Article Text


Occurrence and clinical correlates of REM sleep behaviour disorder in patients with Parkinson’s disease over time
  1. M D Gjerstad1,2,
  2. B Boeve3,
  3. T Wentzel-Larsen4,
  4. D Aarsland1,5,6,
  5. J P Larsen1,2,6
  1. 1
    The Norwegian Centre for Movement Disorders, Stavanger, Norway
  2. 2
    Department of Neurology, Stavanger University Hospital, Stavanger, Norway
  3. 3
    Department of Neurology, Mayo Clinic College of Medicine, Rochester, MN, USA
  4. 4
    Centre for Clinical Research, Haukeland University Hospital, Bergen, Norway
  5. 5
    Department of Psychiatry, Stavanger University Hospital, Stavanger, Norway
  6. 6
    Institute of Clinical Medicine, University of Bergen, Bergen, Norway
  1. Michaela D Gjerstad, The Norwegian Centre for Movement Disorders, Stavanger, Norway, PB 8100, N-4068 Stavanger, Norway; gjmi{at}


Objective: To examine the occurrence and clinical and demographic correlates of REM sleep behaviour disorder (RBD) in patients with Parkinson’s disease (PD) in a community-based cohort over 8 years.

Methods: 231 patients with PD were included in a population-based prevalence study in 1993. Patients were then followed prospectively and reexamined after 4 and 8 years. Semi-structured interviews for information on clinical and demographic data were applied at all study visits. Standardised rating scales of parkinsonism, depression and cognitive impairment were used. The diagnosis of probable RBD (pRBD) was based on a sleep questionnaire. Proportional-odds ordinal logistic regression models for clustered data were used to study the relationship between pRBD and various demographic and clinical variables.

Results: 231 patients were evaluated for RBD in 1993 and, after 4 and 8 years, 142 and 89 patients, respectively, were available for re-evaluation. The frequency of pRBD varied from 14.6% to 27% during the study period. Probable RBD was related to male gender, higher dopaminergic treatment and less severe parkinsonism.

Conclusion: We found that the frequency of pRBD varied over time and that it is associated with male gender, less parkinsonism and higher levodopa equivalent dose. Our findings indicate that dopaminergic therapy may contribute to the expression of RBD and that RBD is symptomatic in earlier stages of PD.

  • Parkinson’s disease
  • RBD
  • prospective
  • sleep

Statistics from

REM sleep behaviour disorder (RBD) in humans was first described by Schenck et al in 1986.1 It is characterised by loss of normal skeletal muscle atonia during rapid eye movement (REM) sleep and associated with vivid dreaming and complex motor activity during sleep.2 RBD may be idiopathic or may precede a neurodegenerative disease, most often the α-synucleinopathies.36 The neuropathologic substrate for RBD is still not clear, although growing evidence implicates brain-stem structures.2 7 8 The Braak staging scheme for Parkinson’s disease (PD) suggests an ascending progression of Lewy body and Lewy neurite pathology from the medulla to the neocortex,9 which may explain the occurrence of RBD years before the development of the motor syndrome of PD and other synucleinopathies, such as dementia with Lewy bodies and pure autonomic failure. The prevalence, development over time and prognostic implication of RBD in patients with PD remain largely unknown.

Fifteen percent of 61 consecutive patients with PD from a movement disorder outpatient clinic fulfilled the minimal diagnostic criteria for RBD according to the International Classification of Sleep Disorders (ICSD).10 In another study, 33% of 33 consecutive patients with PD were diagnosed with RBD or REM sleep without atonia (RSWA) based on polysomnographic recordings.11 These studies note the importance of RBD in patients with PD, but the patient samples were derived from outpatient clinics and cannot be viewed as representative of the population of patients with PD in any community.

To our knowledge, there are no population-based longitudinal studies concerning RBD in patients with PD over time. We have, therefore, in this prospective longitudinal cohort study, explored the occurrence, clinical correlates and progression of RBD in patients with PD.


Study population

A total of 245 patients were diagnosed with PD on 1 January 1993 in a prevalence study in the county of Rogaland, Norway.12 The prevalence rate was 110.9 per 100 000 inhabitants. The patients with suspected symptoms of PD were systematically registered and examined by neurologists with a special interest in movement disorders and diagnosed according to published criteria.13 Patient recruitment was based on hospital files, information from all general practitioners, nursing homes, district nurses and health visitors of the region and has previously been described in detail.12 So far, 22 patients have been examined by autopsy and the clinical diagnosis of PD was confirmed in all cases.14

In 1993, 231 of the 245 patients were evaluated for nocturnal sleeping problems.15 16 Seven patients were re-diagnosed as not having PD during follow-up, 3 patients died between prevalence day and the scheduled examination, 2 patients refused to participate and 2 patients could not be evaluated due to incomplete data. After 4 years (1997), data from 143 patients were available and after 8 years data from 89 patients were available.

An age- and sex-matched control group of 100 well-functioning and generally healthy elderly individuals was selected in 1993 for comparison. These individuals were mainly recruited by their general practitioner after routine visits and had no major diseases that would cause significant disability.15 The control group was examined only in 1993.

Study design

This study is a prospective longitudinal cohort study of suspected features of RBD in patients with PD over 8 years. The original study population of 231 patients at baseline was derived from a community-based prevalence study. Patients were examined with the same standardised examination programme in 1993, 1997 and 2001.

All patients were interviewed and evaluated by a neurologist and a psychiatrist or psychiatric nurse from the study group. A close family member or daily caregiver of the patients with PD was asked to attend the session for support and additional information if necessary.

Clinical assessment

The standardised evaluation programme that was applied in 1993, 1997 and 2001 included a semi-structured questionnaire on disease history and demographic data and the Unified Parkinson’s Disease Rating Scale (UPDRS),17 the Hoehn & Yahr staging,18 the Beck Depression Inventory (BDI) and the Mini-Mental State Examination (MMSE).19 20 The UPDRS was assessed in the “on” state. Hallucinations were diagnosed when the patients or the accompanying family member or caregiver reported hallucinations with or without insight according to the UPDRS, part one, item 3. Due to the variety of dopaminergic treatment, a Levodopa Equivalent Dose (LED) was estimated as follows: LED: (regular levodopa dose × 1) + (levodopa controlled-release dose × 0.75) + (pramipexole dose × 67.0) + (ropinirole dose × 16.67) + (pergolide dose and cabergoline dose × 67.0) + (bromocriptine dose × 10) + ((regular levodopa dose + levodopa controlled-release dose × 0.75) × 0.25)) if taking tolcapone or entacapone)).2123

Assessment of nocturnal features

The minimal diagnostic criteria for RBD defined by the ICSD, at the time of examination, includes movement of limbs or body associated with dream mentation and at least one of the following: potentially harmful sleep behaviours; dreams that appear to be acted out; or sleep behaviours that disrupt sleep continuity (ICSD, 2001).24 For the evaluation of RBD suspected features in this study, a self-constructed questionnaire—the Stavanger Sleepiness Questionnaire (SSQ)—was used. This questionnaire has previously been applied and described in several studies.15 16 25 Patients were asked to rate the severity of motor or vocal activity when sleeping on a scale from 0 to 3. RBD was suspected (referred to as “probable RBD” and abbreviated “pRBD” henceforth) when patients scored 2 (very active during sleep, tend to wake up spouse) or 3 (very active physically and verbally, have been hitting or hurting theirself or caregiver while sleeping). In addition, the patients were asked to report bothersome vivid dreaming or nightmares. Insomnia was diagnosed if the patients reported night-time sleeping problems or used sleeping medication for a minimum of 3 months. Finally, the patients were asked to rate the quality of their night-time sleep from 0 to 3 (0 = without any problems, 1 = a little problematic, 2 = moderately problematic, 3 = difficult). Night-time sleep quality was classified as reduced if a patient scored 2 or more.

Statistical analysis

SPSS version 11.0 (SPSS, Chicago, IL) and Stata (StataCorp, TX, USA) were used for statistical analyses. We used the Mann–Whitney test to compare continuous variables and the Pearson Chi-square test for categorical variables of patients with and without pRBD.

The relationship between pRBD (four categories) and demographic and clinical variables was analysed by proportional-odd ordinal logistic regression models for clustered data using all observations available.26 Covariates used for the multivariate model were time since inclusion, gender, disease duration, UPDRS motor score, MMSE score, BDI score, levodopa dose and hallucination (present or absent). A two-sided p-value <0.05 was considered to be statistically significant.


Occurrence of probable RBD

Table 1 shows the degree of physical activity during sleep in patients with PD according to the SSQ. Data of 231, 143 and 89 patients with PD were available in 1993, 1997 and in 2001, respectively. Thirty-four (14.7%) patients reported features of pRBD at baseline, 39 (27.3%) in 1997 and 13 (14.6%) in 2001. Only 2 individuals (2%) of the healthy elderly reported pRBD features in the control group in 1993.

Table 1 Occurrence of REM sleep behaviour disorder suspect features in patients with Parkinson's disease in 1993, 1997 and 2001 according to the Stavanger Sleep Questionnaire: Are you restless/agitated/or physically active while sleeping during the night?

Factors associated with probable RBD

The clinical and demographic data of patients with PD and with and without pRBD at baseline in 1993 are shown in table 2. Patients with pRBD experienced significantly more often vivid dreaming or nightmares, as well as hallucinations, than patients without pRBD. They were also more often male, with higher daily LED. They had longer disease duration, and reported reduced night-time sleep quality. There was no difference in the amount of dyskinesia or in the percentage spent “off” per day in the two groups.

Table 2 Clinical data of patients with Parkinson's disease and with or without pRBD at baseline in 1993

To further explore the associations between pRBD and clinical and demographic characteristics, the data were then analysed with a clustered ordinal logistic regression model. Time since inclusion, gender, disease duration, UPDRS motor score, BDI score, MMSE score, LED and presence of hallucination were included as independent variables. All available observations from 1993 to 2001 were included. The Wald chi square was 23.32 (8 degrees of freedom, p = 0.003). We found that gender, lower UPDRS motor score and higher LED were significantly related to pRBD (table 3).

Table 3 Proportional-odds ordinal logistic regression model for variables associated with pRBD (four categories).

To evaluate the influence of different dopaminergic treatment on hallucinations and pRBD, patients were divided into three groups: those treated with levodopa alone, those treated with dopamine agonists alone, and those treated with a combination of levodopa and dopamine agonists. The frequency of hallucinations and pRBD between the three groups was then compared using the Chi square test. In 1993, significantly more patients with pRBD used a combination of levodopa and dopamine agonists (p = 0.014), whereas there was no difference in 1997 and 2001, and no difference was seen in respect of the development of hallucinations and the type of dopaminergic treatment used (data not shown).

Course of probable RBD over time

One hundred and forty-three patients were examined in 1993 and 1997, and 89 of these patients were examined at all three study visits. Of the 143 patients examined in 1993 and 1997, 18 (12.6%) had pRBD at baseline and 11 (7.7%) of these reported pRBD at both examination points. Twenty eight (19.6%) of 143 patients had developed new pRBD by 1997.

Of the 89 patients available at all study visits, 13 (14.6%) had pRBD at baseline and 3 (3.4%) of these patients reported pRBD at all three assessments. Four (4.5%) patients experienced pRBD in 1997 and 2001, and 6 (6.7%) patients out of the 89 reported pRBD for the first time in 2001. In addition, we observed that 16 (18%) patients no longer experienced pRBD in 2001 compared with in 1997.

No significant differences were found when comparing the patients with inactive RBD with those continuously reporting RBD (data not shown). None of the patients used clonazepam and no differences were seen in the use of benzodiazepines or antidepressant therapy.


We found that 15% of this community-based cohort of patients with PD had pRBD at baseline. pRBD tended to vary with time and one-third of those with pRBD at baseline were without pRBD 4 years later. pRBD is associated with male gender, less parkinsonism and higher levodopa equivalent dose. We also found that patients with pRBD experienced more frequent vivid dreaming or nightmares, as well as hallucinations, than patients without pRBD. Our findings should be interpreted in the context of the few other studies of RBD in patients with PD.10 11 2732

No solid epidemiological data exists on RBD in any population. The closest approximation of the prevalence of RBD in the general population to date is based on the data by Ohayon et al, which suggests a prevalence of approximately 500 per 100 000 (0.5%).33 If the data from our study is extrapolated to this important question, as we previously demonstrated a point prevalence of 111 patients with PD per 100 000 in 1993 (245 PD patients among 220 000 inhabitants),12 we can estimate a point prevalence of probable clinically significant RBD associated with PD of 15 per 100 000 (34/220 000). There are probably also additional groups of patients with RBD features—clinically insignificant RBD and RSWA in patients with PD, which we could not identify without the use of polysomnography (PSG), suggesting that the prevalence of RBD and RSWA associated with PD is probably far greater than 15/100 000 in our and other community populations. This knowledge may prove useful as epidemiological studies are developed to more specifically assess RBD in the community.

The tendency for RBD to occur in males more than females in this study is consistent with all other studies on RBD.4 34 Why RBD is primarily a male disorder remains unknown.

The finding that patients with pRBD experienced significantly more frequent vivid dreaming or nightmares simply reflects RBD itself, as altered dream mentation, usually of a violent or nightmare quality, is inherent to the disorder.34 35

The higher frequency of visual hallucinations in the patients with pRBD is consistent with other reports,29 31 3638 even despite the lack of significance in the proportional-odds ordinal logistic regression model analysis. The pathophysiological substrate for visual hallucinations in PD and Lewy body disease, in general, is poorly understood. One hypothesis likens some of the phenomena in PD to those in narcolepsy, which is a disorder reflecting multiple wakefulness-REM sleep dissociations. In narcolepsy, cataplexy and sleep paralysis probably represents the normal muscle paralysis of REM sleep invading into wakefulness, whereas hypersomnolence represents REM sleep invading into wakefulness, and hallucinations represent dream imagery of REM sleep invading into wakefulness. RBD, which also commonly occurs in narcolepsy,39 represents the normal muscle tone of wakefulness invading into REM sleep, thereby causing dreams to be acted out and/or dreams to develop in response to excessive motor activity. The higher frequency of visual hallucinations in our patients with PD with pRBD could be interpreted as supporting some contribution of sleep state dissociation underlying both RBD and visual hallucinations in PD.

We also found a higher LED and less severe parkinsonism in the patients with PD with pRBD. These findings may seem contradictory. The use of higher LED patients with pRBD may have contributed to the increased hallucinations, vivid dreaming and lower PD severity found in this group. This interpretation is weakened by the finding that treatment with the dopaminagonist pramipexole is effective in patients with RBD40 and suggests that higher LED would probably decrease the frequency and severity of RBD. Furthermore, as reported by Benbir et al, more frequent visual hallucinations occur in those with RBD than those without, despite equivalent doses of dopaminergic treatment.37 In addition, it is possible that the higher LED could be viewed as supporting the role of dopamine in the pathogenesis of RBD,41 42 or it might suggest that more dopaminergic treatment may provoke a change from previously existing RSWA or clinically insignificant RBD into symptomatic RBD. There is conflicting data on the effects of dopaminergic drugs on RBD and REM sleep atonia, with some investigators reporting increased RBD frequency and/or severity with levodopa43 and pramipexole,44 while others report decreased RBD features.4548 The two reports that suggest worsening RBD features with dopaminergic therapy involved subjects with PD, whereas the other reports that suggest less RBD features with these drugs involved patients with idiopathic RBD, traumatic brain injury or Lewy body dementia.4548 Therefore, dopaminergic therapy in patients with PD may worsen RBD but not in non-PD conditions.

The increased LED in patients with pRBD is not necessarily the reason for their lower UPDRS scores, as clinicians titrate the dopaminergic medication in each individual patient to achieve satisfactory minimisation of PD features without causing unbearable side effects. Less severe parkinsonism in the patients with pRBD indicates, possibly, that pRBD is primarily a manifestation in patients with early PD. This is also supported by the literature and our clinical experience, which suggest that RBD tends to manifest prior to the onset of parkinsonism and then decreases in frequency and severity over time.3 4 33 Our finding therefore supports the notion that RBD is a non-motor symptom of PD in the early stages.49

The findings by other investigators—15% of PD patients meeting the minimal ICSD criteria for RBD in one series,10 and 33% of PD patients with RBD or RSWA based on PSG in another series11—underscore some important, but often overlooked, aspects of dream enactment behaviour, RBD and RSWA. Some investigators equate RSWA with “subclinical” or “preclinical” RBD. Also, some patients exhibit mild motoric behaviour associated with RSWA on PSG, but they and their bed-partners do not report a history of dream enactment behaviour. Such patients could be viewed as having clinically insignificant RBD. We attempted to assess the frequency and clinical correlates of what could be argued as clinically significant RBD in a population-based sample of patients with PD, and the frequency range in our cohort over time —14–28%—is consistent with the Comella study, which probably also reflected clinically significant RBD.10 The higher frequency of RBD in the Gagnon et al study compared with ours probably reflects the inclusion of RSWA and clinically insignificant RBD in their study, as diagnoses were based on PSG findings.11 However, the frequency of what could be considered to be clinically significant RBD in their study was 17% (5 of 29 in the structured interview had a history of dream enactment behaviour10), which is comparable to our data. It could also be argued that, as the frequency data on clinically significant RBD are similar across these studies, and our data were gathered in a population-based sample of patients with PD, perhaps the other data from the Comella et al and Gagnon et al studies are closer to the actual frequencies in PD than could have been surmised previously, as their convenience samples were derived from outpatient clinics.10 11

pRBD fluctuated in prevalence in our study and only 3 out of 89 patients reported RBD suspected symptoms at all three study visits. This may be due to variations in medications and dosages in some cases. Another possible explanation is the increasingly destructed sleep with the progression of PD,50 which again may cause the development of different types of symptomatic or asymptomatic REM sleep-associated phenomena over time.

We acknowledge the limitations in this study. Healthy controls were examined only once and not studied longitudinally. The diagnosis of probable RBD was based on subject reports on a sleep questionnaire. Bed-partner reports may be as good as or better than patient reports for RBD.9 Using this questionnaire, we probably captured dream enactment behaviour that is clinically significant, and therefore our data probably underestimates the true frequency of RBD or the electrophysiologial substrate of REM sleep without atonia,10 which can only be examined by PSG. On the other hand, sleep apnoea may cause RBD similar movements during sleep, which may lead to an overestimation.51 Questions regarding symptoms associated with sleep apnoea were added to the SSQ after 2001 and are not available for our study population. Nevertheless, sleep apnoea may be a probable major contributor to RBD similar movements during sleep. However, our data reflect the knowledge and interest in nocturnal symptoms in patients with PD at the onset of this study (1993), and the same questionnaire was obviously used to maintain consistency over time. Most of our results are also consistent with the current state of knowledge in RBD associated with PD. Prospective studies using PSG in a longitudinal cohort of subjects with PD are warranted to further explore the relationship of RBD.


View Abstract


  • Competing interests: None.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.