 4 weeks, and safety (primary and secondary outcomes) at study endpoint
Figure 2Network of eligible comparisons for efficacy, acceptability, and tolerability at 4 weeks, and safety (primary and secondary outcomes) at study endpoint
Figure 4Network meta-analysis for tolerability at 4 weeks and after 3 months, and safety at endpoint
Comparisons should be read from left to right. Tolerability and safety estimates are located at the intersection between the column-defining treatment and the row-defining treatment. Data are in OR (95% CIs). For tolerability, ORs below 1 favour the column-defining treatment. For safety, ORs above 1 favour the column-defining treatment. Pharmacological treatments are reported in alphabetical order. The certainty of the evidence (according to confidence in network meta-analysis [CINeMA]) was incorporated in this figure as footnotes. BZD-I=intermediate-acting benzodiazepine. BZD-L=long-acting benzodiazepine. BZD-S=short-acting benzodiazepine. DARI=daridorexant. DOXE=doxepin. DOXY=doxylamine. ESZO=eszopiclone. LEMB=Lemborexant. MELA=melatonin. OR=odds ratio. PROP=propiomazine. RAME=ramelteon. SELT=seltorexant. SUVO=suvorexant. TRAZ=trazodone. TRIM=trimipramine. ZALE=zaleplon. ZOLP=zolpidem. ZOPI=zopiclone. *High certainty of evidence, †Moderate certainty of evidence. ‡Low certainty of evidence. §Very low certainty of evidence.
Figure 3Network meta-analysis for efficacy and acceptability at 4 weeks and after 3 months
Comparisons should be read from left to right. Efficacy and acceptability estimates are located at the intersection between the column-defining treatment and the row-defining treatment. For efficacy, data are in standardised mean difference (95% CI), and data above 0 favour the column-defining treatment. For acceptability, data are odds ratio (95% CI), and data above 1 favour the column-defining treatment. Pharmacological treatments are reported in alphabetical order. The certainty of the evidence (according to confidence in network meta-analysis [CINeMA]) was incorporated in this figure as footnotes. BZD-I=intermediate-acting benzodiazepine. BZD-L=long-acting benzodiazepine. BZD-S=short-acting benzodiazepine. DARI=daridorexant. DOXE=doxepin. DOXY=doxylamine. ESZO=eszopiclone. LEMB=Lemborexant. MELA=melatonin. QUET=quetiapine. RAME=ramelteon. SELT=seltorexant. SUVO=suvorexant. TRAZ=trazodone. TRIM=trimipramine. ZALE=zaleplon. ZOLP=zolpidem. ZOPI=zopiclone. *High certainty of evidence, †Moderate certainty of evidence. ‡Low certainty of evidence. §Very low certainty of evidence.
Figure 5Summary of Vitruvian plots for the overall profile of each active treatment and placebo across the seven primary outcomes
Efficacy (participants improved), acceptability (all-cause dropouts), and tolerability (dropouts due to adverse events) are reported both as acute treatment (left) and long-term treatment (right). Safety (participants with adverse events) is reported in the bottom wedge and refers to the outcome at end of treatment. Colour indicates the relative performance of the intervention of interest and the precision of the estimate in comparison with placebo, from green (the intervention is better than placebo), to yellow (unclear whether the drug performs better or worse than placebo), and to red (the intervention is worse than placebo). Estimated event rates are expressed as absolute percentages for active treatments (grey rectangles) and placebo (light blue circles). Coloured wedge titles indicate availability of data for the analyses (see the Results section and the appendix [pp 173–195] for more details). AEs=adverse events.
Figure 5Summary of Vitruvian plots for the overall profile of each active treatment and placebo across the seven primary outcomes
Efficacy (participants improved), acceptability (all-cause dropouts), and tolerability (dropouts due to adverse events) are reported both as acute treatment (left) and long-term treatment (right). Safety (participants with adverse events) is reported in the bottom wedge and refers to the outcome at end of treatment. Colour indicates the relative performance of the intervention of interest and the precision of the estimate in comparison with placebo, from green (the intervention is better than placebo), to yellow (unclear whether the drug performs better or worse than placebo), and to red (the intervention is worse than placebo). Estimated event rates are expressed as absolute percentages for active treatments (grey rectangles) and placebo (light blue circles). Coloured wedge titles indicate availability of data for the analyses (see the Results section and the appendix [pp 173–195] for more details). AEs=adverse events.
Discussion
Benzodiazepines are often prescribed not only for insomnia, but also for multiple indications, including generalised anxiety disorder, panic disorder, social phobia, and seizures.
Before starting patients on benzodiazepines, clinicians should always assess the potential benefit and risks for the individual patient, use caution when prescribing additional medications, aim for the lowest effective dose for the shortest treatment duration possible, and taper patients off benzodiazepines slowly, with regular and frequent follow up.
For the short-term treatment of insomnia, our findings suggest that benzodiazepines with intermediate half-lives, such as temazepam and lormetazepam, have better acceptability than short-acting or long-acting compounds.
The antidepressant drug trazodone is used widely as a hypnotic, and its sedative effects are probably attributable to a combination of antihistaminic and noradrenergic α-1 receptor blockade.
With the exception of quetiapine, all the other H1 receptor antagonists mentioned showed some efficacy in terms of quality of sleep in the short-term, but, among them, only doxepin had evidence to suggest its benefits in terms of number of dropouts and adverse events.
Melatonergic interventions had poor efficacy, with no data in the long-term. In our analysis, lemborexant was the most efficacious orexin antagonist for improving sleep in both the short-term and long-term, whereas seltorexant and suvorexant had a better tolerability profile. Daridorexant, approved by the FDA in January, 2022,
did not show an overall material benefit in the treatment of insomnia disorder.
,
,
,
and the findings from our study differ from those of a systematic review and network meta-analysis about hypnotics for insomnia in older adults,
which included 24 studies (5917 patients) and listed doxepin, zaleplon, and suvorexant among the best options. Age can be a moderator of treatment effect, but it is not clear why older adults would respond better to specific medications than the entire population of adults.
The analysis of separate outcomes for acute and long-term treatment, the higher number of studies, the larger total sample, and the inclusion of unpublished data in our review might explain why we found different results.
It is worth noting that we also analysed polysomnography or actigraphy data whenever available, with results being in line with the primary findings.
We found only very few studies evaluating long-term treatment for insomnia. Clinicians and patients should be aware that most of the pharmacological agents used long term for insomnia have only indications for acute treatment from regulatory agencies. Observational studies evaluating hypnotics in the long term found associations with several safety concerns, including dementia,
fractures,
and infections. However, evidence from observational studies should be interpreted with caution as these studies might be biased by residual confounding.
There are online archives where trials are prospectively registered; however, these archives collect reliable information only about the most recent studies.
By making the dataset fully and freely available, we welcome any information that might help to clarify any mistakes or omissions in our dataset.
To increase the methodological rigour of the contributing evidence, we included only double-blind trials, which were very similar in design and conduct. The poor information in terms of risk of bias assessment might be a matter of reporting; however, we presented full details about the risk of bias of all included studies and CINeMA in the appendix (pp 135–40, 252–316). At visual inspection, the network for the acute timepoints is well connected, but the geometry of the networks for the long-term timepoints showed single-standing nodes, almost always connected only to placebo. Comparisons between many active treatments relied on indirect evidence and were based on the untestable consistency assumption, which might have limited the reliability of the results.
Many, but not all, the drugs included in our analysis are off-patent and available in generic form, which might have important implications in terms of public health policy and recommendations from health technology assessment bodies. Of 30 drugs included in our network, only lorazepam is included in the WHO list of essential medicines,
which makes it available worldwide and also ready to use in low-income and middle-income countries.
We hope that these results will inform shared decision making for patients, carers, clinicians, guideline developers, and policy makers. Future studies should focus on the specific characteristics of patients to provide personalised estimates of comparative effectiveness and individualised predictions regarding the probability of response to treatment and of side-effects.
FDC and AC conceived and designed the study. CDG, EGO, CB, and LA contributed to the methods of the study. AK, FF, GLD, MC, NW, and VDF selected the articles and extracted the data. FDC, EGO, CDG, OE, and AC analysed the data. AC, FDC, EGO, GLD, and OE accessed and verified the data. FDC and AC wrote the first draft of the manuscript. GLD, MC, EGO, VDF, NW, AK, AT, ZM, FF, CDG, DJQ, PC, CB, and LA interpreted the data and contributed to the writing of the final version of the manuscript. All authors agreed with the results and conclusions of this Article. AC, EGO, OE, and FDC had full access to all the data, and AC was responsible for the decision to submit for publication.
