Comparative effectiveness and safety of pharmacological and non-pharmacological interventions for insomnia: an overview of reviews

Background This review aimed to assess the existing evidence regarding the clinical effectiveness and safety of pharmacological and non-pharmacological interventions in adults with insomnia and identify where research or policy development is needed. Methods MEDLINE, Embase, PsycINFO, The Cochrane Library, and PubMed were searched from inception until June 14, 2017, along with relevant gray literature sites. Two reviewers independently screened titles/abstracts and full-text articles, and a single reviewer with an independent verifier completed charting, data abstraction, and quality appraisal. Results A total of 64 systematic reviews (35 with meta-analysis) were included after screening 5024 titles and abstracts and 525 full-text articles. Eight of the included reviews were rated as high quality using the Assessment of Multiple Systematic Reviews 2 (AMSTAR2) tool, and over half of the included articles (n = 40) were rated as low or critically low quality. Consistent evidence of effectiveness across multiple outcomes based on more than one high- or moderate quality review with meta-analysis was found for zolpidem, suvorexant, doxepin, melatonin, and cognitive behavioral therapy (CBT), and evidence of effectiveness across multiple outcomes based on one high-quality review with meta-analysis was found for temazepam, triazolam, zopiclone, trazodone, and behavioral interventions. These interventions were mostly evaluated in the short term (< 16 weeks), and there was very little harms data available for the pharmacological interventions making it difficult to evaluate their risk-benefit ratio. Conclusions Assuming non-pharmacological interventions are preferable from a safety perspective CBT can be considered an effective first-line therapy for adults with insomnia followed by other behavioral interventions. Short courses of pharmacological interventions can be supplements to CBT or behavioral therapy; however, no evidence regarding the appropriate duration of pharmacological therapy is available from these reviews. Systematic review registration PROSPERO CRD42017072527.

List (and define whenever it is necessary) the outcomes for which data were recorded, ideally include prioritization of main and additional outcomes. 7 Section/Topic (Sub-) item # Checklist item Reported on page # interest 6c Include adverse events as (primary or secondary) outcome of interest. Define them and grade their severity (such as mild, moderate, severe, fatal; severity could also be described in the appendix), if appropriate. 7-8 6d b Specify report characteristics (such as language restrictions, publication status, and years considered) used as criteria for eligibility for the OoSRs (see also item 7 Specify full electronic search strategy (algorithm) for at least one database including any limits used (e.g. language and date restrictions-see also subitems 6d and 7c) such that it could be repeated.

Appendix B 8b
Present any additional search process (e.g. algorithm or filter for adverse events, searches in pertinent websites) specifically to identify adverse events that have been investigated. Define what is a SR and provide the process for selecting SRs and its relevant details (screening the title and abstract or full text by at least two reviewers, selection by multiple independent investigators and resolving disagreements by consensus).

9c
Report any attempt to handle overlapping (include one review among multiple potential candidates by choosing for example the most updated SR, the most methodologically rigorous SR or the SR with larger number of primary studies). Report additional search to identify eligible primary studies (e.g. searching in more databases or update the search) and its relevant details. N/A 11. Data collection process 11a Describe the method of data extraction from included SRs (e.g. data collection form, extraction in duplicate and independently, resolving disagreements by consensus). 9 11b Report any processes for obtaining, confirming or updating data from investigators (e.g. contact with authors of included reviews, obtain data from primary studies of included reviews). N/A Section/Topic (Sub-) item # Checklist item Reported on page #

Data items 12
List (and define whenever is necessary) the specific variables for which data were recorded (e.g. PICOS items, number of included studies and participants, dose, length of follow up, results, funding sources) and any data assumptions and simplifications made. State the evaluation of reporting or/and methodological quality (eg. using PRISMA or PRISMA-harms, AMSTAR or R-AMSTAR) of the included reviews. 10 13b e State the evaluation of quality for individual studies that were included in the SRs (inform whether tools such as Jadad or RoB of Cochrane were used by the included reviews) and for the additional primary studies. N/A Specify clearly the method (narrative, meta-analysis or network meta-analysis) of handling or synthesizing data and their details (e.g. state the principal summary measures that were extracted or calculated, how heterogeneity was assessed, what statistical approaches were used if a quantitative synthesis has been conducted).

15b
Describe the software that was used to analyze the data if a quantitative synthesis has been conducted. N/A 15c Report if zero events are included in the studies and how they were handled in statistical analyses, if relevant. N/A 15d Describe methods of any pre-specified additional analyses (such as sensitivity or subgroup analyses, meta-regression). N/A

Review & primary study selection 16a
Provide the details of review selection (e.g. numbers of reviews screened, retrieved, and included and excluded in the overview) and the number of the additional eligible primary studies that were included, ideally with a flow diagram of the overview process.

10
16b Present a flow diagram that gives separately the number of studies focused on harms outcomes. Figure 1 16c c List the studies (full citation) that were excluded after reading the full text and provide reasons. Present results in text or/and tables c of any quality assessment (see also subitems 13a-c): 12, Figure  2, Appendix D • Reporting or/and methodological quality of the included SRs.
12, Figure  2, Appendix D • Inform for the quality of the individual studies that were included in the SRs (report results for sequence generation, allocation concealment, blinding, withdrawals, bias etc.) and for the additional included primary studies.

N/A
• Quality of evidence. N/A 20. Present meta-bias(es) 20 Present results of any assessment of meta-bias(es) (such as publication bias or selective reporting across studies, ROBIS assessment). N/A

Synthesis of results 21a
Summarize and present the main findings of the overview for benefits and harms. If a quantitative synthesis has been conducted, present each summary measure with a confidence interval, prediction interval or a credible interval and measures of heterogeneity or inconsistency. 13-17,

Systematic Reviews with meta-analysis (n=35)
Systematic reviews without meta-analysis (n=29)   One study found no carry over effect after 3 weeks of treatment; the second one found that after 7-8 weeks of treatment, return of sleep variables to pre-treatment baseline after withdrawal, and 1 of 11 patients had marked rebound insomnia and daytime anxiety for the first week off; and the third study after 3 mos of treatment, found withdrawal effects despite tapering dose Zopiclone Triazolam  (3) Relative to placebo, patients reported significant improvement during week 1 (p<0.02); during week 2, the trazodone group did not differ significantly from the placebo group 3 trials reported significant improvements throughout the 6-week treatment period for "quality of sleep."(p<0.001); Pittsburgh Sleep Quality Index improvements were similar for the placebo and trazodone groups Trazodone: 9 No comparator (pre-postintervention) Measures of sleep quality improved significantly post-treatment, improvements were maintained at 6 and 12 months; levels of mindfulness were shown to correlate with quality of sleep Appendix F: Tables of primary studies by treatment comparison for outcomes with more than one included SR or SR+MA ----+ ------SOL, TST * Systematic review with meta-analysis 1 The authors of the Soldatos, 1999 paper did not clearly report which studies were included in the analyses for each outcome. All of the primary studies in this review are listed here but are not included in times cited count.
Hermann, 1993 -- The authors of the Soldatos, 1999 paper did not clearly report which studies were included in the analyses for each outcome. All of the primary studies in this review are listed here and are not included in times cited count. *systematic review with meta-analysis This publication includes two trials 2 only able to ascertain that these three trials were the only ones included in the SR+MA, unable to determine which outcome analyses they contributed to thus they are not counted in the final column *systematic review with meta-analysis Table F4. Primary studies across included systematic reviews that examined antidepressants