Introduction

Top Abstract Introduction Methods Results Discussion References

Increased daytime somnolence has long been recognised as a side effect of dopaminergic drugs, although there were no descriptions on sudden irresistible sleep attacks in clinical trials or national pharmacovigilance databases before the study by Frucht and others.^1-3 Pramipexole and ropinirole were the first dopamine agonists associated with road crashes, but now all dopamine drugs are implicated. ^{3 4} The legitimacy of sleep attacks as a phenomenon distinct from normal somnolence has been questioned, and the associated medicolegal and social issues have generated much controversy.^4-10

We aimed to determine whether sleep attacks do exist or whether they are indistinguishable from drug induced somnolence and excessive sleepiness, the prevalence of sleep attacks, which dopamine drugs are implicated, whether attacks are predictable and preventable, and whether they are treatable.

	Methods

Top Abstract Introduction Methods Results Discussion References

We searched for the key words "Parkinson" or "Parkinson's disease" or "parkinsonism," and "sleep" or "narcoleptic" or "attacks" and a combination of these between July 1999 and May 2001. We did not search prior to these dates because there were no reported cases of sleep attacks until 1999.³ We identified relevant publications from electronic searches of three general biomedical databases (Medline, Embase, Pascal) and from hand searches of major neurological journals (Brain, Annals of Neurology, Archives of Neurology, Neurology, Movement Disorders, Journal of Neurology Neurosurgery and Psychiatry, Journal of Neurology, Nervenarzt) and general medical journals (Lancet, New England Journal of Medicine, British Medical Journal), the abstracts of congresses (13th international congress on Parkinson's disease, international symposium on gait disorders, second international congress on mental dysfunction in Parkinson's disease, American Academy of Neurology 52nd meeting, 10th meeting of the European Neurological Society, sixth international congress of Parkinson's disease and movement disorders), and the reference lists of relevant articles.

Sleep attacks are defined as events of overwhelming sleepiness that occur without warning or with a prodrome sufficiently short or overpowering to prevent protective measures.¹⁰ Most of the reports we surveyed were issued before this definition was published, so the inclusion criteria for sleep attacks in these studies varied greatly. To overcome this methodological problem, we used the reported characteristics of the onset of the event and the reliability of witnesses to the event to grade these phenomena into sleep attacks (definite, probable, and possible), sleep episodes, and sleep events not otherwise specified (box).

	Results

Top Abstract Introduction Methods Results Discussion References

We identified 20 publications covering sleep events in 124 patients (table 1). We classified sleep events into 96 sleep attacks (5 definite, 8 probable, and 83 possible), 4 sleep episodes, and 23 sleep events not otherwise classified. Overall, 38 patients were taking ropinirole, 32 pramipexole, 13 bromocriptine, 23 either lisuride or piribedil, 5 pergolide, 2 apomorphine, and 1 cabergoline. Eight patients were receiving levodopa monotherapy, and in two the dopamine agonist was not specified. Complete information on sex, age, and duration of exposure to the drug before the onset of a sleep attack was available in 33 patients (table 2). The age ranged from 34-87 years, and two thirds of the patients were male. Parkinson's disease was staged from Hoehn and Yahr 1 to 3, and the duration of disease ranged from 1-20 years. Two patients had sleep events on first exposure to the drug, but in the others they occurred with recurrent exposures (2 weeks to 20 years).¹¹

Nature of sleep events

Circumstances around and before sleep attacks
The circumstances at the time of the sleep event were described in 29 patients, with several reporting more than one event. In 17 cases the sleep event happened during driving, leading to road crashes in 10. In 12 cases non-driving attacks occurred during standing (3 patients), walking (1), talking (3), eating (3), writing (1), and grooming a pet (1). Twenty patients had recurrent sleep events.

Clinical and electrophysiological descriptions of time course
We identified six reports on 16 separate sleep events describing their exact time course: three with complete information and three with rudimentary information. Electroencephalography was performed in six patients, ^{5 14 21} but only three sleep events were captured in two of the patients. ^{5 21}

Epworth sleepiness scale
None of 79 patients with positive scores on the Epworth sleepiness scale had had sleep events in the past. ^{25 26} The scale comprises eight questions about falling asleep in inappropriate but tiring situations. Possible ratings for each question are 0 (never), 1 (slight chance), 2 (moderate chance), and 3 (high chance). A score of 14 or higher is considered indicative of excessive daytime sleepiness.²⁶ Of the four patients tested who had had sleep events none showed a clear positive Epworth score (0, 8, 11, 12). ^{11 27} Lang found the scale to have a sensitivity for prediction of falling asleep while driving of less than 50%.²⁸

Prevalence of sleep events
Two retrospective and seven prospective studies, totalling 1787 patients with Parkinson's disease, gave data on prevalence on sleep events. They occurred in 6.6% (range 0%-30%) of patients taking dopamine drugs (table 3). In Montastruc and others' study of 236 patients with Parkinson's disease, more sleep events occurred with ropinirole (41%) than with bromocriptine (36%), lisuride (27%), or piribedil (30%).²⁹ But differences between agonists were not significant according to Hobson and others.³⁰

Treatment of sleep events
Treatment strategies were reported in 30 of the 123 patients (table 4). Active treatment was effective with modafinil in one patient with waves of sleepiness and with amantadine in one patient with possible sleep attacks. In 25 patients the drug dose was either reduced (10 patients) or discontinued (15 patients), which prompted either complete (22 patients) or partial (3) cessation of sleep events. Switching from one dopamine agonist to another was reported in three cases and brought about initial remission in two patients but recurrence in one. One of the two patients for whom remission of sleep attacks was initially reported represented after several months (W Pirker, personal communication, 2001). An adjustment of the intake schedule to the patients' need to be awake at certain times was successfully attempted in two patients taking levodopa.¹⁵ To avoid car crashes, one patient with episodic waves of sleepiness would pull over and take a nap if tiredness occurred while driving.¹⁵

	Discussion

Top Abstract Introduction Methods Results Discussion References

Up to 30% of patients taking dopamine agonists for Parkinson's disease have sleep attacks. These are a class phenomenon, and their prediction, prevention, and treatment are yet to be solved.

The concept of sleep attacks has been disputed by several authors, who speculate that these episodes are but exaggerated daytime drowsiness.^3-8 But recent detailed first hand reports by medical specialists are of particular value as they support the existence of two types of sleep events. ^{10 11 14 20}

A limitation of our review is that publication bias might have led to an overestimation of the prevalence of sleep attacks. Also data were derived from clinics dealing with movement disorders, which might not be representative.

Dealing with sleep attacks and their consequences
Driving presents problems in patients with sleep events because of their decreased awareness of drowsiness over time and because passengers may not notice their reduced vigilance while driving.⁴ However, some potentially disastrous outcomes can be prevented.⁴ By definition the patients in our review experienced sleep episodes. Prevention of car crashes relies on the detection of patients at risk and advising them not to drive.

Secondary prevention has been proposed by using the Epworth sleepiness scale, multiple sleep latency test, and polysomnography as screening tools.⁴ Polysomnograms of 27 patients showed no strong association between excessive daytime sleepiness and disrupted night sleep.³¹ Analysis of past sleep events suggested that low sensitivity of screening tools might be increased by adding the inappropriate sleep composite score.³⁰ This score was devised to examine the likelihood of falling asleep during five potentially stimulating activities. In contrast to most of the items of the Epworth sleepiness scale, sleep episodes while driving, eating, working, conversing, and doing household chores were thought to represent problematic and pathological excessive daytime sleepiness.³⁰ The available data are limited and do not allow either to be considered a reliable predictor of future risks of sleep events.

Little is known about general risk factors of sleep events. Contrary to the findings of Montastruc and others more (70:53) patients in our review had sleep event while taking ergot dopamine agonists. ^{23 29} These findings have to be treated with caution, however, because of publication and selection bias. Preliminary data suggest that the risk of a sleep event is higher in patients with dysautonomia or males. ^{23 29} Currently neither a narcolepsy-like Parkinson's disease phenotype nor higher doses of dopamine agonists, duration of treatment, Hoehn and Yahr stage, age, or prior episodes of falling asleep can be considered specific risk factors for sleep events. ^{6 9 13 29 30} An intake of sedatives was also not predicative.²⁹

With no reliable risk factor available, effective treatment is crucial to counteract the consequences of sleep attacks. Various strategies such as dose alterations or active treatment have been attempted but not in a controlled and prospective way. As switching from one dopamine agonist to another can lead to a recurrence of symptoms,³ it seems preferable to reduce the dose of drug. However in only 7 of 30 cases in which treatment modalities were provided did a reduction of dopamine agonist dose result in complete remission of sleep events. Remission was incomplete (but presumably tolerable) in three patients, whereas in 15 the drug was stopped by the doctor. Why this was done was never specified. Because long term side effects can be more disabling than disease related symptoms, low dose therapy that just meets the needs of the patient has become the rule.³² A reduction of dose would lead to a worsening of motor symptoms, counter to the primary aim of treating Parkinson's disease. But even in those patients for whom a reduction is initially tolerated, disease progression will inevitably lead to an increased dose, possibly with a renewed risk of later sleep events. These strategies therefore can not be considered good and permanent solutions.

Driving safety
The health authorities of Canada, the European Union, and the United States have asked the manufacturers of pramipexole and ropinirole to advise doctors to warn patients not to drive or engage in comparable risky activities while taking the drugs. ^{4 10} We found 17 cases where the sleep event occurred during driving, leading to road crashes in 10 cases. Despite the potential danger from driving, experts believe that sleep attacks are too infrequent to recommend that patients taking dopamine agonists for Parkinson's disease stop driving. ^{4 7 8 10 33} Studies suggest that whether or not patients with Parkinson's take dopamine drugs they do not cause more road crashes than age matched controls. ^{4 34} Recommendations for driving, other than informing patients of a potential risk, should be made with caution until more data are available.