eFigure. Flowchart of the Study Population
eTable 1. Model Fitness Assessed by Bootstrap Validation
eTable 2. Number of Missing Observations for Each Covariate Used in the Analyses
eTable 3. Baseline Characteristics of the Included vs Excluded Population
Customize your JAMA Network experience by selecting one or more topics from the list below.
Identify all potential conflicts of interest that might be relevant to your comment.
Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.
Err on the side of full disclosure.
If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.
Not all submitted comments are published. Please see our commenting policy for details.
Zhang X, Molsberry SA, Pavlova M, Schwarzschild MA, Ascherio A, Gao X. Association of Sleepwalking and REM Sleep Behavior Disorder With Parkinson Disease in Men. JAMA Netw Open. 2021;4(4):e215713. doi:10.1001/jamanetworkopen.2021.5713
Is the presence of sleepwalking, either alone or co-occurring with rapid eye movement (REM) sleep behavior disorder, associated with higher odds of having Parkinson disease in men?
In this cross-sectional study of 25 694 men from the Health Professional Follow-up Study, both probable sleepwalking and REM sleep behavior disorder were significantly associated with higher odds of having Parkinson disease.
This study suggests that sleep regulation might be associated with Parkinson disease–related neurodegeneration.
Previous studies conducted among patients with Parkinson disease (PD) reported that parasomnias other than rapid eye movement (REM) sleep behavior disorder (RBD), particularly sleepwalking (SW), are associated with PD severity. However, it remains unclear whether the presence of SW is associated with altered odds of having PD in a population-based study.
To evaluate whether probable SW, either alone or co-occurring with probable RBD, is associated with higher odds of PD in men.
Design, Setting, and Participants
This cross-sectional study included 25 694 men from the Health Professionals Follow-up Study, a population-based cohort of male health professionals in the US with information on probable SW and probable RBD. Data collection took place between January 2012 and June 2018, and data analysis took place from July 2020 to October 2020.
Probable SW and probable RBD were measured by questions adapted from the Mayo Sleep Questionnaire in 2012.
Main Outcomes and Measures
PD, confirmed after review of medical records by a movement disorder specialist.
Of the 25 694 studied men (mean [SD] age, 75.6 [7.4] years), 223 (0.9%) had probable SW, 2720 (10.6%) had probable RBD, and 257 (1.0%) had PD. After adjusting for potential confounders (eg, age, smoking, caffeine intake, chronic disease status, and other sleep disorders), compared with individuals without probable SW and probable RBD, participants with probable SW, probable RBD, and both probable SW and probable RBD had higher odds of PD, (probable SW: odds ratio [OR], 4.80; 95% CI, 1.61-14.26; probable RBD: OR, 6.36; 95% CI, 4.83-8.37; both probable SW and probable RBD: OR, 8.44; 95% CI, 3.90-18.27).
Conclusions and Relevance
In this cross-sectional study of a male population, probable sleep parasomnias, including both SW and RBD, were associated with higher odds of having PD. PD-related neurodegeneration may impair arousal regulation during sleep.
Parkinson disease (PD) is the fastest-growing neurological disease in prevalence, disability, and deaths globally since 1990.1 PD is a movement disorder, but several nonmotor features have been frequently observed among patients with PD, which may be associated with progressive neurodegeneration.2 Identifying nonmotor symptoms is critical for understanding disease pathology and clinical significance.3
Rapid eye movement (REM) sleep behavior disorder (RBD) is one of the most extensively studied nonmotor features of PD.2,3 Small case series and descriptive studies conducted among patients with PD have suggested that sleepwalking (SW), a parasomnia that emanates from non-REM sleep (usually slow-wave sleep), appeared to have a high prevalence in patients with PD (approximately 10%).4,5 Given that SW is generally considered a rare condition in adults (1.0%-2.3%, according to a meta-analysis6), this prevalence in patients with PD appears unusually high, but to our knowledge, no study to date has included a group of individuals without PD in direct comparison for risk estimation. Furthermore, in additional studies of patients with PD, the presence of non-REM parasomnias was also associated with worse symptoms, cognitive impairment, and depression,7,8 which could be associated with dysfunction of arousal and motor control.4,8 Therefore, we conducted a cross-sectional study using data from the Health Professionals Follow-Up Study (HPFS) to examine the association between probable SW and probable RBD, measured by a validated questionnaire, and PD in more than 25 000 US men.
We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cross-sectional studies. This study was based on the HPFS, a cohort that started in 1986 with 51 529 male health professionals in the United States. The study protocol was approved by the institutional review boards (IRBs) of the Brigham and Women’s Hospital and the Harvard T.H. Chan School of Public Health, and the IRBs allowed participants’ completion of questionnaires to be considered implied consent. Participants received mailed questionnaires every 2 years; in 2012, it included questions on probable SW and probable RBD for the first time. Among the 27 093 participants who responded to the probable SW and probable RBD questions, 1343 individuals (5.0%) were excluded because they did not have a sleep partner to answer questions concerning the participant’s probable SW and probable RBD (eFigure in the Supplement).
Detailed procedures for PD ascertainment have been described previously.9 Briefly, PD cases that occurred before 2012 were confirmed by a movement disorder specialist reviewing medical records for those who reported having received a PD diagnosis on the regular follow-up questionnaires.9
The adapted Mayo Sleep Questionnaire was used to assess probable SW and probable RBD by asking whether the participant’s sleep partner had noticed the participant walking or acting out dreams while sleeping.2,10 In a validation study (n = 176), the question of probable SW had 100% sensitivity and 86% specificity, while the question of probable RBD had 98% sensitivity and 74% specificity relative to polysomnography results.10 Based on answers to these 2 questions, participants were categorized in 4 groups: no probable SW or probable RBD, probable SW only, probable RBD only, and both probable SW and probable RBD.
Potential confounders were assessed via questionnaires, including age, smoking status, weight and height, diet (eg, alcohol and caffeine consumption), history of major chronic diseases (eg, hypertension and diabetes), restless legs syndrome (RLS), and sleep-related parameters (eg, sleep duration, insomnia symptoms, and use of hypnotics).11,12
Statistical analyses were performed from July to October 2020, using SAS version 9.4 (SAS Institute). Statistical analyses were 2-tailed (α < .05). Odds ratios (ORs) with 95% CIs from the Firth penalized logistic regressions were estimated, and P values were calculated using the Wald test, adjusting for age, smoking status (never, past, or current), body mass index (calculated as weight in kilograms divided by height in meters squared; <21, 21-24.9, 25-29.9, 30-34.9, or ≥35), alcohol consumption (0, 1-9.9, 10-19.9, 20-29.9, or ≥30 g/d), caffeine consumption (mg/d), hypertension (yes or no), diabetes (yes or no), total sleep duration (<6 hours, 6-8 hours, >8 hours), excessive daytime sleepiness (rarely or never, some days, or most days), hypnotics use (yes or no), antidepressant use (yes or no), and RLS (yes or no). When the case number is small, a penalized logistic regression allows the computation of reliable, finite estimates of coefficients if separation in a data set is observed,13 and model fitness was evaluated by bootstrap validation adjusting for optimism to exclude the risk of overfitting. The model fitness ranged from good to very good; optimism-corrected C-statistic ranged from 0.739 to 0.851 (eTable 1 in the Supplement). We used missing values indicators for individuals missing information on a covariate so that each model included the same number of individuals. We were aware that this method could be subject to residual confounding, but we determined that this was likely to be modest because of the small number of individuals missing information on each covariate (eTable 2 in the Supplement).
Interaction between probable SW and probable RBD was examined by multiplying the 2 factors and entering them in the model after adjusting for the aforementioned covariates. We further tested an interaction between these sleep disorders and age, smoking, and caffeine consumption in relation to the odds of having PD, and we examined the association of disease duration by conducting a subgroup analysis stratified by years since PD diagnosis. Because RLS, another common sleep and movement disorder, was associated with a risk of PD in the HPFS,11 we also performed a sensitivity analysis by excluding participants with RLS.
At the time of assessment, the 25 694 participants had a mean (SD) age of 75.6 (7.4) years. A total of 223 (0.9%) had probable SW, 2720 (10.6%) had probable RBD, and 257 (1.0%) had documented cases of PD. Participants with probable SW or probable RBD were more likely to have longer sleep duration compared with those without sleep parasomnias (18 [17.1%] or 462 [17.8%] vs 2407 [12.7%]), prevalent daytime sleepiness (20 [19.1%] or 466 [18.0%] vs 2291 [12.1%]), hypnotics use (42 [39.3%] or 497 [19.3%] vs 3224 [17.0%]), and RLS (9 [9.7%] or 134 [6.3%] vs 961 [5.7%]) (Table 1). The excluded population were older and had more sleep complaints (eg, excessive daytime sleepiness, hypnotics use) compared with the included population (eTable 3 in the Supplement). In our cohort, 8 of 11 (72.8%) of patients with PD and probable SW also had probable RBD.
Participants with single or both probable parasomnias were compared to address misclassification. Individually, both probable SW and probable RBD were significantly associated with PD in the fully adjusted model (probable SW: adjusted OR, 4.80; 95% CI, 1.61-14.26; probable RBD: adjusted OR, 6.36; 95% CI, 4.83-8.37) (Table 2). Participants with both probable SW and probable RBD had an OR of 8.44 (95% CI, 3.90-18.27), although the multiplicative interaction between probable SW and probable RBD was not significant. Similar results were observed after excluding participants with RLS (Table 2). No significant interaction was found with age, smoking status, or caffeine consumption.
After stratification by PD duration, probable SW without probable RBD was only significantly associated with PD among those with disease duration of 4.1 years to 8 years (OR, 6.27; 95% CI, 1.32-29.81) and more than 8 years (OR, 12.05; 95% CI, 3.22-45.04), although this could be limited by small sample size in the shorter PD duration groups (Table 3). In contrast, probable RBD without probable SW was consistently associated with higher odds of PD across disease duration groups (≤4 years: OR, 6.47; 95% CI, 4.33-9.60; 4.1-8 years: OR, 5.35; 95% CI, 3.33-8.57; >8 years: OR, 7.11; 95% CI, 4.43-11.41) (Table 3).
To our knowledge, this is the first population study examining the association of SW with a major neurological disease with a general population as a reference group. We found that probable RBD with probable SW had a stronger association with PD than probable RBD alone.
SW occurs most often during arousal from the non-REM sleep phase. It manifests as a complex disorder, involving motor dyscontrol, mental confusion, and amnesia.14 Pathologic changes occasionally occur in the ventral mesopontine junction, which contains dopaminergic neurons14 and thus shares a common pathway with PD development. Furthermore, a genome-wide parametric linkage analysis for SW suggested a candidate region on chromosome 20, which includes the adenosine deaminase gene.15 Difference in adenosine metabolism could be associated with A1 and A2A adenosine receptor antagonism, which has been shown to directly interact with dopaminergic signaling in animal studies.16
RBD is one of the strongest prodromal nonmotor symptoms for PD that has been identified to date.3 The RBD-PD association could be explained by α-synuclein deposition, a common pathology in both diseases.3,17 When assessed by questionnaire, participants may not be able to correctly separate SW from other nighttime movements caused by different etiologies, such as RBD.18 Questionnaire-assessed SW had a sensitivity of 100% and specificity of 86% relative to polysomnography confirmed cases.10 In a previous polysomnography study, 80% of patients with PD and a history of SW had co-occurring RBD.7,8 Consistently, 72.8% of patients in our cohort with PD and probable SW also had probable RBD. Analyses were done separately for single or both parasomnias in this study to address the potential misclassification associated with this low specificity.
Based on our results, a longer duration of PD was associated with probable SW and with a stronger association between probable SW and PD than between probable RBD and PD. This could be because of chance, given that there were fewer cases in the shorter PD duration groups. However, this observation is consistent with a previous study conducted in patients with PD, which found that individuals with longer disease duration were more likely to have SW.5 Arousal-related disorders, including SW, were more common among patients with PD who had longer treatment, greater severity of disease, and more depression and anxiety.7,8 Patients with PD who had more severe cognitive impairment had a higher risk of SW but did not have a higher risk of probable RBD.8 Progressive neurodegeneration and changes in neurotransmitter levels may further influence arousal regulation during the progression of PD. While RBD has been recognized as a prodromal symptom of PD, SW appears to be associated with symptomatic PD and associated sleep disruption.
Sleep disruption in PD is common. A previous survey reported that 64% of individuals with PD had sleep problems.19 Sleep disruption is multifactorial; is sometimes considered as a nonmotor symptom of PD; is sometimes attributed by patients to nighttime tremor, neuropsychiatric symptoms, or postural discomfort that is difficult to counteract because of bradykinesia; and has been hypothesized to also relate to circadian disruption. Even early in the disease, patients with PD may have an altered BMAL1 expression and melatonin secretion.19 Patients with PD may have reduced amplitude of melatonin secretion, particularly those who reported sleepiness.20 Patients with RBD and PD have altered expression of several genes involved in circadian rhythm regulation including human period gene 1 (PER1), PER2, and PER3, a trend for the later dim light onset of melatonin secretion.21 As dysregulation of the circadian control of sleep may lead to more fragmented sleep,22 sleep fragmentation of both REM and NREM sleep may be a major contributing factor to the higher frequency of probable SW and probable RBD in patients with PD. Although sleep fragmentation was not assessed in this study, we observed a higher percentage of participants with SW and RBD having excessive daytime sleepiness and either shortened (<6 hours) or prolonged (>8 hours) sleep duration.
Compared with previous studies on SW and PD, this study is the first to include non-PD control participants from a large cohort.5 However, it does have limitations. One major limitation of this study is the questionnaire-based method for assessing sleep disorders, which could introduce misclassification, as addressed previously in this article. However, the Mayo Sleep Questionnaire has been validated against clinical diagnosis and proved appropriate for use in large, population-based studies.2,3,17,23 Due to the cross-sectional nature of this study, we are unable to make conclusions about the temporal relationships between probable SW and PD.10 Evidence from future prospective studies is needed to address the value of SW as a potential prodromal marker for PD. Information on the childhood history of SW was not collected, limiting our ability to examine the potential early-life risk predisposition. Of note, the HPFS cohort consists of male health professionals who are predominantly White and who, at the time of this study, had a mean (SD) age of 76 (7.4) years, younger than the age with peaked prevalence globally (85-89 years).1 Thus, results from this study might not be able to generalizable to other populations with different characteristics.
In this study, the presence of parasomnia, including both probable RBD and probable SW, was more common among men with PD, relative to those without PD. Prospective cohort studies in which long-term follow-up data are available with a reasonable number of SW and PD cases are warranted for better understanding the association.
Accepted for Publication: February 22, 2021.
Published: April 13, 2021. doi:10.1001/jamanetworkopen.2021.5713
Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2021 Zhang X et al. JAMA Network Open.
Corresponding Author: Xiang Gao, MD, PhD, Department of Nutritional Sciences, The Pennsylvania State University, 109 Chandlee Lab, University Park, PA 16802 (firstname.lastname@example.org).
Author Contributions: Dr Gao had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Ascherio, Gao.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Zhang.
Critical revision of the manuscript for important intellectual content: Molsberry, Pavlova, Schwarzschild, Ascherio, Gao.
Statistical analysis: Zhang, Molsberry.
Obtained funding: Ascherio, Gao.
Conflict of Interest Disclosures: Dr Pavlova reported receiving grants from Jazz Pharmaceuticals, Lundbeck, and Biomobie and personal fees from Sanofi, Oakstone, AudioDigest, and Mass Medical Consultation during the conduct of the study. Dr Schwarzschild reported receiving personal fees from Eli Lilly and Co, Denali Therapeutics, Prevail Therapeutics, and Parkinson Study Group for advisory services to nQ Medical, Partner Therapeutics, and Chase Therapeutics outside the submitted work, receiving personal fees from Pennsylvania State University during the conduct of the study, and holding a patent for Adenosine A2A knockout mouse line with royalties paid via Massachusetts General Hospital. No other disclosures were reported.
Funding/Support: The Health Professional Follow-up Study cohort is funded by the National Institutes of Health through grant U01 CA167552. This study is funded by the National Institute of Neurological Disorders and Stroke through grant 1R01NS102735-01A1.
Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Disclaimer: The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Create a personal account or sign in to: