Parkinson’s disease (PD), the most common neurodegenerative movement disorder, has an annual incidence of 13.4 per 100,000 in the United States.1 Inability to fall asleep and stay asleep, disturbed motor activity during sleep, and excessive daytime somnolence are common in PD patients.2 In a community-based sample of 239 PD patients in Norway in the mid-1990s, 60% reported having some sleep problem; 27% of the 239 patients rated their overall nighttime problem as moderate-to-severe; and the most commonly reported sleep symptoms included frequent awakening (sleep fragmentation) and early awakening.3 Sleep problems in PD have been attributed to many diverse causes; for example, prescribed therapy for PD, such as dopaminergic treatment, or excessive nocturia, due to dysautonomia in PD, among many others.4 Although the impact of sleep disturbances—one of the many different non-motor manifestations of PD—have received recognition relatively recently, James Parkinson, in 1817, wrote that persons with PD were often “constantly sleepy” and “exhausted.”5
Health-related quality of life (HRQOL) includes physical, social, and mental health. Previous research suggests that insomnia is associated with poor HRQOL, especially with depressive symptoms, in PD. A study of an Australian PD cohort found that nocturnal sleep disruption was associated with depressive symptoms early in the disease course.6 In another study, insomnia was related to longer PD duration and worse HRQOL, including depression.7 Yet another study showed that insomnia was associated with greater pain and depressive symptoms.8 Also, short sleep duration (≤6 hours) and early morning awakening are strongly correlated with fatigue in PD.9
Although these and other studies have described associations of insomnia with HRQOL in PD,10 none have analyzed the relative associations of different types of sleep disturbance symptoms with HRQOL in PD. We hypothesize, based on previous research, that insomnia and daytime sleepiness will each be associated with worse HRQOL, including depressive symptoms, in PD. As an exploratory analysis, we analyze the relative contributions of additional sleep symptoms with HRQOL in PD; these include snoring and other indicators of sleep-disordered breathing.11,12
A group of 371 adults with idiopathic PD from three counties in central California were enrolled between 1998 and 2006 in a population-based study of predictors of PD onset.13 Eligibility for that study included being diagnosed with PD within the last 3 years, being English- or Spanish-speaking, and being over age 18 years. Participants received routine care for their PD from general neurologists in this region. Of the 371 participants in that study of PD incidence, 254 were eligible for and subsequently enrolled in a follow-up study of determinants of progression in PD.14 The diagnosis of PD was confirmed for all study participants by movement-disorder specialists at enrollment in the previous study and was reassessed in this follow-up study. The study was approved by the UCLA IRB (#G06–07–055), and all subjects provided informed consent after they received full explanation of study procedures.
The MOS Sleep measure (Appendix 1; online supplement) is a 12-item, self-administered measure assessing six specific sleep-problem symptoms: sleep disturbance: initiation (2 items), sleep disturbance: maintenance (2 items), awakening short of breath or with headache (1 item), daytime somnolence (3 items), sleep inadequacy (2 items), and snoring (1 item).15 Items were developed from a literature review and piloted with adults who were outpatients in an academic clinic and in a rural health clinic setting. Support for the reliability and validity of the MOS Sleep Measure in the general U.S. population and among patients with chronic medical conditions has been reported.16 Internal consistency reliability for multi-item scales exceeded 0.70 for all but Daytime Somnolence (Cronbach’s α=0.63). Impairment on the MOS Sleep measure was associated with worse mental health and with lower work productivity and quality.17 The frequency with which each problem has been experienced during the previous 4 weeks is rated on a 6-point scale, ranging from “None of the Time” to “All of the Time,” except one of the sleep-disturbance: initiation items on usual length of time to fall asleep, which has 5 categories of duration. All scores are transformed linearly to range from 0 to 100; higher scores indicate more of the attribute implied by the scale name (e.g., more snoring or more inadequate sleep).15,18
The SF−36 (Version 2.0) is a generic, health-related quality of life survey administered in the RAND Medical Outcomes Study, with 36 items defining 8 scales: Physical Functioning, Role Limitations Due to Physical Health, Role Limitations Due to Emotional Problems, Pain, Emotional Well-Being, Energy, General Health, and Social Functioning. A Physical Health Composite score (PCS) and Mental Health Composite score (MCS) are derived from the 8 scales. All scales and composite scores are calculated as T-scores with a mean of 50 and a standard deviation (SD) of 10. Age- and gender-adjusted general population norms are calculated. Evidence for its reliability and validity in PD patients has been reported.19
Depressive symptoms were assessed by the Patient Health Questionnaire (PHQ−9) and the Geriatric Depression Scale (GDS−15). The PHQ−9 is a brief, self-administered depression assessment measure found to be accurate in medical settings.20 PHQ−9 items map onto nine depression symptoms from the DSM-IV-TR, including one item: “trouble falling or staying asleep, or sleeping too much.” Subjects answer how many days over the past 2 weeks they had the symptom—0: None; 1: Several Days; 2: More Than Half the Days; and 3: Nearly Every Day. A symptom severity measure (range: 0–27) is the sum of the numerical responses for all questions. The PHQ−9 was self-administered, although a research assistant could assist any study participant who had difficulty writing responses. The GDS−15 is a self-administered, 15-item Yes/No questionnaire (range: 0−15), which does not contain any items about sleep. The GDS−15 performs well as a screening tool in distinguishing depressed from non-depressed PD patients.21
Movement-disorder specialists evaluated and rated stage of PD with the Modified Hoehn & Yahr score.22,23 These assessments were conducted with patients who were off Parkinson's disease medication since the previous evening (“off” state). Stage 0 is no sign of disease, and Stage 5 is wheelchair-bound or bedridden.
Age, gender, education level, marital status, employment status, medical comorbidities, and duration of diagnosis of PD were obtained through survey and interview questions at the baseline assessment for the PD incidence study and updated at the follow-up assessment.
Data collection for the follow-up study took place over a 25-month period beginning in June 2007. Sleep, depression, the SF−36, and disease-stage measures were collected through self-administered and interviewer-administered questionnaires or in-person examination by a movement-disorders specialist. The sleep measures were added to the study partway through the data collection period.
Demographic and clinical characteristics of 153 included and 101 excluded participants were compared by use of bivariate statistics (t-tests, Wilcoxon rank-sum, chi-square). Reasons for exclusion were the following: either the MOS Sleep measure or SF−36 was not completed or they were completed more than 23 days apart from each other (N=70); disease-stage was not assessed with the Hoehn & Yahr scale (N=30); or the diagnosis of PD was not confirmed at the follow-up examination (N=1). MOS Sleep measure and SF−36 responses were compared with age- and gender-adjusted general population norms18,24 by t-tests. Zero-order correlations between the MOS Sleep measure and SF−36 HRQOL scales were used to determine whether there were any suppression effects in the subsequent multivariate analyses.
To determine relative associations of each MOS Sleep scale with the SF−36 health-related QOL scales and depressive symptoms, we used linear-regression models, adjusting for age, gender, and PD stage (Hoehn & Yahr). For each of the 12 models (eight SF−36 scales, two SF−36 composite scores, and two depression measures), backward stepwise regression was used to determine a reduced model that included only variables whose associations with the dependent variable (SF−36 scales or depression measures) were a p value <0.05 (two-tailed). Analyses were conducted with SAS Version 9.2 (SAS Institute, Cary, NC).
Characteristics of the 153 subjects included in this analysis are presented in Table 1. Of 254 subjects enrolled in the PD progression study, characteristics of the 153 included and 101 excluded subjects were not different, except that the 153 included subjects had a PD duration of 4.6 years (SD: 2.2), versus 6.4 years (SD: 2.1) years (p<0.0001) for those excluded.
TABLE 1.Characteristics of Parkinson’s Disease Sample (N=153)
| Add to My POL
|Mean (SD) or N (%)|
|Age, years, mean (SD); range: 43–92||72.2 (9.2)|
|Women, N (%)||62 (40.5)|
|Currently married, N (%)||114 (75.0)|
|Highest level of education, N (%)a|
| Less than high school diploma||29 (18.9)|
| High school diploma||56 (36.6)|
| Technical or trade-school diploma||19 (12.4)|
| College diploma or graduate school||42 (27.5)|
|Ethnicity, N (%)|
| White or European American||116 (75.8)|
| Latino or Hispanic||27 (17.7)|
| Other||10 (6.5)|
|Work status, N (%)|
| Employed||32 (21.1)|
| Unemployed||6 (4.0)|
| Retired||108 (71.1)|
| Other: disability (3), emeritus (1), housewife (1), temp work (1)||6 (4.0)|
|Duration of PD diagnosis, years, mean (SD)||4.6 (2.2)|
|UPDRS Motor Score, on medication, mean (SD)||16.9 (10.6)|
|Hoehn & Yahr Stage, off medication, N (%)|
| Stage 0: No signs of disease||0|
| Stage 1: Unilateral disease||12 (7.8)|
| Stage 1.5: Unilateral plus axial involvement||7 (4.6)|
| Stage 2: Bilateral disease, without impairment of balance||50 (32.7)|
| Stage 2.5: Mild bilateral disease with recovery on pull test||45 (29.4)|
| Stage 3: Mild-to-moderate bilateral disease; some postural instability; physically independent||29 (19.0)|
| Stage 4: Severe disability; still able to walk or stand unassisted||6 (3.9)|
| Stage 5: Wheelchair-bound or bedridden unless aided||4 (2.6)|
|Mean number of medical comorbidities (SD)||0.69 (1.13)|
|Median (IQR)||0 (0, 1)|
Compared with age- and gender-adjusted U.S. general population norms, study participants reported worse sleep-maintenance, disturbance, snoring, sleep inadequacy, and daytime somnolence (all p <0.001). Participants also reported significantly worse HRQOL than the general population on seven of the eight SF−36 scales (no difference in Pain scores) and on both SF−36 Composite scores (Table 2).
TABLE 2.MOS Sleep Measure and SF–36 (version 2.0) Scores of Study Participants and Comparison With U.S. General Population Norms (N=153)
| Add to My POL
|Study Sample Mean (SD) N=153||Age- and Gender-Adjusted Norms18,41||p|
|MOS Sleep measure (range: 0–100)a|
| Sleep disturbance: Initiation||21.8 (25.6)||19.2||NS|
| Sleep disturbance: Maintenance||34.9 (24.5)||28.3||0.001|
| Awakening short of breath or with headache||9.8 (20.0)||8.1||NS|
| Daytime somnolence||39.7 (20.7)||26.9||<0.0001|
| Sleep inadequacy||40.8 (26.8)||27.5||<0.0001|
| Snoring||37.3 (33.4)||26.7||<0.0001|
|SF–36 version 2.0 (T-scores)b|
| Physical Functioning||36.8 (11.5)||42.5||<0.0001|
| Role Limitations: Physical||39.1 (11.2)||43.5||<0.0001|
| Role Limitations: Emotional||42.7 (11.9)||46.5||0.0002|
| Pain||46.3 (10.9)||47.0||NS|
| Emotional Well-Being||48.0 (10.3)||51.6||<0.0001|
| Energy||45.9 (11.2)||49.7||<0.0001|
| General Health||42.0 (9.3)||46.9||<0.0001|
| Social Functioning||44.3 (10.3)||48.0||<0.0001|
|SF–36 (version 2.0) Composite scores|
| Physical Health||39.2 (10.0)||43.2||<0.0001|
| Mental Health||48.3 (11.0)||51.5||0.0005|
Awakening short of breath or with headache and disturbances in sleep initiation were broadly associated with worse HRQOL, with significant associations of these sleep symptoms and seven of the eight SF−36 scales (Role Limitations: Physical was the exception), and both SF−36 Composite scores (Table 3). Daytime somnolence was significantly and uniquely associated with five SF−36 scales and the Physical Health Composite score. In contrast, snoring was not associated with any SF−36 scale or composite, and sleep inadequacy was only associated with Pain. (The Sleep Disturbance: Maintenance scale was excluded from multivariable regression models because of suppression effects.)
TABLE 3.Unique Associations of Types of Sleep Symptoms and Parkinson’s Disease Severity With Health-Related Quality of Life and Depression, From Multivariate Regression Models (N=153)
| Add to My POL
|Independent Variables From Backward Stepwise Regression Models|
|Types of Sleep Disturbances: MOS Sleep Scales||Disease Severity|
|Dependent variable (SF–36 scale or depression measure)||Sleep-Initiation Disturbance||Awakening Short of Breath or With Headache||Daytime Somnolence||Sleep Inadequacy||Hoehn & Yahr Stage|
|β (95% CI)||β (95% CI)||β (95% CI)||β (95% CI)||β (95% CI)|
|SF–36 Physical Functioning scale||–0.08 (–0.15, –0.03)||—||–0.12 (–0.19, –0.05)||—||–6.64 (–8.53, –4.75)|
|SF–36 Role Limitations: Physical scale||–0.08 (–0.14, –0.02)||–0.11 (–0.19, –0.02)||–0.14 (–0.21, –0.06)||—||–5.25 (–7.31, –3.21)|
|SF–36 Role Limitations: Emotional scale||–0.10 (–0.17, –0.03)||–0.12 (–0.21, –0.02)||—||—||–2.97 (–5.38, –0.56)|
|SF–36 Pain scale||—||–0.13 (–0.22, –0.05)||—||–0.09 (–0.15, –0.03)||–1.77 (–3.90, 0.36)|
|SF–36 Emotional Well-Being scale||–0.12 (–0.18, –0.06)||–0.17 (–0.25, –0.09)||—||—||–1.02 (–2.98, 0.94)|
|SF–36 Energy scale||–0.08 (–0.14, –0.01)||–0.17 (–0.25, –0.09)||–0.13 (–0.21, –0.05)||—||–3.08 (–5.17, –0.99)|
|SF–36 General Health scale||–0.11 (–0.16, –0.05)||–0.09 (–0.16, –0.02)||–0.08 (–0.14, –0.01)||—||–2.49 (–4.26, –0.72)|
|SF–36 Social Functioning scale||–0.08 (–0.14, –0.02)||–0.10 (–0.18, –0.02)||–0.11 (–0.18, –0.04)||—||–3.86 (–5.82, –1.89)|
|SF–36 Physical Health Composite score||–0.05 (–0.11, –0.01)||–0.10 (–0.17, –0.03)||–0.13 (–0.20, –0.07)||—||–5.22 (–6.94, –3.49)|
|SF–36 Mental Health Composite score||–0.11 (–0.18, –0.05)||–0.15 (–0.23, –0.06)||—||—||–1.13 (–3.29, 1.03)|
|Depression: PHQ–9a||—||0.07 (0.03, 0.12)||—||0.06 (0.03, 0.09)||1.46 (0.43, 2.49)|
|Depression: GDS–15||0.05 (0.03, 0.07)||0.05 (0.02, 0.07)||—||—||0.59 (–0.07, 1.25)|
In the multivariable models, awakening short of breath or with headache was most strongly associated with worse depression on both Depression measures (p ≤0.0006; Table 3). Sleep inadequacy was also associated with more severe depressive symptoms on the PHQ−9 (p ≤0.0002), and sleep initiation disturbance was associated with worse Depression as assessed by the GDS−15 (p <0.0001). Snoring and daytime somnolence were not associated with Depression symptoms.
Our findings confirm that PD patients experience a disproportionate number of sleep complaints and worse health-related quality of life than the general population.25–27 Although it is known that the majority of PD patients have sleep disturbances affecting their ability to fall asleep, their ability to maintain sleep, or daytime somnolence,28 we are unaware of studies that have evaluated the relative contributions of different types of sleep disturbances and symptoms in PD to HRQOL. As hypothesized, insomnia and daytime somnolence had strong negative associations with HRQOL, but we also uncovered significant unique associations of awakening with shortness of breath or with headache with worse HRQOL in PD. Our results are different from those of Martinez-Martin, 2006,29 demonstrating a weak association between sleep disturbances and HRQOL, but are in line with a later study by the same author showing that sleep and fatigue are associated with poorer HRQOL30 and another study demonstrating a similar correlation, utilizing the Parkinson’s Disease Sleep Scale (PDSS) by Scaravilli et al.31 We may presume that different sleep assessment inventories may have contributed to these differences.
Waking up with “shortness of breath or headaches” does not appear to be more common in PD than in the general population, but, when it is present, it is associated with worse HRQOL. Although patients with PD are not more likely to have worse symptoms of OSA than the general population, some PD patients have noteworthy OSA symptoms. Possible mechanisms may be related to the wearing off overnight of PD medication, resulting in anxiety or re-emergence of PD symptoms such as increased muscular rigidity. Levodopa therapy may also be associated with the emergence of irregular and rapid breathing, alternating with brief periods of apnea, in a pattern consistent with a central origin. This temporal relationship of respiratory disturbance to the administration of levodopa may suggest a peak-dose drug effect.32 One additional theory focuses on an underlying autonomic dysregulation of breathing associated with REM sleep,33 which is more likely to wake up patients during the latter part of the night; hence, waking up short of breath or with a headache. The underlying pathophysiology of sleep-disordered breathing in PD may be less dependent on weight (i.e., obesity) as a predisposing factor to the anatomical restriction at the level of the hypopharynx because body mass index (BMI) does not change before the onset of PD,34 and PD patients tend to have lower BMI than control subjects.35 However, as noted above, other mechanisms could include respiratory dyskinesia, rigidity of ribcage musculature, and the possible negative impact of dopaminergic therapy on respiratory tone.36
Other findings that emerge from these data are that daytime sleepiness and difficulties initiating sleep are associated with virtually every domain of HRQOL. In PD, sleep-initiation difficulties are often related to depression, wearing-off effects of dopaminergic drugs, and movement disorders of sleep, such as restless legs syndrome.7 Hypersomnia in PD is most likely centrally mediated, but could well be related to prescribed dopaminergic agents that are used to treat the disease, and comorbid psychiatric disorders. Although underlying sleep disturbances due to inadequately-controlled nighttime PD symptoms, such as periodic leg movements or nocturia seen in the later stages of PD, may cause sleep-maintenance difficulties, they may also contribute to sleepiness during the day due to their contribution to poor and insufficient sleep.
Depressive symptoms are common in patients with PD. In this sample, there were significant associations of worse scores on one or both depression measures with waking up feeling short of breath or with a headache and with initiation-of-sleep disturbance. These types of sleep disturbance were commonly associated with problems in most domains of HRQOL and were significantly associated with both SF−36 Composite scores. This underscores the complex relationships of sleep with depression and HRQOL. Given the high frequency of depression in people with PD, sleep disturbance must be assessed as both a primary symptom of PD in addition to a marker of an underlying depressive disorder. Depression measures such as the GDS−15 that do not screen using items about sleep and movement disturbance may be better tools for depression assessment in those with PD,14 minimizing confounding due to the occurrence of these sleep disturbances as a primary symptom in PD independent of depression.
This is a cross-sectional study of associations, limiting any interpretation regarding causality. Also, there may be residual confounding or inadequate adjustment due to unmeasured covariates. Whereas one group of investigators concluded that the generic SF−36 may not be sufficient for assessing HRQOL in PD, others found that it performed equally with or better than two disease-targeted measures.19 The PHQ−9 contains a question specifically addressing sleep, which may confound the relationship between sleep disturbance and score on the PHQ−9. Although we did not conduct polysomnography in this study, at least one study of treatment of sleep problems in epilepsy found greater improvement in patients’ self-reports of sleep on the MOS Sleep Measure than from polysomnogram assessments. The authors of that study posited the explanation that even “relatively small improvements in sleep continuity might be perceived as a relief to the patient,” but that additional research was needed to understand the implications of differences in these two assessment methods.37 The MOS Sleep Scale was not included in the recent Movement Disorders Society taskforce review of sleep measures;38 however, it has been shown to have good psychometric properties in patients with restless leg syndrome39 and in a patient population with pure sleep-related seizure.40
In conclusion, our community-dwelling sample of PD patients report more significant sleep disturbances, particularly sleep-maintenance insomnia, daytime sleepiness, and inadequate sleep, than a general population sample. Virtually every HRQOL domain is also affected in PD. Furthermore, although we confirmed our hypothesized associations of insomnia and of daytime somnolence with HRQOL in PD, we also identified strong associations of awakening short of breath or with headache and HRQOL in PD. The selective association of certain sleep disturbance symptoms—some of which were unanticipated—suggests that assessing specific symptoms may be worthwhile in the clinical setting. Although we speculate about possible mechanisms for symptoms such as awakening short of breath or morning headaches, their etiology in PD patients is not known and warrants further investigation, given their strong associations with HRQOL.
We thank Honghu Liu, Ph.D for statistical consulting assistance; Jeff Bronstein, M.D., Ph.D., for assisting in UPDRS and Hoehn & Yahr data collection; and Aaron Cook, M.P.H., for technical assistance in manuscript preparation.