Controlled Clinical Trial of Zolpidem for the Treatment of Insomnia Associated With Attention-Deficit/ Hyperactivity Disorder in Children 6 to 17 Years of Age
OBJECTIVE. The goal was to evaluate the hypnotic efficacy of zolpidem at 0.25 mg/kg per day (maximum of 10 mg/day), compared with placebo, in children 6 through 17 years of age who were experiencing insomnia associated with attention-deficit/hyperactivity disorder.
METHODS. An 8-week, North American, multicenter, double-blind, placebo-controlled, parallel-group study was conducted. Patients underwent stratification according to age (6–11 years [N = 111] or 12–17 years [N = 90]) and were assigned randomly to receive treatment with the study drug or placebo (in a 2:1 ratio). The primary efficacy variable was latency to persistent sleep between weeks 3 and 6. Secondary efficacy variables also were assessed, and behavioral and cognitive components of attention-deficit/hyperactivity disorder were monitored. Safety was assessed on the basis of reports of adverse events, abnormal laboratory data, vital signs, and physical examination findings. The potential for next-day residual effects also was assessed.
RESULTS. The baseline-adjusted mean change in latency to persistent sleep at week 4 did not differ significantly between the zolpidem and placebo groups (−20.28 vs −21.27 minutes). However, differences favoring zolpidem were observed for the older age group in Clinical Global Impression scores at weeks 4 and 8. No next-day residual effects of treatment were associated with zolpidem, and no rebound phenomena occurred after treatment discontinuation. Central nervous system and psychiatric disorders were the most-frequent treatment-emergent adverse events (>5%) that were observed more frequently with zolpidem than with placebo; these included dizziness, headache, and hallucinations. Ten (7.4%) patients discontinued zolpidem treatment because of adverse events.
CONCLUSION. Zolpidem at a dose of 0.25 mg/kg per day to a maximum of 10 mg failed to reduce the latency to persistent sleep on polysomnographic recordings after 4 weeks of treatment in children and adolescents 6 through 17 years of age who had attention-deficit/hyperactivity disorder-associated insomnia.
Sleep disturbances in children are common, and they also affect the children's families adversely. Sleep disorders diminish higher-level cognitive functions, flexibility, the ability to reason, and the ability to think abstractly.1 Inadequate sleep has been linked to increased injury rates and negative metabolic sequelae.2,3 Improvement in a child's sleep leads to improvements in parental sleep time, mood, and behavior.4
Although there has been research on sleep hygiene/behavioral therapies, there are no published, randomized, controlled trials of pharmaceutical therapies for pediatric sleep disturbance.5 Despite this lack of safety and efficacy data, several medications are commonly prescribed. Their use is dictated by prescribers’ familiarity and parental acceptance of sleep medications for their children.6,7 Drugs administered for this purpose have not been true hypnotic agents but have been drugs with other therapeutic indications that manifested drowsiness as a side effect.
An increased prevalence of insomnia has been described for children with attention-deficit/hyperactivity disorder (ADHD), regardless of whether they are receiving stimulants.8,9 Such patients are treated frequently with α2-adrenergic receptor agonists such as clonidine, in an off-label manner.6,10
The primary objective of this study was to evaluate the hypnotic efficacy of zolpidem at 0.25 mg/kg per day (maximal dosage of 10 mg/day), compared with placebo, for children 6 to 17 years of age who were experiencing ADHD-associated insomnia. The secondary objectives included evaluations of safety, the potential for residual effects/rebound insomnia after discontinuation of treatment, and the effects of treatment on components of ADHD.
This was a multicenter, stratified (6 to 11 years and 12 to 17 years of age, with imbalanced randomization [2:1]), double-blind, placebo-controlled, parallel-group study conducted in the United States (41 sites). Patients were included if they were 6 to 17 years of age (inclusive), had been diagnosed as having ADHD, as defined on the basis of Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision criteria and clinical interviews, and complained about childhood insomnia, defined as repeated difficulty with sleep initiation or consolidation that occurred despite adequate age and appropriate time and opportunity for sleep (Appendix).11 Patients were required to have latency to persistent sleep (LPS) of >30 minutes, according to baseline polysomnographic results, and a sleep disturbance not attributable to direct physiologic effects of an abused drug or misused prescription medication. Patients underwent stabilization with all therapies, including ADHD therapy, for ≥1 month. Female patients of childbearing potential had negative pregnancy test results and used birth control.
Patients were excluded if they had other sleep disorders diagnosed with baseline polysomnography, other major psychiatric disorders (but not obsessive-compulsive disorder), or a history of substance abuse and/or dependence. Previous adverse experience with zolpidem, use of pharmacologic sleep aids that the patient was unwilling to discontinue, or current use of rifampicin and/or sertraline also disqualified patients. Institutional review boards approved this study. Appropriate written consents were obtained from each patient and/or parent or legal guardian before enrollment in the study. Children provided written assent in accordance with local institutional review board guidelines.
Eligible patients were assigned randomly to receive zolpidem (0.25 mg/kg per day, to a maximum of 10 mg/day) or placebo. The zolpidem dosage was based on an open-label, pharmacokinetic/pharmacodynamic study with 64 children that showed linear kinetics and age-related differences, indicating more-rapid clearance in younger children.12 Investigators recommended a dose of 0.25 mg/kg, prepared as an oral formulation at 2.5 mg/mL. The placebo was matched with respect to color and flavor. The investigational product (IP) was given as a single dose within 30 minutes before bedtime, in the sleep laboratory. The randomization was centralized and generated by the study sponsor.
The study was conducted in 3 segments (Fig 1). Segment A involved screening (up to 21 days), during which no IP was administered. Segment B was an 8-week, double-blind, treatment phase for assessment of efficacy and safety. Segment C was a 7-day follow-up phase without administration of the IP, for assessment of safety and rebound effects. Enrollment lasted from March to August 2006. Patients were scheduled to make 9 site visits, that is, at screening, at baseline (segment A), at weeks 1, 2, 3, 4, 6, and 8 of treatment (segment B), and during discontinuation of treatment at week 9 (segment C).
Efficacy was assessed with polysomnography, which evaluated multiple parameters of sleep, and with the Clinical Global Impression (CGI) scale, which has 2 components, that is, Clinical Global Impression-Severity (CGI-S), assessing the severity of insomnia, and Clinical Global Impression-Improvement (CGI-I), assessing improvement in insomnia. CGI-I was scored from 1 (worsening) to 7 (improvement) at each visit for visits 3 through 9. CGI-S was scored from 1 (normal) to 7 (severe) at each visit for visits 2 through 9. CGI-S scoring may seem to be reversed,13 because convention dictates that a scale's lower value represents a “bad” rating and a higher value a “good” rating. In this study, CGI scoring was completed by the investigator from an interview with the child, on the basis of improvement (CGI-I–child) and severity change (CGI-S–child) from baseline. At the same visits, the investigator completed CGI scoring from an interview with the parent/legal guardian for both improvement and severity (CGI-I–parent/guardian and CGI-S–parent/guardian).
The primary outcome measure was LPS, which was evaluated with polysomnography and was assessed between weeks 3 and 4 or between weeks 4 and 6. The 2 key secondary efficacy end points were CGI-I–child scores at week 4 and CGI-S–child scores. The other secondary criteria included CGI-I–parent/guardian scores, CGI-S–parent/guardian scores, and polysomnographic assessments of wake time after sleep onset (WASO), number of awakenings after sleep onset (NAASO), and total sleep time (TST). These measures were divided by the time in bed, which differed according to age, to obtain the proportion of WASO, the proportion of NAASO, and sleep efficiency. Actigraphic measures of sleep characteristics included LPS and TST and were obtained between weeks 3 and 4. The secondary efficacy measures focusing on ADHD were ADHD Rating Scale-IV and Conners’ Continuous Performance Test II (CPT-II) scores.
Safety assessments were based on adverse event (AE) reports, pertinent abnormal laboratory values, vital signs, and physical findings. Next-day residual effects of treatment were measured by using the Pediatric Daytime Sleepiness Scale. Rebound effects were assessed as the changes in the actigraphically measured LPS and TST from baseline to each of the first 2 nights of the follow-up period. The safety analysis was performed with the total treated population. Data used for assessment of potential residual and rebound effects were analyzed by using an analysis of covariance model.
The primary analysis was based on the change in LPS from baseline to the postbaseline polysomnogram recorded once between weeks 3 and 6 (considered as week 4). Data were analyzed with an intent-to-treat approach. Changes in LPS from baseline were analyzed by using an analysis of covariance model with treatment and age groups as fixed effects and baseline values as the covariate, with a 2-sided significance level of 5%. No interaction was added in the model. The model containing the treatment-age group interaction was explored and provided to support the primary model. If the P value of this interaction term was <10%, then the results according to age group were also provided. The main secondary end points, CGI-I–child and CGI-S–child scores at visit 6 (week 4), were analyzed after adjustment for multiple comparisons. This allowed hierarchical analysis only if the primary variable was significant. Other secondary variables were analyzed for exploratory purposes, with adjusted least-squares (LS) mean differences and 95% confidence intervals (CIs).
The other secondary analyses, for CGI-I, CGI-S, TST/sleep efficiency, WASO, NAASO, and actigraphic measures (LPS and TST), were performed similarly to the LPS analyses except for the CGI-I analyses. The CGI-I analyses were performed as analyses of variance, because no baseline values were assessed for this parameter; it is a measure of change during the treatment period versus the pretreatment period. The treatment-age group interaction was examined by using a model that included treatment and age groups as fixed factors plus the interaction of age group and treatment group. Analyses according to age group were performed with the model including the age group-treatment group interaction only if this interaction term was significant at the 10% level. Analyses of LPS changes from baseline according to treatment and age groups were performed by using LS means. Descriptions of CGI changes from baseline according to treatment and age groups were provided by using baseline-adjusted means. Robustness was assessed by using an analysis of variance model with 2 fixed factors (treatment and age group), using rank values instead of raw values for the change in LPS, compared with baseline.
For ADHD analyses with ADHD Rating Scale-IV, missing items were replaced by the means of the nonmissing items if ≤30% of the items were missing; otherwise, the total score was set to “missing.” Changes from baseline to weeks 4 and 8 for the ADHD Rating Scale-IV total score were analyzed with the same analysis of covariance model as for the primary analysis of polysomnographically measured LPS. For the CPT-II results, changes from baseline for numbers of commissions and omissions and reaction time were derived for visits 6, 8, and 9 and were analyzed with an analysis of variance model based on rank values with 2 fixed factors (treatment and age group).
Determination of Sample Size
Sample size was determined on the basis of the ability to detect a difference of 20 minutes (SD: 35 minutes) in polysomnographically measured LPS between treatment groups for the change from baseline to assessment between weeks 3 and 4 with a power of 90%, a type 1 error of 5%, and a 2-sided test. Accounting for a dropout rate of ∼20%, this study was planned to include 189 patients (126 treated with zolpidem and 63 treated with placebo), resulting in ≥100 patients being exposed to zolpidem and 50 to placebo for 8 weeks.
After initial screening of 529 patients, 201 were recruited and randomly assigned. Of these, 136 were allocated to receive zolpidem and 65 placebo. No patients were lost to follow-up monitoring. Fifteen patients in the zolpidem group and 8 patients in the placebo group discontinued treatment. Reasons for discontinuation in the zolpidem group included AEs (10 patients), lack of efficacy (4 patients), and poor compliance (1 patient). The only reason for discontinuation in the placebo group was patient request (8 patients). Patient dispositions are presented in Fig 2.
Baseline demographic features were similar (Table 1). A history of ADHD was reported by 127 patients (93.4%) in the zolpidem group and 61 patients (93.8%) in the placebo group. The mean durations of ADHD from the time of diagnosis were >6 years for both groups; the mean severities of ADHD were equivalent. There were no major differences in baseline sleep measures, including LPS, TST, WASO, and NAASO, or the severity of insomnia, as measured with CGI-S–child and CGI-S–parent/guardian scores.
Nonhypnotic medications were used for sleep by 41 patients (30.1%) in the zolpidem group and 23 patients (35.4%) in the placebo group. In the zolpidem group, clonidine (22 patients [16.2%]) was used most frequently; in the placebo group, drugs with other indications that manifested drowsiness as a side effect (9 patients [13.8%]) and antihistamines (8 patients [12.3%]) were used. The most frequently reported ADHD medications were psychostimulants in both the zolpidem group (123 patients [90.4%]) and the placebo group (61 patients [93.8%]).
No significant difference between treatment groups in LPS at week 4 was detected. The LS mean reduction from baseline was −20.28 minutes with zolpidem and −21.27 minutes with placebo. Baseline-adjusted mean changes at week 4 for sleep efficiency, proportions of WASO, and proportions of NAASO did not differ significantly (Table 2). For actigraphic measures at week 4, baseline-adjusted means for TST and LPS did not differ significantly between groups (Table 2).
On the basis of CGI-I–child assessments, the zolpidem group showed greater improvement, compared with the placebo group, at week 4 (LS mean difference: 0.4 [95% CI: 0.05–0.85]) (Table 3). A treatment-age interaction was observed (Table 3). At weeks 4 and 8, the mean CGI-I–child scores were greater with zolpidem for the 12- to-17-year age group but not the 6- to 11-year age group.
For CGI-S–child scores at week 4, the baseline-adjusted mean decrease was greater for the zolpidem group (LS mean difference: −0.64 [95% CI: −1.095 to −0.187]) (Table 2). At week 4, CGI-S–child scores with the treatment-age interaction model were similar to the CGI-I–child scores. At week 8, baseline-adjusted mean improvement in CGI-S–child scores was greater in the zolpidem group for the 12- to 17-year age group (LS mean difference: −1.57 [95% CI: −2.284 to −0.852]) but not for the 6- to 11-year age group (LS mean difference: 0.08 [95% CI: −0.559 to 0.729]).
Mean CGI-I–parent/guardian scores at week 4 did not differ (LS mean difference: 0.4 [95% CI: −0.04 to 0.77]) (Table 2). In contrast, baseline-adjusted mean CGI-S–parent/guardian scores were improved with zolpidem (LS mean difference: −0.55 [95% CI: −1.010 to −0.088]). Treatment-age interactions were observed for both of these variables. For CGI-I and CGI-S variables at weeks 4 and 8, mean values showed a greater improvement with zolpidem for the 12- to 17-year age group but not for the 6- to 11-year age group.
Baseline-adjusted mean changes in ADHD Rating Scale-IV scores and CPT-II omission and commission errors at weeks 4 and 8 did not differ between groups. The median change versus baseline for the average reaction time in the CPT-II showed a greater increase with zolpidem in the intention-to-treat population at week 4 (median: placebo: 0 milliseconds; zolpidem: 10 milliseconds; P = .0282) but not at week 8 (median difference: placebo: 8 milliseconds; zolpidem: 17 milliseconds; P = .5162). Analyses of actigraphic measures confirmed the results observed with polysomnographic sleep parameters.
A total of 201 patients were exposed to IP. The median exposure time was 56 days for both groups, with 62.5% of zolpidem-treated patients and 47.7% of placebo-treated patients experiencing ≥1 treatment-emergent AE, including dizziness (zolpidem: 23.5%; placebo: 1.5%), headache (zolpidem: 12.5%; placebo: 9.2%), and hallucinations (zolpidem: 7.4%; placebo: 0.0%) (Table 4).
IP administration was discontinued permanently because of treatment-emergent AEs for 10 patients (7.4%) in the zolpidem group, compared with 0 patients (0.0%) in the placebo group (Table 5). The AEs were mostly psychiatric (6 patients [4.4%]) or central nervous system related (2 patients [1.5%]). The main AE leading to discontinuation was hallucination (5 severe hallucination episodes involving 4 patients). Among the 136 zolpidem-treated patients, 10 (7.4%) experienced 13 hallucinatory episodes. Of these 13 episodes, 8 were mild or moderate and 5 were severe. One patient in the placebo group experienced a serious AE (impulse control disorder); no serious AEs were observed with zolpidem. There were no deaths.
The Pediatric Daytime Sleepiness Scale baseline-adjusted mean decrease at end of treatment was greater with zolpidem, but this difference was not statistically significant (P = .0567). Both groups experienced worsening from baseline in LPS on postdiscontinuation nights 1 and 2, but the difference was not significant. Similarly, TST on postdiscontinuation night 1 was lower than baseline values for both groups. The effect was not observed on night 2, and the primary efficacy measure did not differ significantly in the groups.
The primary outcome measure of LPS did not differ between zolpidem-treated and placebo-treated patients after 4 weeks (LS mean: −20.28 vs −21.27 minutes). Analyses of the secondary sleep measures (measured with polysomnography and actigraphy) at week 4 also showed no difference between groups, but CGI scores based on interviews indicated favorable effects of zolpidem in adolescents.
Insufficient dosing, on the basis of age, could account in part for the differences in the older group for CGI-child and CGI–parent/guardian scores at weeks 4 and 8. Pharmacokinetic data show that zolpidem is metabolized rapidly in children.12 An alternative explanation for the results is that insomnia in pediatric patients with ADHD may be a different neurophysiologic disorder than insomnia in adults, for whom zolpidem and similar hypnotic agents are indicated for relief. The difference between LPS measured in a sleep laboratory and the clinical report of difficulty falling asleep (clinical LPS) also must be considered in the interpretation of these results. The age group-dependent differences in CGI scores may reflect age-associated differences in zolpidem pharmacokinetics,12 γ-aminobutyric acid neurotransmission, or both. Alternatively, they could reflect an age-associated difference in the ability of this type of assessment instrument to detect change.
Zolpidem had neither a positive nor a negative effect on ADHD severity after 4 and 8 weeks of treatment. This lack of positive effect was not surprising, because the intervention was not designed to improve core ADHD symptoms; there also was no significant improvement in insomnia.
The proportion of patients who experienced ≥1 treatment-emergent AE was higher with zolpidem, although the majority of AEs were mild to moderate. These findings are consistent with the safety and tolerability profile obtained from the pharmacokinetic study.12
The main AEs leading to discontinuation of zolpidem administration were psychiatric and occurred more frequently in the younger group. Hallucinations, which led to the withdrawal of 4 patients, occurred within 15 to 30 minutes after dosing, occurred more prominently among boys, and occurred among patients receiving a lower dose of medication without a history of hallucinations. Full recovery occurred within 1 day.
No next-day residual effects were noted, as measured with the Pediatric Daytime Sleepiness Scale, consistent with data from adult studies.14,15 Rebound worsening in LPS on postdiscontinuation nights 1 and 2 and in TST on postdiscontinuation night 1, compared with baseline, was observed in both groups, without significant difference. This differs from reports of placebo-controlled studies of zolpidem in adults,14,15 in which sleep worsened relative to measurements recorded during treatment but not baseline values. The lack of difference between the drug and placebo groups suggests that this might have been a nonspecific effect of study drug discontinuation and was not zolpidem specific. The potential for drug interactions should be noted. Responses to zolpidem for children who were receiving different medications for insomnia and/or ADHD before the study were not evaluated.
Zolpidem administered at a dose of 0.25 mg/kg per day (to a maximum of 10 mg/day) did not reduce the latency of sleep onset significantly, on the basis of polysomnographic and actigraphic recordings after 4 weeks, in children and adolescents 6 through 17 years of age who were experiencing ADHD-associated insomnia. Neuropsychiatric symptoms constituted the most-frequent treatment-emergent AEs (noted for >5% of patients) observed with zolpidem, compared with placebo, and included dizziness, headache, and hallucinations. Favorable results in subjective measures of hypnotic efficacy in the older age group suggest that such patients may respond to zolpidem differently from younger patients. The severity of insomnia in a subset of pediatric patients with a lack of response to behavioral interventions demands further study of the pharmacotherapy of insomnia in children and adolescents, particularly patients with comorbid ADHD and oppositional defiant disorder. Head-to-head trials against currently used agents may be warranted, to determine whether various subgroups of patients respond differently.
APPENDIX: PROPOSED DEFINITION OF PEDIATRIC INSOMNIA11
The following criteria, adapted from the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision, attempt to take into account the unique aspects of pediatric insomnia, including normal developmental changes in sleep habits, the fact that caregivers rather than the insomnia-sufferer typically voice the complaint of insomnia, and the potential effects of childhood insomnia on the family. The criteria are as follows. (1) The complaint is significant difficulty (defined on the basis of frequency, severity, and/or chronicity) initiating or maintaining sleep. The difficulty is viewed as problematic by the child and/or a caregiver. (2) The sleep disturbance causes clinically significant impairment in school performance, behavior, mood, learning, or development for the child, as reported by the child and/or a caregiver. (3) The sleep disturbance does not occur exclusively in the context of an intrinsic dyssomnia such as narcolepsy, restless leg syndrome, or sleep-related breathing disorders, a circadian rhythm disorder, or a parasomnia. (4) The sleep disturbance is not attributable to either the direct physiologic effect of a drug or the abuse or misuse of prescribed medications.
Dr Reed has received research support from the National Institute of Child Health and Human Development, the State of Ohio Department of Health, and the US Health Resources and Services Administration.
- Accepted January 12, 2009.
- Address correspondence to Jeffrey L. Blumer, PhD, MD, Pediatric Pharmacology/Critical Care, Rainbow Babies and Children's Hospital, 11100 Euclid Ave, Cleveland, OH 44106. E-mail:
Financial Disclosure: Dr Blumer has received research support from, acted as a consultant for, and/or served on a speakers bureau for Abbott, AstraZeneca, Bristol-Myers Squibb, Cubist, GlaxoSmithKline, MedImmune, Merck, Novartis, Pfizer, Roche, sanofi-aventis, Sepracor, and Wyeth; Dr Findling receives or has received research support from, acted as a consultant for, and/or served on a speakers bureau for Abbott, Addrenex, AstraZeneca, Bristol-Myers Squibb, Forest Laboratories, GlaxoSmithKline, Johnson & Johnson, Eli Lilly, Neuropharm, Novartis, Organon, Otsuka, Pfizer, sanofi-aventis, Sepracor, Shire, Solvay, Supernus Pharmaceuticals, Validus, and Wyeth; Dr Shih has received research support from sanofi-aventis as a member of the independent data safety monitoring committee for the pediatric zolpidem study on which this work is based; Dr Soubrane is currently an employee of sanofi-aventis; and Dr Reed has received research support from, served as a consultant for, and/or served on a speakers bureau for Abbott, Astellas, AstraZeneca, Bayer, Bristol-Myers Squibb, Eli Lilly, Enzon, Forest Laboratories, GlaxoSmithKline, Janssen, Johnson & Johnson, Merck, Novartis, Organon, Pfizer, Roche, Sankyo, sanofi-aventis, Somerset, UCB Pharma, and Wyeth-Ayerst.
What's Known on This Subject
Sleep disorders are common in children, affecting family interactions, quality of life, and school performance. Behavioral therapies have yielded inconsistent responses. To date, no drugs have been evaluated rigorously for safety and effectiveness as soporific agents for these patients.
What This Study Adds
This study provides the first evaluation of the effectiveness and safety of zolpidem, a nonbenzodiazepine hypnotic agent, in children and adolescents with insomnia associated with ADHD. The dose evaluated was chosen after careful pharmacokinetic study in children and adolescents.
- ↵Valent F, Brusaferro S, Barbone F. A case-crossover study of sleep and childhood injury. Pediatrics.2001;107 (2). Available at: www.pediatrics.org/cgi/content/full/107/2/e23
- ↵Mindell JA, Durand VM. Treatment of childhood sleep disorders: generalization across disorders and effects on family members. J Pediatr Psychol.1993;18 (6):731– 750
- ↵Mindell JA, Emslie G, Blumer J, et al. Pharmacologic management of insomnia in children and adolescents: consensus statement. Pediatrics.2006;117 (6). Available at: www.pediatrics.org/cgi/content/full/117/6/e1223
- ↵Owens JA, Rosen CL, Mindell JA. Medication use in the treatment of pediatric insomnia: results of a survey of community-based pediatricians. Pediatrics.2003;111 (5). Available at: www.pediatrics.org/cgi/content/full/111/5/e628
- ↵Ring A, Stein D, Barak Y, et al. Sleep disturbances in children with attention-deficit/hyperactivity disorder: a comparative study with healthy siblings. J Learn Disabil.1998;31 (6):572– 578
- Copyright © 2009 by the American Academy of Pediatrics