Dayvigo Evidence Base Graded by GRADE: What the Clinical Trials Actually Show

What Is Lemborexant and Why Does GRADE Grading Matter Here?

Lemborexant is a small-molecule competitive antagonist at both orexin receptor subtypes (OX1R and OX2R). By blocking wake-promoting orexin/hypocretin signaling rather than broadly suppressing CNS activity, it aims to reduce sleep latency without the next-day impairment profile that limits benzodiazepines and Z-drugs in clinical practice. [1]

The GRADE (Grading of Recommendations Assessment, Development and Evaluation) framework rates evidence quality across four domains: risk of bias, inconsistency, indirectness, and imprecision. Applying it to insomnia pharmacotherapy matters because sleep trials differ widely in outcome measurement, duration, and comparator choice, making raw p-values an unreliable guide to real-world confidence. [2]

How GRADE Domains Apply to Insomnia Trials

Risk of bias in sleep studies is shaped heavily by blinding fidelity and whether polysomnography (PSG) or patient-reported outcomes (PROs) serve as primary endpoints. PSG-based outcomes reduce performance bias but add cost and may not reflect bedroom conditions. PRO-based outcomes, like subjective sleep onset latency (sSOL), carry higher risk of expectation bias but better reflect the patient's lived experience.

Inconsistency and indirectness are relevant because most DORA trials enrolled adults with chronic insomnia disorder per DSM-5 criteria but excluded common comorbidities like obstructive sleep apnea, depression, and shift-work disorder. Estimates may not transfer directly to those populations.

SUNRISE-1: The Core PSG Evidence

SUNRISE-1 is the highest-quality individual trial in lemborexant's evidence base. Published in JAMA Network Open (2019), it enrolled 291 adults aged 55 and older with insomnia disorder and used a double-blind, placebo- and active-controlled crossover design with in-laboratory PSG. [1]

Primary Endpoints and Effect Sizes

The primary endpoint was sleep efficiency (SE) on night 1 and nights 2 through 3. Lemborexant 5 mg produced a mean SE of 78.0% versus 72.7% for placebo (P<0.001). Lemborexant 10 mg achieved 80.2% SE versus the same placebo arm. Both doses also significantly reduced LPS (latency to persistent sleep): 10 mg cut LPS by a mean of 17.1 minutes versus placebo on night 1 (P<0.001). [1]

Against zolpidem ER 6.25 mg, lemborexant 10 mg showed statistically superior sleep efficiency at nights 2 through 3, and lemborexant 5 mg showed non-inferiority. This active-comparator data is rare in insomnia pharmacology and moves the GRADE certainty for the SE endpoint versus zolpidem into the moderate range.

Next-Morning Function: The Differentiating Signal

SUNRISE-1 included a morning balance and driving simulator assessment at 9 hours post-dose. Zolpidem ER 6.25 mg showed statistically significant impairment on the driving simulation score at hour 9 (P<0.05 versus placebo). Lemborexant 5 mg did not differ from placebo on that measure. Lemborexant 10 mg did show a residual impairment signal, which informed the FDA labeling warning about morning driving after the 10 mg dose. [3]

This functional outcome is not always captured in GRADE summaries that focus only on PSG metrics. Its inclusion raises the clinical relevance of the 5 mg dose profile.

SUNRISE-2: Long-Duration Evidence

SUNRISE-2 addressed a gap that is common in sleep pharmacology: most trials run 4 weeks or fewer, leaving long-term efficacy and safety uncertain. Published in Sleep (2020), SUNRISE-2 randomized 949 adults (aged 18 and older) with insomnia disorder to lemborexant 5 mg, lemborexant 10 mg, or placebo for 12 months, with subjective sleep outcomes as primary endpoints and monthly PSG in a sub-cohort. [4]

Twelve-Month Efficacy Outcomes

At month 1, lemborexant 10 mg reduced sSOL by a mean of 18.4 minutes versus 7.1 minutes for placebo. That gap was maintained at month 6 (17.1 vs. 6.2 minutes) and month 12 (16.4 vs. 6.0 minutes), demonstrating durability without dose escalation. [4]

Wake after sleep onset (WASO), a measure of sleep maintenance, also improved: lemborexant 10 mg reduced subjective WASO by 30.4 minutes at month 12 versus 12.6 minutes for placebo (P<0.001). The 5 mg dose produced intermediate but still statistically significant reductions at all time points.

Rebound and Withdrawal Assessment

One of the GRADE indirectness concerns with short trials is that they cannot capture rebound insomnia or withdrawal, both of which affect the benefit-harm calculation in chronic use. SUNRISE-2 evaluated sleep in the week after treatment discontinuation. Neither dose showed statistically significant rebound insomnia versus placebo, a finding that distinguishes DORAs from benzodiazepines at a mechanistic level. [4]

That finding carries moderate GRADE certainty: it is based on a single trial with subjective measurement, though the absence of a rebound signal was consistent across both doses.

Network Meta-Analysis Context: How Does Lemborexant Rank?

Single-trial evidence is necessary but not sufficient for GRADE certainty. Two network meta-analyses published after the SUNRISE program provide indirect comparison data across hypnotics.

A 2022 Lancet meta-analysis by Crescenzo et al. (N=30 trials, 15 interventions) ranked lemborexant 10 mg among the most effective interventions for sleep-onset latency and subjective sleep quality, with a surface-under-the-cumulative-ranking (SUCRA) score of 84 for sSOL. [5] That analysis rated the overall evidence certainty as moderate due to heterogeneity in trial duration and outcome definitions, consistent with a formal GRADE downgrade for inconsistency.

A separate 2022 BMJ meta-analysis by De Crescenzo et al. Covering 154 trials found that most benzodiazepines and Z-drugs outperformed placebo for short-term sleep outcomes but carried higher residual sedation risk. Lemborexant was rated favorably on the safety-outcome balance, particularly for older adults. [6]

Applying the GRADE Certainty Ratings Systematically

The table below summarizes GRADE certainty by outcome domain based on the available evidence as of mid-2025:

| Outcome | Best Evidence | GRADE Certainty | Key Downgrade Reason | |---|---|---|---| | PSG sleep efficiency vs. Placebo | SUNRISE-1 (N=291, PSG) | High | None identified | | Latency to persistent sleep vs. Placebo | SUNRISE-1 (N=291, PSG) | High | None identified | | Subjective SOL vs. Placebo, 12 months | SUNRISE-2 (N=949) | Moderate | Single trial; PRO endpoint | | WASO vs. Placebo, 12 months | SUNRISE-2 (N=949) | Moderate | Single trial; PRO endpoint | | PSG SE vs. Zolpidem ER 6.25 mg | SUNRISE-1 (N=291) | Moderate | Single active-comparator trial | | Next-morning driving (5 mg) | SUNRISE-1 sub-study | Moderate | Small sub-sample, simulator not on-road | | Rebound insomnia absence | SUNRISE-2 (N=949) | Moderate | Single trial, subjective measure | | Long-term safety beyond 12 months | Post-marketing surveillance | Low | No RCT data beyond 12 months |

FDA Regulatory Evidence: What the Agency Weighted

The FDA approved lemborexant in December 2019 based on SUNRISE-1 and SUNRISE-2, plus two additional Phase 2 dose-ranging studies and a thorough QTc study. [3] The agency's review identified no clinically meaningful QT prolongation at approved doses, which is a relevant safety signal for any sedative in older adults receiving polypharmacy.

The prescribing label carries two notable GRADE-relevant items. First, the agency required a scheduled substance designation (Schedule IV) despite the mechanistic difference from benzodiazepines, citing the possibility of abuse, though no head-to-head abuse-potential RCT was required for approval. Second, the label specifies that patients should not drive until at least 8 hours after taking the 10 mg dose, based on the SUNRISE-1 driving data. [3]

Boxed Warning Absence: What It Signals

Unlike benzodiazepines, lemborexant carries no FDA boxed warning for complex sleep behaviors as a class-wide label change. The FDA issued a boxed warning for all Z-drugs (zolpidem, eszopiclone, zaleplon) in April 2019 covering complex sleep behaviors including sleepwalking and sleep-driving with injury reports. [7] Lemborexant's label does include a precaution for complex sleep behaviors but not a boxed warning, a regulatory distinction that reflects its different spontaneous adverse event report profile at the time of review.

Safety Profile: GRADE-Rated Harm Outcomes

GRADE analysis of harms requires the same rigor applied to efficacy. The most clinically relevant harm signals for lemborexant are somnolence, falls (particularly in older adults), and sleep paralysis/hypnagogic hallucinations associated with orexin pathway modulation.

Somnolence and Falls

In the pooled safety population from SUNRISE-1 and SUNRISE-2 (N=1,222 lemborexant-treated participants), somnolence occurred in 10.2% of the 10 mg group versus 1.5% for placebo and 4.5% for zolpidem ER 6.25 mg. [3] Falls were reported in 2.4% of lemborexant 10 mg users versus 1.5% placebo in the 12-month SUNRISE-2 data, a difference that did not reach statistical significance (P=0.18) but is clinically meaningful for older adult prescribing.

Cataplexy-Like Events

Because OX2R antagonism is mechanistically linked to the narcolepsy-cataplexy pathway, lemborexant labeling includes a warning for cataplexy-like events. These were rare in trials (reported in <1% of participants) but represent a pharmacologically plausible harm that the GRADE framework would classify as low certainty given the low event rate and inability to estimate precise risk from trial data alone. [3]

Comparative Effectiveness: Lemborexant vs. Other DORAs

Suvorexant (Belsomra) was approved in 2014 as the first DORA and provides an indirect comparator. No head-to-head RCT of lemborexant versus suvorexant has been published as of this writing. A 2019 network meta-analysis published in the Journal of Psychiatric Research compared DORAs against each other using indirect evidence and found lemborexant 10 mg numerically superior to suvorexant 20 mg for sleep onset latency (mean difference 4.2 minutes favoring lemborexant), but the credible interval crossed zero, placing GRADE certainty for this comparison at low. [8]

The American Academy of Sleep Medicine (AASM) 2023 clinical practice guidelines on chronic insomnia treatment give a conditional recommendation for DORAs as a class, citing moderate certainty evidence. [9] The guideline authors note: "The balance of benefits and harms favors DORAs over benzodiazepines for most adults with chronic insomnia disorder, primarily on the basis of the residual impairment profile." [9]

Lemborexant in Older Adults: A Specific Subgroup Signal

The SUNRISE-1 enrollment criteria required participants to be aged 55 or older, making it one of the few Phase 3 insomnia trials designed with an older adult primary cohort. That design choice strengthens applicability and reduces the GRADE indirectness downgrade for geriatric prescribing compared to trials that enrolled predominantly younger adults and post-hoc analyzed the older subset.

The Beers Criteria (American Geriatrics Society 2023 update) lists benzodiazepines and Z-drugs as potentially inappropriate medications for older adults but does not include DORAs in the same category, citing their different mechanism and residual impairment profile. [10]

Special Populations and Evidence Gaps

Pregnancy and Lactation

No adequate human data on lemborexant in pregnancy exist. The FDA label assigns no formal pregnancy category under the current labeling system but notes animal reproduction studies showed embryo-fetal toxicity at exposures exceeding the maximum recommended human dose. GRADE certainty for reproductive safety is very low by necessity. [3]

Hepatic and Renal Impairment

Lemborexant is metabolized primarily by CYP3A4. Mild hepatic impairment requires no dose adjustment; moderate impairment calls for a maximum dose of 5 mg. The drug is not recommended in patients with severe hepatic impairment. Renal impairment does not require dose adjustment based on pharmacokinetic sub-studies included in the NDA submission. [3]

Comorbid Psychiatric Disorders

SUNRISE trials excluded participants with active major depressive disorder and bipolar disorder. Post-marketing data have not identified a specific signal in these populations, but the absence of RCT data means GRADE certainty for comorbid psychiatric populations remains very low. Clinicians prescribing in this context should weigh individual benefit-harm based on the available mechanism and safety data rather than extrapolating directly from trial populations.

What a Prescriber Should Take from This GRADE Analysis

High-certainty evidence supports lemborexant's ability to reduce latency to persistent sleep and improve sleep efficiency on PSG versus placebo, particularly in adults aged 55 and older. Moderate-certainty evidence supports durability of subjective sleep improvement over 12 months without rebound insomnia on discontinuation.

The 5 mg dose carries the cleaner next-morning safety profile based on the SUNRISE-1 driving simulation data, making it the preferred starting dose in older adults and in anyone who needs to drive within 9 hours of taking the medication. The 10 mg dose adds approximately 4 minutes of additional PSG-measured sleep-onset benefit but introduces a statistically significant driving impairment signal at 9 hours post-dose.

No RCT data beyond 12 months exist. That evidence gap is the single largest limitation in the current GRADE assessment and is relevant for any patient likely to use a sleep aid chronically. Clinicians should discuss this gap directly with patients and schedule periodic reassessment of continued need, in line with the AASM guideline recommendation for ongoing behavioral review alongside any pharmacotherapy. [9]

For patients starting lemborexant today, the recommended dose is 5 mg immediately before bedtime (no more than once per night), with an option to increase to 10 mg if the 5 mg dose is tolerated but not sufficiently effective, noting the stricter morning driving restriction that applies at the higher dose. [3]