Post hoc analyses were performed in the subgroup of subjects aged ≥ 65 years from Study 304, Study 106, and Study 108 [13, 16, 17]. Each study was conducted in accordance with the principles of the International Conference on Harmonization guidelines for Good Clinical Practice and the Declaration of Helsinki. Trial protocols were approved by the appropriate institutional review boards. All subjects provided written informed consent prior to participation. Detailed study methodologies were previously published [13, 16, 17].

2.1 Study Design2.1.1 Study 304

Study 304 was a 1-month, global, multicenter, randomized, double-blind, PBO-controlled, active-comparator, parallel-group, fixed-dose, phase 3 study in females aged ≥ 55 years and males aged ≥ 65 years who met the criteria for insomnia disorder as defined in Diagnostic and Statistical Manual of Mental Disorders, 5th edition [18] and had evidence of sleep maintenance insomnia. The study was previously reported by Rosenberg et al. [13].

Study 304 included a 2-week PBO run-in period, followed by a 5:5:5:4 randomization to LEM 5 mg (LEM5), LEM 10 mg (LEM10), zolpidem tartrate extended-release 6.25 mg (ZOL), or PBO for 30 days (Online Resource 1A).

2.1.2 Study 106

Study 106 was a single-center, randomized, double-blind, PBO-controlled, active‑comparator, four-period, incomplete crossover study in healthy subjects aged ≥ 21 years as reported previously [16].

Subjects were randomized to four eight-night treatment periods (separated by a washout of ≥ 14 days) (Online Resource 1B). On days 1 and 8 of each treatment period, subjects were administered a bedtime dose of LEM 2.5 mg (not reported here), LEM5, or LEM10. On days 2–7, subjects were instructed to take the study drug before bedtime at home. Zopiclone 7.5 mg, an active control, was taken at bedtime on days 1 and 8 only (with PBO administered on the days in between). In the PBO period, PBO was taken for 8 days. All subjects received PBO and zopiclone and two of the three doses of LEM.

2.1.3 Study 108

Study 108 was a multicenter, single-dose, randomized, double-blind, PBO-controlled, active-comparator, four-period, crossover, phase 1 study in healthy female subjects aged ≥ 55 years and healthy male subjects aged ≥ 65 years. This age range matched that of Study 304. Subjects had regular sleep timing and duration and no insomnia, as previously reported [17].

The randomization phase comprised four 1-day treatment periods with a minimum 14-day washout interval between each (Online Resource 1C). Subjects were randomized to one of four sequences to receive LEM5, LEM10, ZOL, and PBO administered as a single dose within 5 min before bedtime.

2.2 Assessments2.2.1 Sleep Parameters

In Study 304, polysomnograms (PSGs) were performed at baseline (two consecutive PSGs during the single-blind run-in period), after the first two nights (nights 1 and 2), and the last two nights (nights 29 and 30) of treatment to assess latency to persistent sleep (LPS), determined as minutes from lights-off to the first epoch of 20 consecutive epochs of non-wakefulness; wake after sleep onset (WASO), measured as minutes of wake from LPS until lights-on; WASO in the second half of the night (WASO2H), defined as minutes of wake during the interval from 240 min after lights-off until lights-on; and sleep efficiency (SE), defined as the proportion of minutes spent asleep per total time in bed. For the indicated time points, each sleep outcome measure was determined from the average of the paired PSGs during the baseline and treatment period.

2.2.2 Postural Stability

In Studies 304 and 108, postural stability was assessed using an ataxiameter to measure body sway during 60 s in units of 1/3° angle of arc, with higher values indicating more body sway. For the outcome of postural stability, the effect of alcohol on body sway was used as a threshold for indicating impairment [19]. A 7-unit difference between treatments in the change from time-matched baseline was considered clinically meaningful [17]. In Study 304, postural stability was assessed at baseline (mornings after the pair of PSGs) and at the beginning (days 2 and 3) and end (days 30 and 31) of treatment within 5 min of morning awakening (approximately 8 h post dose). In Study 108, postural stability was assessed at baseline and after each treatment, within 5 min of middle-of-the-night awakening (approximately 4 h post dose) and within 5 min of morning awakening (approximately 8 h post dose).

2.2.3 Driving Performance

In Study 106, next-morning driving tests were assessed on days 2 and 9 post treatment using a standardized on-the-road driving test, as described previously [16]. Subjects were asked to drive 100 km for 1 h on a highway. They were instructed to drive with a steady position between the delineated boundaries of the slower (right) traffic lane while maintaining a constant speed of 95 km/h (59 mph). The primary outcome variable was standard deviation of lateral position (SDLP) in centimeters. A clinically meaningful effect was defined as mean change from PBO in SDLP ≥ 2.4 cm [16].

2.2.4 Cognitive Performance

In Studies 304 and 108, cognitive performance was assessed using a computerized cognitive performance assessment battery (CPAB), which comprised nine tasks assessing various aspects of memory and attention. Output variables from the nine tasks were combined to derive scores for four cognitive domains: power of attention (ability to focus and process information), continuity of attention (ability to maintain attention), quality of memory (ability to store and retrieve information), and speed of memory retrieval (time to retrieve stored information). Higher values indicate better continuity of attention and quality of memory (measured in units); lower values indicate better power of attention and speed of memory retrieval (measured in ms). In Study 304, the CPAB was administered at baseline (following the two PSGs after having taken PBO) and at the beginning (days 2 and 3) and end (days 30 and 31) of randomized treatment in the morning, following the postural stability assessment. In Study 108, the CPAB was administered after the postural stability test (starting within 15 min of middle-of-the-night awakening) and within 15 min of morning awakening.

2.2.5 Auditory Awakening Threshold

In Study 108, subjects were awakened via the auditory awakening threshold (AAT) ≥ 4 h after bedtime and (preferably) after 5 consecutive minutes of nonrapid eye movement stage 2 (N2) sleep. AAT was assessed using an audiometer that delivered tones at increasing volumes, from 15 dB up to a maximum of 105 dB [17]. The tones increased until the subject said, “I’m awake.” The decibel level on this response was scored as the AAT. Subjects who did not awaken to the maximum tone were awakened by the technician [17].

2.2.6 Return-to-Sleep Latency

In Study 108, return-to-sleep latency (RSL) was measured after participants completed the middle-of-the-night assessments. Overhead lights were turned off, and the time to return to sleep was measured by PSG. The RSL was defined as the duration in minutes from lights-off after middle-of-the-night assessments to the first epoch of N2, nonrapid eye movement stage 3 (N3), or rapid eye movement sleep.

2.3 Safety

The incidence and severity of adverse events (AEs) and their relatedness to treatment were recorded. Vital signs, electrocardiograms, and laboratory parameters were also evaluated.

2.4 Statistical Analysis

For Study 304, PSG, body sway, and CPAB findings were analyzed for the elderly subgroup, i.e., randomized subjects who were ≥ 65 years of age (received at least one dose of study drug and had at least one post dose primary efficacy measurement). For Study 106, driving performance was analyzed for the subgroup of subjects ≥ 65 years of age who received at least one dose of study drug and who had sufficient pharmacodynamic data to derive at least one SDLP measure. For Study 108, body sway, CPAB, and AAT findings were analyzed for the subgroup of subjects ≥ 65 years of age who had sufficient pharmacodynamic data to derive at least one outcome variable. For all studies, AEs were similarly analyzed for subjects ≥ 65 years of age in the safety analysis set (randomized subjects who received at least one dose of study drug and had at least one post dose safety assessment).

For Study 304, PSG and body sway measures and CPAB findings were analyzed using the mixed-effect model for repeated measurements, with factors for age group (65–75 years and > 75 years), region, treatment, visit (time point), and treatment-by-visit interaction as fixed effects and baseline value as a covariate, and with log transformation for LPS only. For all measures, missing values were not imputed and assumed to be missing at random. For body sway, subjects with extreme values (values outside of the physiological range, n = 2) were excluded.

For Study 106, SDLP (primary outcome) was analyzed using repeated-measures analyses of variance. Secondary endpoints included symmetry analysis of individual changes from PBO in SDLP. McNemar test was used to compare the subjects with a mean difference in SDLP > 2.4 cm (reflecting impairment) to those subjects with a mean difference in SDLP < − 2.4 cm (reflecting improvement).

For Study 108, statistical analyses for RSL included a repeated mixed-effects model, as described previously [17], for the subgroup of subjects ≥ 65 years of age. Mean change from baseline and associated 95% confidence intervals (CIs) were calculated for body sway, AAT, and CPAB domains for 4 h post dose (middle of the night) and 8 h post dose (morning). Statistical significance was taken to be indicated by nonoverlapping 95% CIs. No adjustments were made for multiple comparisons.