Reliability of a single objective measure in assessing sleepiness

Abstract

Study objectives: To evaluate reliability of single objective tests in assessing sleepiness.

Design: Subjects who completed polysomnography underwent a 4-nap multiple sleep latency test (MSLT) the following day. Prior to each nap opportunity on MSLT, subjects performed the psychomotor vigilance test (PVT) and divided attention driving task (DADT). Results of single versus multiple test administrations were compared using the intraclass correlation coefficient (ICC) and adjusted for test administration order effects to explore time of day effects. Measures were explored as continuous and binary (i.e., impaired or not impaired).

Setting: Community-based sample evaluated at a tertiary, university-based sleep center.

Participants: 372 adult commercial vehicle operators oversampled for increased obstructive sleep apnea risk.

Interventions: N/A.

Measurements and results: AS CONTINUOUS MEASURES, ICC WERE AS FOLLOWS: MSLT 0.45, PVT median response time 0.69, PVT number of lapses 0.51, 10-min DADT tracking error 0.87, 20-min DADT tracking error 0.90. Based on binary outcomes, ICC were: MSLT 0.63, PVT number of lapses 0.85, 10-min DADT 0.95, 20-min DADT 0.96. Statistically significant time of day effects were seen in both the MSLT and PVT but not the DADT. Correlation between ESS and different objective tests was strongest for MSLT, range [-0.270 to -0.195] and persisted across all time points.

Conclusions: Single DADT and PVT administrations are reliable measures of sleepiness. A single MSLT administration can reasonably discriminate individuals with MSL < 8 minutes. These results support the use of a single administration of some objective tests of sleepiness when performed under controlled conditions in routine clinical care.

Keywords: Excessive sleepiness; divided attention driving task; multiple sleep latency test; objective tests; psychomotor vigilance test.

Figure 1

ICC values of MSLT (latency), PVT (raw number of lapses), DADT (tracking error) at 10 min and 20 min, explored as both a continuous (white) and binary measure (gray), showing highest reliability for the DADT.

Figure 2

Time of day effects of continuous measures. Change in MSLT mean sleep latency (squares), PVT number of lapses (triangles), and DADT tracking error (circles), across test administrations from time 1 to time 4. The DADT shows the least change across the day.

Figure 3

Time of day effects of binary measures. Change in impairment on MSLT defined by latency < 8 minutes (squares), PVT defined by number of lapses > 3 (triangles), and DADT defined by tracking error > 250 cm (circles) across test administrations from time 1 to time 4. The DADT continues to show little change. See Supplemental S3.