We read with interest the recent article by Shpirer et al. 1 regarding polysomnographic evaluation of patients with excessive daytime sleepiness and Parkinson's disease (PD). Their study found no correlation between objective measures of daytime sleepiness in multiple sleep latency testing (MSLT) and subjective daytime sleepiness as measured by the Epworth Sleepiness Scale (ESS). They also found no effect of dopamine agonist therapy on MSLT results. In contrast, other studies of PD have shown relationships between MSLT-defined sleepiness and such measures as medication, sleep architecture, and disease duration. [2][3][4] All the aforementioned studies using MSLT in PD have used a single day of MSLT recording data, with the implicit assumption that sleepiness in PD is a relatively stable trait. It is conceivable that some of the discrepancies in studies using MSLT data could be caused by a lack of stability in this trait in PD. In our laboratory, we have had the opportunity to study PD patients using MSLT on two consecutive days, allowing us to examine the degree of short-term stability of this trait. We report below on these patients, who were not among our original case series. 3 Nine patients with idiopathic PD and excessive daytime sleepiness reported by their significant others (Significant Other Epworth Sleepiness Scale [SOESS] Ͼ 10) were studied as part of a larger investigation of daytime sleepiness in PD. Mean (SD) age was 61 (9) and mean (SD) duration of PD was 10 years (4.5). There were 7 men and 2 women. Mean (SD) SOESS was 17 (1.6). Hoehn-Yahr scores were I (n ϭ 3), II (n ϭ 1), II.5 (n ϭ 4), and III (n ϭ 1). Mean (SD) scores on UPDRS (motor subscale), Geriatric Depression Scale, and Mini-Mental State Exam were 21 (7.5), 4 (3.8), and 27 (2.9), respectively. All patients were on a combination of carbidopa/ levodopa (mean daily levodopa dose 542 mg, SD 355) and dopamine agonist therapy (mean pergolide equivalents 1.5 mg, SD 1.0).
MSLTs were performed at 8 am, 10 am, noon, 2 pm, and 4 pm using conventional procedures. Each test was conducted for a maximum of 20 minutes (if no sleep occurred) or was terminated at 15 minutes subsequent to the first 30-second epoch scored as sleep. MSLTs were performed on two consecutive days. Medication type and dosages were constant across days. MSLTs were analyzed as the time to reach the first epoch of Stages 1, 2, or REM sleep. All recordings were independently scored by two board-certified sleep specialists. We computed daily medians for each patient's 5 nap opportunities on both Day 1 and Day 2.
For Scorer A, the mean (SD) sleep latencies on Days 1 and 2 were 6.1 (6.4) and 5.8 (6.0) minutes, respectively, yielding a moderately high interday reliability (Spearman ϭ 0.73; P ϭ 0.03). For Scorer B, the mean (SD) sleep latencies on Days 1 and 2 were 9.3 (6.7) and 8.2 (7.4) minutes, respectively, yielding a somewhat lower interday reliability of (Spearman ϭ 0.53; P ϭ 0.14). To examine possible inter-rater discrepancy in scoring more carefully, we examined the extent of inter-rater discrepancy on the basis of individual naps on each day. Out of 90 nap opportunities, sleep onset was specified within 5 minutes or less on 81% of all naps.
Our findings suggest that MSLT-defined sleepiness in a PD sample retains a modest degree of stability across two consecutive days of testing. However, discrepancies among studies of this phenomenon may be caused by variability from day-to-day and between scorers. Other approaches to examining sleepiness/alertness in this population may be warranted, such as the Maintenance of Wakefulness Test.