Bound to supine sleep: Parkinson’s disease and the impact of nocturnal immobility

BACKGROUND
Impaired nocturnal mobility is a well-known problem in Parkinson's disease (PD), and clinical experience suggests a predominance of supine body position during sleep. However, this assumption - and potential consequences - still awaits objective validation by a polysomnography-based and adequately controlled study.


METHODS
Clinical and polysomnographical analysis of 80 consecutive PD patients and 80 control subjects carefully matched for age, sex, body mass index and apnea-hypopnea index.


RESULTS
PD patients slept twice as much in supine position than control subjects (62.2 ± 32.9% vs. 34.2 ± 28.5%, p < 0.001). In PD, but not in control subjects, more supine sleep correlated with fewer changes in body position (rho = -0.434, p < 0.001). Longer PD disease duration was an independent predictor of more supine sleep in multiple linear regression analysis (β = 0.389, p < 0.001); conversely, more supine sleep was associated with higher apnea-hypopnea index and daytime sleepiness.


CONCLUSIONS
We confirmed that supine sleep is common in PD, and increases with longer disease duration. Our findings indicate that supine sleep may contribute to the overall disease burden by deteriorating sleep-disordered breathing and daytime vigilance.

1. Studies with human subjects must have been conducted in accordance with the Declaration of Helsinki. All persons must have provided informed consent prior to being included in the study.
2. Studies with animal subjects must have been conducted in accordance with the Guide for the Care and Use of Laboratory Subjects as adopted by the US National Institutes of Health and/or according to the requirements of all applicable local, national and international standards.
3. Protocols with animal or human subjects must have been approved by the relevant local committee(s) charged with ensuring subject protection. Studies that entail pain or distress will be assessed in terms of the balance between the distress inflicted and the likelihood of benefit. 4. The authors declare that the manuscript is original, that it is not being considered for publication elsewhere, and that it will not be submitted elsewhere while still under consideration for Parkinsonism & Related Disorders or after it has been accepted by Parkinsonism & Related Disorders. 5. All authors have seen and approved the manuscript in the form submitted to the journal. The authors declare that they have conformed to the highest standards of ethical conduct in the submission of accurate data and that they acknowledge the work of others when applicable. 6. All sources of financial support for the work have been declared in the Acknowledgements section of the manuscript. Any additional conflicts of interest must also be declared. Please include declarations of any consultancy or research funding received from relevant companies from three years prior to performance of the research until the time of manuscript submission. If the research is supported by internal funds, that should be stated as well.
To indicate compliance with the preceding declaration and that you have obtained agreement from all of the authors of this paper to declare their compliance as well, please place an x here: _X__ In cases of uncertainty please contact an editor for advice.

*Author Declaration
Dear Dr. Bonifati Thank you for giving us the opportunity to submit a revised version of our manuscript. We also would like to thank the three reviewers for their thoughtful comments, which helped to improve the quality of the manuscript. Our responses to the reviewers' concerns are written in italics, and the changes in the manuscript are highlighted with red font. We hope you will now find the current version suitable for publication.
I again certify that all coauthors have seen and agree with the contents of the manuscript. It is not under review at any other publication.
Thank you again for reconsidering our manuscript for publication.

Reviewer #1
Sleep disorders are such complex problems that one must be careful in deducing too much, especially from a single, retrospective study. Your study appears to have been well done and the ms is well written. I have some reservations and suggestions.
We thank Reviewer #1 for this encouraging comment and the helpful suggestions.
In the highlights section, I suggest noting on the first line that this was a retrospective study. This indicates to the reader that you were unable to control for many variables that might be important.
We added a statement that this was a retrospective study.
In the introduction, I suggest deleting the statement that "nocturnal immobility is recognized as a disease-defining feature." Perhaps you mean "common disease feature?" I am not aware of any criteria for defining PD that include nocturnal immobility.
We changed the wording according to the reviewer's suggestion.
I applaud your matching PD and controls for BMI and AHI, but, as you know, there are many other items you might have chosen. Your mean ages were the same. Did you purposely match for age? Please state.
Thank you for this comment. We understand that perfect matching is not feasible. For this study, we deliberately matched the two groups for age, sex, BMI and AHI, because we believe these four factors most likely impact the distribution of different body positions during sleep.
You state that one body position you measured was "upright." Did some subjects sleep upright in a bed? Please clarify The used sensor is able to detect "upright" body position, i.e. when sitting or standing up; however, no sleep in this position was detected in this study. We apologize for this confusion and have now clarified this issue in the Methods section.
You state that "probably only few conditions lead to such a pronounced increment in supine..." Unless you have data to support this, I suggest deleting. I suspect that many neurological and orthopedic disorders are associated with this, and, in the U.S., pediatricians now have babies sleep on their backs to avoid SIDS. This might possibly lead to most adults learning to do this.
We deleted this statement, as we indeed lack any supporting data.

*Response to Reviewers
Most importantly, you need to list the many weaknesses in your study. We have no information on why sleep studies were obtained, which may have been quite different in the two groups. We don't know how many were smokers, had pain syndromes, had back problems, had RBD, had breathing disorders, were obese, what medications they took, whether supine sleeping correlated with nocturia, etc etc. You don't have to list every issue you can think of, but you should mention a few since they are relevant and impossible to match for.
Thank you for these suggestions. We appreciate your concerns, and are now discussing the limitations inherent to such a retrospective study in more detail.
Most importantly, you should acknowledge that you found correlations, not cause and effect relationships. You have not provided any data to indicate that less sleeping supine is helpful. Perhaps a trial of having people sleep with something like a small ball attached to their sleep attire to make it impossible to sleep supine would be an inexpensive intervention to test your hypothesis.
I am not sure many will agree with your suggestion to include supine sleeping in studies of treatment interventions. There are so many possible things to study and your single study does not evaluate effects of reducing the time in the supine position.
We agree that our non-interventional study does not provide any evidence that improved nocturnal mobility and less supine sleep would also lead to meaningful changes in quality of life of PD patients. We understand the difference between correlation and cause and effect relationship, and have now tempered our conclusions accordingly.
However, we do believe supine sleep and nocturnal immobility should be addressed in studies on sleep quality in PD and that this issue has so far been too neglected.

Reviewer #2
The primary aim of this article is to determine body position of PD patients during sleep. This can be determined also by sleep partner report or videography. Expensive polysomnogram is not needed.
Patients were selected retrospectively from referrals to a sleep lab. Since it is not routine practice to order PSG for all PD patients, presumably these PD patients had suspicion of possible sleep apnea so they were referred for PSG. The data or these patients cannot be extrapolated to all PD patients.
We absolutely agree that polysomnography is expensive and information on body position could be achieved cheaper and by other means. Indeed, among our patients, no polysomnography was actually done primarily to assess distribution of body positions during sleep. Rather we wished to benefit from the many polysomnographies that had been performed in PD patients in our department, and set out to analyze, whether PD patients really do sleep more in supine position.
We also agree that a potential referral bias cannot be excluded. However, we believe we reasonably accounted for this risk by using a rater-independent propensity score matching with inclusion of four major variables. In addition, due to the high prevalence of sleep-wake disturbances in PD and their possible diagnostic, therapeutic and medico-legal implications, polysomnography has become an integral diagnostic procedure in PD patients in our department. Thus, a significant referral bias seems quite unlikely, because polysomnography was performed in most PD patients independent of the presence or absence of (suspected) sleep apnea or any other sleep disorder.
We admit, on the other hand, that our PD population is representative for a tertiary center, and our findings cannot be generalized to all PD patients.
We added these limitations to our discussion.

Reviewer #3
This is an interesting and well written paper confirming that supine sleep is, first, more common in PD than in normal controls and second, increases with disease duration. Reported below, there are a few questions that add more information.
We thank Reviewer #3 for this appraisal of our study.
1. The multiple regression included disease type and found no influence of disease type with the akinetic-rigid form. However other type of approach could be used to determine whether there might be some association between this form and the propensity for supine position during sleep, such as cluster analyses or group comparisons.
We also calculated a group comparison (ANOVA) for association of disease type with amount of supine sleep; we did not find significant differences in this analysis ( 3. Did the author explore possible relation between PSG findings and time of supine position? We did correlation analysis of all polysomnographic findings listed in Table1 and below; however, we did not find any significant correlation (Pearson's r). We summarized these negative findings in the Result section ("supine sleep position did not correlate with any sleep stage, sleep efficiency or arousal index").

Supine position [%]
Pearson's r p Finally, a graph about the correlation between disease duration and more supine sleep in PD should be added in Figure 1.
We added a scatter plot to figure 1.
We thank Reviewer #3 for the helpful comments. All authors report no potential conflicts of interest and have no disclosure.

Subjects, clinical assessment and matching procedure
This was a retrospective chart review. In a first step, we selected 119 consecutive PD patients who received whole-night video-polysomnography (PSG) as an integral diagnostic procedure between 2004 and 2009. In order to compare their data on sleep positions and nocturnal mobility to an adequate control group, we considered 248 consecutive nonneurological patients from our data base, in which PSG was performed during the same time for suspected sleep disorders. We used propensity score matching as a rater-independent and unbiased method to generate two groups accurately matched for age, sex, body mass index (BMI, body mass divided by the square of individual's height) and apnea-hypopnea index (AHI, number of apnea and hypopnea events per hour of sleep) [5]. By this means, we eventually included 80 PD patients and 80 matched control subjects.
Diagnostic procedure, clinical assessment and calculation of levodopa dose equivalents (LDE) are identical to previous studies [5]. We ascertained sleepiness using the Epworth Sleepiness Scale (ESS), with scores of ≥10 indicating excessive daytime sleepiness (EDS). The study was approved by the local Ethics Committee.

Polysomnography assessment
All patients had overnight PSG recordings, using a multi-channel recording system (Embla, RemLogic TM ). Scoring of sleep stages and respiratory events was performed visually using standardized criteria [5]. Apnea was defined by a cessation of oro-nasal airflow longer than 10 seconds, and hypopnea by a reduction of oro-nasal airflow by at least 50% lasting more than 10 seconds and accompanied by an arousal or SaO 2 reduction of ≥ 3%. A body position sensor was attached by an elastic velcro strap at the level of the lower sternum and

Statistical analysis
Statistical analyses were performed using SPSS (version 21

Clinical and polysomnographic findings
Demographic, clinical and polysomnographic findings are summerized in  (Fig. 1B). In PD, but not in controls, multiple linear regression analysis revealed that more supine sleep was independently associated with higher AHI (β=0.260, p=0.015 and age as second significant associate: β=0.349, p=0.001).
Periodic limb movements in sleep did not correlate with supine position in either group.
Likewise, supine sleep position did not correlate with any sleep stage, sleep efficiency or arousal index, and group comparison did not reveal any difference in supine sleep between PD patients with various disease type (akinetic-rigid, tremor-dominant, equal). Finally, the use of dopamine agonists had no impact on total amount of supine position, and correlation analysis between LDE of dopamine agonists and of long-acting dopaminergic drugs at bedtime with supine position did not demonstrate any significant associations.

Predictors of supine sleep in PD
Linear regression analysis identified longer disease duration to be independently associated with more supine sleep in PD (β=0.389, p<0.001) (Fig. 1C). is therefore increasingly advocated [7].

Discussion
In this line, our results and those of many other groups confirm that SDB is the most obvious problem caused by lying in supine position. Supine position increases the risk of upper airway collapses, causes more and longer apnoeic events, and necessitates a higher pressure from CPAP devices [8,9]. On the other hand, it is unlikely that increased SDB severity represents the sole explanation for the negative association between supine sleep and daytime sleepiness. Even healthy people with poor sleep quality spend more time in supine position than good sleepers [12], and this observation suggests a mutual influence of supine position and poor sleep quality. In our study, however, we failed to demonstrate any correlation between polysomnographic measures of sleep consolidation and amount of supine position, which is in contrast to previous studies reporting an adverse impact of nocturnal immobility on subjective sleep quality.

Subjects, clinical assessment and matching procedure
This was a retrospective chart review. In a first step, we selected 119 consecutive PD patients who received whole-night video-polysomnography (PSG) as an integral diagnostic procedure between 2004 and 2009. In order to compare their data on sleep positions and nocturnal mobility to an adequate control group, we considered 248 consecutive nonneurological patients from our data base, in which PSG was performed during the same time for suspected sleep disorders. We used propensity score matching as a rater-independent and unbiased method to generate two groups accurately matched for age, sex, body mass index (BMI, body mass divided by the square of individual's height) and apnea-hypopnea index (AHI, number of apnea and hypopnea events per hour of sleep) [5]. By this means, we eventually included 80 PD patients and 80 matched control subjects.
Diagnostic procedure, clinical assessment and calculation of levodopa dose equivalents (LDE) are identical to previous studies [5]. We ascertained sleepiness using the Epworth Sleepiness Scale (ESS), with scores of ≥10 indicating excessive daytime sleepiness (EDS). The study was approved by the local Ethics Committee.

Clinical and polysomnographic findings
Demographic, clinical and polysomnographic findings are summerized in Table 1 (Fig. 1B). In PD, but not in controls, multiple linear regression analysis revealed that more supine sleep was independently associated with higher AHI (β=0.260, p=0.015 and age as second significant associate: β=0.349, p=0.001).
Periodic limb movements in sleep did not correlate with supine position in either group.
Likewise, supine sleep position did not correlate with any sleep stage, sleep efficiency or arousal index, and group comparison did not reveal any difference in supine sleep between PD patients with various disease type (akinetic-rigid, tremor-dominant, equal). Finally, the use of dopamine agonists had no impact on total amount of supine position, and correlation analysis between LDE of dopamine agonists and of long-acting dopaminergic drugs at bedtime with supine position did not demonstrate any significant associations.

Predictors of supine sleep in PD
Linear regression analysis identified longer disease duration to be independently associated with more supine sleep in PD (β=0.389, p<0.001) (Fig. 1C).

Discussion
Compared to carefully matched control subjects, PD patients slept almost twice as much in supine position and less often changed nocturnal body posture. In addition, we identified longer PD duration as independent predictor of more supine sleep. These results have been expected, and may be regarded as inevitable consequences of nocturnal immobility and axial rigidity. However, more than confirming a commonplace, our study suggests relevant clinical implications of this predominance for supine sleep position, as indicated by the association with increased severity of sleep-disordered breathing and excessive daytime sleepiness.
In this line, our results and those of many other groups confirm that SDB is the most obvious problem caused by lying in supine position. Supine position increases the risk of upper airway collapses, causes more and longer apnoeic events, and necessitates a higher pressure from CPAP devices [8,9]. On the other hand, it is unlikely that increased SDB severity represents the sole explanation for the negative association between supine sleep and daytime sleepiness. Even healthy people with poor sleep quality spend more time in supine position than good sleepers [12], and this observation suggests a mutual influence of supine position and poor sleep quality. In our study, however, we failed to demonstrate any correlation between polysomnographic measures of sleep consolidation and amount of supine position, which is in contrast to previous studies reporting an adverse impact of nocturnal immobility on subjective sleep quality.