Authors for correspondence: Naohisa Uchimura, Kurume University School of Medicine, 67 Asahi-Machi, Kurume 830-0011, Japan (fax 0942 35 6041, e-mail email@example.com). Eric P. Prinssen, F. Hoffmann-La Roche Ltd, Building 72/148, Grenzacherstr. 124, CH-4070 Basel, Switzerland (fax +41 61 688 1895, e-mail firstname.lastname@example.org).
Abstract: Influenza-associated neuropsychiatric symptoms include parasomnias such as sleepwalking which is a common sleep disturbance in childhood. Oseltamivir is a widely used antiviral drug for influenza. Recently, sleepwalking-like events have been reported in patients with influenza receiving oseltamivir. We investigated whether oseltamivir itself has effects on sleep. In this crossover study, healthy Japanese male volunteers were randomized into two treatment groups, each of which comprised two double-blind 4-day treatment periods. In the first period, group A received 75 mg oseltamivir (evening dose) on day 3, followed by 75 mg b.i.d. on day 4, and placebo in the second period. Group B received the same treatments, but in reverse order. Polysomnographic assessments were performed on all four nights of each treatment period. Pharmacokinetics were assessed during a 2-day open-label phase beginning on day 12. Thirty-one volunteers aged 20–24 years were enrolled. No volunteer had electroencephalographic abnormalities, and no abnormal behaviour was observed. Sleep parameters measured over the whole night and during early- and late sleep periods (first and last thirds of the night) were very similar for oseltamivir and placebo, although the amount of stage 2 sleep in the middle sleep period was slightly greater with oseltamivir. Pharmacokinetics for oseltamivir phosphate in groups A and B were very similar, but for oseltamivir carboxylate, AUC and Cmax values were higher in group B, probably because this group received oseltamivir on the evening of day 11. Oseltamivir was well tolerated. Oseltamivir did not produce clinically relevant changes on nocturnal polysomnographic variables in young Japanese men.
Oseltamivir (Tamiflu®; Roche) is an oral neuraminidase inhibitor that is readily absorbed from the gastrointestinal tract and subsequently converted by hepatic carboxylases from the phosphate prodrug to its active carboxylate form . Oseltamivir is indicated for the treatment and prophylaxis of influenza [2,3] and is used widely around the world. During the 2005–06 and 2006–07 influenza seasons, increased numbers of neuropsychiatric events were reported in patients with influenza with and those without exposure to oseltamivir, particularly in Japan. These events were uncommon – the rate of events from spontaneous reporting was estimated at 99 per million in children and adolescents aged ≤16 years  – but nevertheless provoked discussion relating to the central nervous system (CNS) tolerability of the drug [5,6]. Ongoing pharmacovigilance by Roche , reports from the Japanese Ministry of Health, Labour and Welfare (MHLW) for the 2005–06 and 2006–07 seasons [7,8] and examinations of US health claims databases [4,9–11] indicated no causal relationship between oseltamivir and neuropsychiatric events. Moreover, reports of low levels of penetration of oseltamivir into the CNS [12,13] and lack of interaction of the drug with molecular targets relevant to mood, cognition and behaviour  indicate that oseltamivir has limited or no potential for induction or exacerbation of CNS adverse events (AEs) and that such events are most likely linked to influenza itself .
One aspect of CNS function that remains unexplored in persons receiving oseltamivir is sleep. Individuals with sleep disorders, especially those of arousal, may display symptoms that resemble the neuropsychiatric events that have been reported in patients with influenza with and without exposure to oseltamivir [15,16]. As sleep patterns have not been studied in patients taking oseltamivir, the present trial was undertaken to explore whether sleep abnormalities or related effects possibly leading to arousal disorders occur in persons taking this agent.
Study design. This study was carried out at three sites in Japan: Kitasato University East Hospital Clinical Trial Center, Sagamihara, Kanagawa; the National Institute of Mental Health, National Center of Neurology and Psychiatry, Kodaira, Tokyo, and the Japan Somnology Center, Neuropsychiatric Research Institute, Shibuya, Tokyo. The study followed a placebo-controlled, randomized, crossover design with two double-blind 4-day treatment periods each featuring nightly polysomnography (PSG) and ended with an open-label treatment stage for pharmacokinetic assessment. Safety was also assessed. The study was carried out in accordance with the provisions of the Declaration of Helsinki, relevant clauses of the Japanese Pharmaceutical Affairs Law and Good Clinical Practice. The study was approved by ethical review boards at each centre. Volunteers provided written and fully informed consent before enrolment.
Study population. Healthy male volunteers aged 20–24 years and with a body mass index of 17.6–26.4 kg/m2 were screened for eligibility up to 28 days before the start of the study. The main exclusion criteria were the following: sleep disorder in the preceding 6 weeks; a history of sleepwalking or night terrors (including family history) or a history of convulsions, sleep apnoea or restless legs syndrome; an irregular sleep pattern; a history of addiction to or dependency on drugs or alcohol or any neuropsychiatric disorder or other medical condition that might affect study outcomes; abnormal electroencephalogram (EEG) at screening; regular use of hypnotic or sedative drugs or use of psychotropic drugs in the preceding 5 weeks; and use of the uricosuric agent probenecid.
Treatments. Volunteers eligible to enter the study were randomly allocated to one of the two treatment sequences (groups A and B) both of which consisted of two treatment periods (period 1, days 1–5; period 2, days 8–11) separated by a 2-day drug-free period. The full schedule is shown in table 1. Treatment was double-blinded and placebo-controlled for periods 1 and 2. Allocation to groups was based on a pre-determined table held by the study administrator. Group A received oseltamivir in period 1 and placebo in period 2, and group B received the same treatments but in reverse order; all the treatments were given orally. Oseltamivir was only given on the third and fourth days of each treatment period, as a single dose of 75 mg on the evening of the third day and 75 mg twice daily on the fourth day. Placebo was given on the second day (morning and evening doses) and the morning of the third day in both periods, and no treatments were given on the first day of both periods. Period 1 (only) included one additional no-treatment day (day 5). On day 12, all volunteers received oseltamivir 75 mg as an open-label single dose for 24-hr pharmacokinetic assessment (which ended on day 13). Volunteers made one follow-up visit on days 15, 16 or 17.
Table 1. Treatment schedule. Dosing during both double-blind periods was twice daily and for open-label was once daily.
Treatment period 1 (double-blind)
Treatment period 2 (double-blind)
O, oseltamivir; P, placebo.
1Oseltamivir; 75-mg capsule, evening (placebo capsule in morning).
2Oseltamivir; 75 mg morning and evening.
3Oseltamivir; 75-mg capsule, morning (nothing in evening).
Group A (n = 16)
Group B (n = 15)
The volunteers remained under investigators’ observation at the study sites from day 1 to day 13 to monitor safety and pharmacokinetics, but were allowed to return home on days 6 and 7 if investigators found no problems on day 5. During this period, meals were provided at scheduled intervals and drinks containing caffeine or alcohol were prohibited, as was grapefruit or its juice. Strenuous exercise was also prohibited.
Procedures and assessments. At screening, medical history and baseline characteristics were recorded, including vital signs, EEG, standard 12-lead electrocardiogram (ECG), laboratory tests (haematology, blood chemistry and urinalysis), and indicators of renal function (creatinine clearance). Vital signs and AEs were monitored daily throughout the study, ECG was measured on days 1, 4, 5, 8 and 11, and laboratory tests on days 1, 3, 8, 10 and 13; all these assessments were repeated at the follow-up visit.
Polysomnography was carried out during the first and second double-blind periods (on the evenings of days 1–4 and 8–11). This included EEGs, ECGs, electromyograms of mentalis and anterior tibial muscles and assessments of eye movement. PSG scoring was evaluated with the criteria developed by Rechtschaffen and Kales , and arousal was evaluated according to the criteria of the American Sleep Disorders Association . The principal PSG measures were total sleep time, sleep efficiency (ratio of total sleep time to time in bed), sleep latency, number and duration of nocturnal wakings, time spent awake after sleep onset, respiratory rate during sleep and the amount of time spent in each sleep stage. Sleep stage measures were evaluated for the whole night and for early-, middle- and late sleep periods (total sleep period for each volunteer divided into three parts of equal duration). All volunteers were monitored via video monitors at each site to capture any abnormal sleep behaviours such as sleep walking and night terrors, and video recordings were synchronized with PSG data. Study sites submitted all video and PSG data to an EEG analysis specialist (HK) for qualitative assessment, and all results were reviewed by a second assessor (NU).
Pharmacokinetics were determined on day 12 (before dosing and up to 12 hr after dosing) and day 13 (24 hr after dosing), with blood samples being handled and analysed by a single independent laboratory (BASi Analytics Ltd, Kenilworth, UK). Primary parameters were the area under the curve of plasma concentration versus time extrapolated to infinity (AUCinf) and peak plasma level (Cmax). Secondary parameters were AUC to the last measured concentration (AUClast), time to Cmax (tmax), elimination half-life (t1/2), the elimination rate constant (kel), apparent clearance (CL/F), CL/F adjusted for body-weight (CL/F/BW), mean residence time (MRTinf) and apparent volume of distribution (Vd/F).
Safety was evaluated via recording of AEs, laboratory testing and physical and vital signs. AEs were defined as all undesirable or unintentional signs, symptoms or diseases and were classified according to preferred terms based on Medical Dictionary for Regulatory Activities (MedDRA; International Federation of Pharmaceutical Manufacturers and Associations (IFPMA) Geneva, Switzerland) Version 10.0, while laboratory test abnormalities were defined as those that deviated from normal values at the study sites. Serious AEs (SAEs) were defined as those resulting in death or morbidity or that caused hospitalization. AEs were classified as unrelated, slightly related, possibly related or probably related to study medication and as mild, moderate or severe, as judged by the investigator.
Data handling. Quality of data was assured by standard operating and audit procedures. The target sample size of 30 was calculated to confer a power of ≥90% at a two-sided significance level of 5% by McNemar’s test and was based on a projected 30% difference between oseltamivir and placebo in incidence of abnormal behaviours during sleep and a 33% inconsistency in response ratio.
The three pre-specified analysis populations were the per-protocol set of persons with no serious deviations from the study protocol, the safety set of all those who received at least one dose of study medication, and the pharmacokinetic analysis set of those who received study medication with no serious deviation from protocol and who yielded a measurement of oseltamivir phosphate in plasma. No compensatory methods were used to account for missing data. Summary statistics for each sleep parameter were calculated for each day of the study, and mean values calculated from days 3 and 4 and days 10 and 11. Analysis of variance (anova) was performed with a mixed model for sleep parameters, and the differences in adjusted mean values between active drug and placebo with 95% confidence intervals (CIs) were determined. AEs and laboratory test results were analysed descriptively. Plasma levels of oseltamivir below the level of quantification were reported as zero.
In total, 31 volunteers were enrolled, 16 in group A (oseltamivir in period 1 and placebo in period 2) and 15 in group B (placebo in period 1 and oseltamivir in period 2). There were no withdrawals, and all 31 persons were included in the per-protocol, safety and pharmacokinetic populations. Both groups were demographically similar (table 2).
Table 2. Demographic data by randomized groups at baseline.
Group A (n = 16)
Group B (n = 15)
S.D., standard deviation.
Age (mean ± S.D., years)
21.1 ± 0.9
21.7 ± 1.1
Height (mean ± S.D., cm)
173.6 ± 7.0
175.2 ± 6.3
Weight (mean ± S.D., kg)
64.3 ± 8.9
64.9 ± 6.5
Body mass index (mean ± S.D., kg/m2)
21.3 ± 1.7
21.2 ± 2.2
Summary statistics for total night sleep parameters are shown in table S1, and least square mean differences between placebo and oseltamivir (with 95% CIs) for total night’s sleep and for each of the three sleep periods (early, middle and late) are illustrated in fig. 1. For the total night’s sleep parameters, there were no differences between treatments in the amounts or proportion of any sleep stage, and all 95% CIs for estimated treatment difference crossed zero. When sleep parameters were assessed separately for each sleep period, no differences between oseltamivir and placebo were found, except for the amount and proportion of stage 2 sleep in the middle period, which were both greater in volunteers treated with oseltamivir (95% CIs did not cross zero) (fig. 1).
For oseltamivir phosphate (parent drug), pharmacokinetic parameters on days 12 and 13 in groups A and B were very similar. For oseltamivir carboxylate (active metabolite), AUC and Cmax values were higher in group B than in group A, probably because volunteers in group B received active drug during the second double-blind period (days 10 and 11), 1 day before pharmacokinetics was determined. Overall, tmax of parent drug ranged from 1–4 hr and t½ from 1.2–8.9 hr; corresponding ranges for the carboxylate metabolite were 3–7 hr and 4.5–10.3 hr. For parent drug, clearance values ranged from 310–844 l/hr, the apparent volume of distribution was 733–4180 l, and the mean residence time was 2.4–6.4 hr. Pharmacokinetic characteristics of the parent drug and metabolite are summarized in table 3 for groups A and B.
Table 3. Mean pharmacokinetic parameters (with standard deviations).
Group A (n = 16)
Group B (n = 15)
AUCinf, area under the curve of plasma concentration versus time extrapolated to infinity; AUClast, AUC to the last measured concentration; CL/F, apparent clearance; CL/F/BW, CL/F adjusted for body-weight; Cmax, peak plasma level; kel, elimination rate constant; MRTinf, mean residence time; tmax, time to Cmax; t½, elimination half-life; Vd/F, apparent volume of distribution.
Oseltamivir phosphate (parent drug)
170.7 ± 44.8
173.0 ± 39.6
62.3 ± 24.2
61.7 ± 19.2
167.0 ± 44.4
167.9 ± 37.9
2.3 ± 0.7
2.0 ± 0.8
1.9 ± 0.6
2.4 ± 1.9
0.39 ± 0.12
0.38 ± 0.15
471.5 ± 136.0
461.2 ± 136.6
7.3 ± 1.6
7.0 ± 1.6
3.7 ± 0.5
3.8 ± 1.0
1279.7 ± 479.3
1475.5 ± 864.6
Oseltamivir carboxylate (active metabolite)
2973.8 ± 528.0
4653.5 ± 1085.7
242.0 ± 43.3
346.3 ± 51.4
2585.1 ± 389.5
4077.6 ± 882.3
5.1 ± 0.9
4.3 ± 0.7
6.9 ± 1.5
7.4 ± 1.2
0.10 ± 0.02
0.10 ± 0.01
13.0 ± 2.3
11.9 ± 1.6
Tolerability and sleep behaviour.
There were 25 AEs in 15 of the 31 volunteers, none of which had a definite causal relationship to oseltamivir treatment as judged by the investigator. All AEs were classified as mild; there were no SAEs or deaths and no noteworthy changes in vital signs. The system organ classes with two or more events were ‘investigations’, ‘skin and subcutaneous tissue disorders’ and ‘musculoskeletal and connective tissue disorders’; the events are described by preferred term in table 4. All three skin disorders were reactions to the EEG electrodes or to the fixing tape. No volunteer had abnormal EEGs while awake or asleep, and no abnormal behaviour was seen during sleep periods. One volunteer reported waking during the night (MedDRA preferred term: middle insomnia), during treatment with oseltamivir, but there were no associated abnormal EEG findings, and no abnormal behaviours were observed by video monitoring. Oseltamivir had no effects on respiratory rate during sleep or on 12-lead ECGs taken on arousal from sleep. Scatter diagrams with coefficients of correlation showed a statistically significant negative correlation between creatinine clearance and AUCinf of active metabolite (group A: −0.652, p =0.006; group B: −0.663, p =0.007).
Volunteer numbering was by centre: Kitasato University East Hospital (1–15), National Centre of Neurology and Psychiatry (17–21, 23–30) and Japan Somology Centre (33–35).
AEs were classified according to the preferred terms with the Medical Dictionary for Regulatory Activities (MedDRA) Version 10.0, while laboratory test abnormalities were defined as those that deviated from normal values at the study sites.
In this double-blind, crossover study, oseltamivir had virtually no discernible effects on sleep parameters or night-time behaviour in young Japanese men, either qualitatively or quantitatively, as shown by polysomnographic testing. There was no effect on the amount and proportion of each sleep stage when this was assessed for the total night or when each of the three sleep periods were assessed separately. One exception to this was the amount and proportion of stage 2 sleep in the middle sleep period, which was slightly greater with oseltamivir than with placebo. This apparent difference is unlikely to have been a treatment-related effect, given the absence of any other treatment differences in this sleep period and a lack of any effects on stage 2 sleep in other sleep periods. Moreover, this difference was not associated with any abnormal behaviour or AEs. The lack of marked changes in sleep in our study is consistent with the results of an animal study, in which supratherapeutic doses of oseltamivir had no effect on hexobarbital-induced sleeping time in mice .
Parasomnias can result in complex behaviours during sleep, the most common of which are arousal disorders and rapid eye movement (REM) sleep behaviour disorder [15,19]. Of particular interest in the current context are arousal disorders, including sleepwalking and night terrors, which may result in suicide or potentially violent or injurious behaviours [19,20]. Arousal disorders are seen during the first hour or two of the night after an abrupt arousal from slow-wave sleep . In the present study, oseltamivir caused no changes relative to placebo in the proportion of time spent in any of the sleep stages, including deep sleep and REM sleep, or in the number of wakings or the total time spent awake. In a recent review of 3051 neuropsychiatric events (NPAEs) in patients with influenza, spontaneously reported to Roche over an 8-year period , parasomnias were reported, although they were not common (2.4%); injuries and accidents (1.1%), suicidal ideation and suicidal events (0.5%) and panic attack (0.4%) were also reported . The review found no evidence that any NPAEs in patients with influenza were more common in those treated with oseltamivir than in untreated patients, suggesting that their aetiology may be influenza-related.
There is a body of evidence linking influenza itself with disturbance of sleep and EEG activity. Influenza has been shown to be associated with changes in murine EEG patterns , and the cytokine-mediated acute phase response to influenza in mice, typified by somnolence and temperature change, is induced by the recognition of viral double-stranded RNA by toll-like receptors . Kleine–Levin syndrome, a recognized, albeit rare, complication of influenza, is characterized by hypersomnolence and disturbances of cognition and behaviour . In children with influenza, abnormal EEG patterns such as focal slowing appear to be closely associated with neurological complications such as delirious behaviour and febrile seizures [25,26] and visual hallucinations . Consequently, the delirious behaviour, hallucinations and other neuropsychiatric events that have been reported in patients with influenza  may be part of the acute phase response to influenza infection linked to sleep disturbance and abnormal EEG. Indeed, the review of NPAEs found that delirium and delirium-type symptoms, which were associated with 89% of all the events reported, occurred most often during the first 2 days of influenza. This coincides with the recommended treatment time window for oseltamivir so it is difficult to differentiate between the effects of disease and treatment . However, together with a large body of evidence , the absence of any effects of oseltamivir on sleep and nocturnal EEG activity in the current study suggests that an effect of oseltamivir on sleep is not a feasible mechanism for producing delirium-like events in patients with influenza.
The tolerability profile of oseltamivir in this study was consistent with previous observations, and no new safety signals were identified. All AEs were mild, and none were judged by investigators to be ‘probably’ related to oseltamivir treatment. A single case of nocturnal waking, although deemed ‘possibly’ associated with oseltamivir treatment, was not associated with any EEG changes. There were no changes in vital signs (including cardiac parameters) and no effects on respiratory rate during sleep. These findings are concordant with collated data showing no important safety concerns that might limit the use of oseltamivir in any patient group [4,29]. There were no unexpected pharmacokinetic findings in the present trial, with any apparent changes in exposure to oseltamivir being linked to the timing of pharmacokinetic sampling relative to the crossover treatment sequence. The pharmacokinetics of oseltamivir have already been shown to be similar in both Japanese and Caucasian individuals , and the present results suggest no grounds for any change in this view. Plasma levels in healthy volunteers and infected patients are also very similar , and recent case studies have reported very low penetration of oseltamivir and oseltamivir carboxylate into the CNS in healthy individuals  and infected patients [31,32]. It is therefore likely that the findings of the current study can be extrapolated from healthy volunteers to patients with influenza.
In conclusion, the present study shows an absence of adverse effects of oseltamivir on nocturnal EEG, sleep or sleep-related behaviour in young Japanese men. Pharmacokinetics were similar to that seen in other studies to date, and there were no new or unexpected safety signals. Oseltamivir is thus considered unlikely to be associated with sleep disturbance or with parasomnias, in accord with the findings of other studies that show very low CNS exposure to the drug and a failure to establish any link with neuropsychiatric effects generally.
The authors acknowledge the support of the following sub-investigators who assisted with this study: Yoshio Otani and Yasuhiko Ikeda (Kitasato University East Hospital Clinical Trial Center), Miho Murata and Hirokuni Tagaya (National Institute of Mental Health) and Kenichi Hayashida and Yasunori Oka (Japan Somnology Center). For expert advice on statistics, we thank Drs Nelson Kinnersley and Remy Luthringer. Support for third party writing assistance for this manuscript was provided by F. Hoffmann-La Roche Ltd. Craig R Rayner, Stephen Toovey, Brian E Davies, Yoshio Hosaka, Masaichi Abe and Eric P Prinssen are (CRR, ST, YH, MA, EPP) or were (BD) employees of or consultants (ST) to F. Hoffmann-La Roche Ltd or Chugai Pharmaceutical Co. Ltd.