Daytime sleepiness and other sleep disorders in Parkinson’s disease

Methods.

Consecutive patients with PD (new and established) seen at the Baylor College of Medicine PD Center and Movement Disorders Clinic completed sleep questionnaires over a 3 month period. No patient refused to participate. The survey included the Epworth Sleepiness Scale (ESS),6 sections from the National Sleep Foundation sleep survey7 (a 1,014 person survey of sleep habits in the United States), and specific questions regarding falling asleep while driving: “Recently, have you fallen asleep while you were driving? If so, have you fallen asleep while the car was still moving?” Input from the spouse or caregiver was allowed. If patients were not able to complete the entire questionnaire in clinic, they were subsequently interviewed over the phone or in person. We also asked about symptoms of restless legs syndrome (RLS): 1) urge to move legs with unpleasant sensation; 2) motor restlessness; 3) worsening with inactivity and temporary improvement with activity; and 4) worsening at night.8 All patients who answered affirmatively to all four cardinal criteria for the diagnosis of RLS were subsequently interviewed to ensure that they did have RLS. Medical records were reviewed to corroborate a correct diagnosis of PD, to assess for the presence of dementia, and to collect information about current medications and treatment history, duration and severity of PD (Hoehn and Yahr stage [H&Y]), prior diagnosis of RLS, and the presence of dementia. Patients were classified as having dementia if they had a documented Mini-Mental State Examination (MMSE) of less than 25, had neuropsychiatric testing that concluded dementia, or carried a formal diagnosis of dementia in our clinic.

Data were incorporated into a database and statistical analysis was performed on SPSS 10.0 (Chicago, IL). Prediction of ESS used a stepwise regression model, including 1) duration of PD; 2) age; 3) H & Y ; 4) sex; 5) dementia; 6) use of levodopa; 7) use of any DA; 8) history of thalamotomy; 9) history of pallidotomy; and 10) history of deep brain stimulation (DBS). A second model substituted individual DA (pramipexole, ropinirole, pergolide) for “use of any DA.” Predictors of “falling asleep while driving” used a backwards logistic regression analysis based on the Wald statistic and the same 10 predictors. Again, a second model substituted individual DA (pramipexole, ropinirole, pergolide) for “use of any DA.” To minimize any misinterpretation of predictor exclusion, as predictors may be highly correlated with the outcome,9 the criterion for inclusion and exclusion of a variable was set at a level of α ≤ 0.05 and α ≥ 0.15, respectively, for both analyses. Furthermore, stepwise selection of predictors was specified where predictors are removed from the equation one at a time, but could be retained if doing so would help prediction. Analyses were performed using SPSS Regression and SPSS Logistic Regression. Subsequent analyses between predicted group membership proportions were based on the χ² statistic. To determine the effects of drug dose on sleep, we used Spearman ρ, employing a dose equivalence formula of pergolide (1 mg): pramipexole (1 mg): ropinirole (2.5 mg): levodopa (100 mg): controlled release levodopa (74 mg).10

Results.

A total of 320 patients completed the survey over a 3 month period. Eleven were subsequently excluded because the diagnostic criteria for PD were not met. Four were excluded because they were in blinded drug trials and therefore their exact medications were not known, and two were excluded because we were unable to contact them to complete or verify data. Therefore, 303 (183 male, 60.4%) were analyzed. Patient demographics are summarized in table 1.

Questions from the National Sleep Foundation survey demonstrated numerous sleep abnormalities. Patients reported a mean of 6.3 ± 1.7 hours of actual sleep per night out of 7.4 ± 1.7 hours spent in bed attempting sleep. They reported an average of 2.7 ± 2.5 awakenings per night, and 4.5 ± 3.8 naps per week. “Some daytime sleepiness” was reported by 265 (87.5%) despite the fact that 133/265 (50.1%) of these patients “felt [they] got enough sleep the night before.” These patients most often ascribed this to medications (49.5%), their medical condition (34.7%), and age (26.4%).

Prediction of ESS scores.

ESS scores were also markedly abnormal (11.1 ± 5.7). In 152 (50.2%) patients, the score was above the traditional score, suggesting pathologic hypersomnolence (>10). Based on multiple regression analysis, duration of PD (r = 0.20, p = 0.001), male sex (Male_adj = 11.9, CI_95% 11.7 to 12.2; versus Female_adj = 9.7, CI_95% 9.4 to 10.1; Δ_{male versus female} = 2.2, CI_95% 1.8 to 2.6, p = 0.001), use of any DA (DA_adj = 11.9, CI_95% 11.6 to 12.1; versus No DA_adj = 9.8, CI_95% 9.5 to 10.2; Δ_{DA use versus no DA use} = 2.0, CI_95% 1.6 to 2.4, p = 0.003), and higher H&Y (r = 0.18, p = 0.004) did reliably improve prediction of higher ESS scores. Combined, these four predictors strongly predict higher ESS, R = 0.36, R²_adj = 0.12, F(4,298) = 10.8, p < 0.0001. The second regression analysis including the three individual DA showed that pramipexole (Pramipexole_adj = 11.7, CI_95% 11.4 to 12.8; versus No Pramipexole_adj = 10.7, CI_95% 10.4 to 11.0, p < 0.0001) and pergolide (Pergolide_adj = 12.5, CI_95% 11.9 to 13.0; versus No Pergolide_adj = 10.8, CI_95% 10.6 to 11.1, p < 0.0001) predicted higher ESS where ropinirole did not. This difference, however, might be attributed to different powers of the groups: pramipexole (n = 119), pergolide (n = 43), and ropinirole (n = 29). After excluding nine patients who were on multiple DA, and adjustment by covariates (duration of PD, male sex, and higher H&Y), adjusted ESS scores did not vary among the pramipexole, pergolide, and ropinirole groups (p = 0.94) (figure). Adjusted ESS scores, however, were higher in a subgroup of patients on any DA monotherapy (10.8 ± 1.7, n = 34) compared to a subgroup taking levodopa monotherapy (10.0 ± 1.7, n = 100), Student’s t-test(132) = 2.4, p < 0.02, two-tailed. ESS scores also showed a mild but significant correlation with an overall larger dose of dopaminergic (levodopa and DA) medications (r = 0.15, p = 0.01). ESS scores did not clearly correlate with the dose of DA (r = 0.13, p = 0.09) although there was a trend toward a correlation with higher doses.

• Download figure
• Open in new tab
• Download powerpoint

Figure. Sleep scores of different dopaminergic medications in PD. Comparisons of various dopaminergic medications for Epworth Sleepiness Scale (ESS) scores (raw and adjusted data), n = 303. Patients taking multiple dopamine agonists (n = 9) are not included. There were only four patients taking pergolide monotherapy and four taking ropinirole monotherapy. Despite apparent differences in the raw scores, analysis of covariance tests showed that only patients taking no dopaminergic medication had lower ESS. All dopaminergic medication groups were statistically similar.

Discussion.

We found that EDD and episodes of falling asleep while driving were very common in PD. They correlated with each other, with certain demographic features (especially older age), and with dopaminergic medications, particularly DA. ESS scores also mildly correlated with higher doses of dopaminergic medications, but were similar among patients taking the three major DA (pramipexole, pergolide, ropinirole).

EDD has been reported in most series that have compared patients with PD to controls.11-17⇓⇓⇓⇓⇓⇓ In the only other large survey that investigated specific factors associated with EDD, a sleep questionnaire was administered to 245 patients with PD in western Norway.16 They found that EDD correlated with more severe PD, a longer duration of dopaminergic treatment, diminished cognitive function, and hallucinations. Multiple sleep latency tests on two patients demonstrated reduced latency to sleep while on pramipexole, but not a reduced latency to REM sleep, as is seen with narcolepsy.4

The cause of EDD is probably multifactorial, but can be segregated into intrinsic abnormalities in PD, the effects of medications for PD, or a combination of both. Nocturnal sleep is markedly abnormal in both treated and untreated PD.14 In one survey, 98% of patients reported some sleep abnormality.18 The most common subjective nighttime complaints include stiffness and an inability to move, nocturia, dystonia, and sleep fragmentation.19-22⇓⇓⇓ Evaluations of patient with PD have also demonstrated increased periodic limb movements of sleep (PLMS),23,24⇓ sleep apnea, fragmented sleep architecture, reduced slow wave and REM sleep, and REM behavioral disorder.20

Somnolence has been reported as an adverse effect in all medications used to treat PD. This could result from their effect on either nocturnal sleep or daytime arousal. One controlled trial in humans found that levodopa produced as much drowsiness as triazolam, a soporific benzodiazepine.25 The levodopa-induced drowsiness, however, subsided over 11 days, whereas somnolence from triazolam continued. Polysomnographic studies of patients with PD and controls taking dopaminergic medications have yielded inconsistent and often contradictory results. Some have shown generally improved sleep,26,27⇓ whereas others have shown disrupted sleep patterns.28-31⇓⇓⇓ Animal studies show that low doses of DA appear to increase REM sleep, whereas higher doses may inhibit it.32,33⇓ Therefore, higher doses of DA may cause daytime somnolence by worsening nocturnal sleep patterns.

Daytime sleepiness may also result from a direct daytime soporific effect of DA. Chronic dopaminergic treatment has been shown to phase advance intrinsic melatonin levels.34 This could potentially lower the threshold for daytime sleepiness. Ionophoretic administration of D2/D3 agonists into the ventral tegmental area (VTA) dopaminergic nucleus (A10) induces both narcolepsy and cataplexy in narcoleptic Doberman dogs.35 Similar administration of the drug into the nigrostriatal nucleus (A9) caused cataplexy but did not induce sleep.36 Pure D1 agonists did not produce sleep or cataplexy, and the effects of D2/D3 stimulation were reversed by dopamine antagonists. Ionophoretic injections of apomorphine into the VTA also induced sleep in rats.37 Therefore, DA may cause sleepiness at dopaminergic sites other than the nigrostriatal system.

Falling asleep while driving is surprisingly common in the general population. Surveys show a frequency of 6% to 48.1%.7,38-41⇓⇓⇓⇓ The highest figures are reported by long haul truck drivers. The percentage of accidents attributed to sleepiness also varies markedly but may be as high as 30%.42 Motor vehicle accidents are also much more common in patients with sleep disorders, such as sleep apnea, occurring as much as 15 times that of the normal population.39 Therefore, the frequency of falling asleep while driving in our patients with PD, while probably greater than that of the normal population, is not higher than that seen in people with sleep disorders.

Our study suffers from several potential weaknesses. First, our patients tend to be more severe and may not be representative of all patients with PD. Second, although we did not change our prescription patterns after the initial “sleep attack” report,2 some referring physicians may have changed medications if patients reported sedation, thus biasing results. Third, pramipexole was used more than other DA in this group, possibly reducing the power of comparison among the three groups. Fourth, reporting of sleep characteristics is subject to recall bias. Fifth, we did not use polysomnography or other objective measures to quantify sleep parameters. Finally, we were unable to statistically account for the large number of other non-PD medications used by our patients that could affect alertness. Despite these limitations, we feel that our results demonstrate that daytime drowsiness and falling asleep while driving are common in PD, and that they correlate with both more advanced disease and the use of all dopaminergic drugs. We therefore recommend that patients be educated about the soporific effects of both PD and medications for PD, but we have not altered our prescribing practices among the individual medications.