Test-retest reliability of the 5-minute psychomotor vigilance task in working-aged females.

Background: The psychomotor vigilance task (PVT) is a commonly used test that effectively assesses neurobehavioral alertness. The originally developed PVT is 10 minutes in duration, which presents practical and logistical issues, particularly when administered to large samples or on a repetitive basis. More recently the PVT has been used in both 3- and 5-minute formats. While both of these durations have been shown to be field sensitive to identify impairments from sleep- and fatigue-related interventions, the 5-minute version has been suggested to be more valid than the 3-minute. However, while these have shown field-validity in a number of working populations, there is a paucity of data reporting the test-retest reliability statistics of the 5-minute PVT, particularly in working-aged females. The purpose of the study was to examine the test-retest reliability of a comprehensive set of PVT variables for the 5-minute PVT in a population of working-aged females (20-63 years).

New Method: Participants reported to the laboratory on two separate days and performed a 5-minute PVT on each occasion. Outcome measures included the mean reaction time (MRT), fastest and slowest 10% of reaction times (F10RT% and S10RT%, respectively), standard deviation of reaction times (SDRT) as well as error-based metrics including major and minor lapses, anticipations, and false starts. In addition, total errors (ERR) were computed as a composite of all types of errors. Reliability statistics were reported as intraclass correlation coefficients (ICCs), standard error of measurement (SEM, SEM%), and minimal difference to be considered real (MD, MD%). Systematic error was also evaluated between sessions.

Results: Overall high reliability was shown for the MRT (ICC = 0.79, SEM% = 4.14%) and F10RT% (ICC = 0.83, SEM% = 4.43%) variables, with moderate relative reliability (based on ICCs) for the false starts, ERR, and sleepiness scale (ICC = 0.50 - 0.70) variables but these all exhibited poor absolute reliability (based on SEM% values of 32.60 - 168.69%). Poor relative reliability was found for the SDRT and S10RT% variables (ICCs < 0.50) but the S10RT% variable had good absolute reliability (SEM% = 7.12%). The minor and major lapses and anticipations variables had too few of event occurrences for a confident determination of the reliability.

Conclusions: The finding that the MRT variable displayed systematic error (P = 0.01) indicating that a learning curve may have been present, but the F10RT% did not show systematic error, suggests the F10RT% may be the most reliable PVT variable in a 5-minute duration test. These findings provide researchers and practitioners with reliability statistics that may help in determining which variable(s) to use, and which to avoid when specifically conducting 5-minute PVT assessments, particularly in a population of working-aged females. These results suggest that the 5-minute PVT can be used in place of the 10-minute version, if used appropriately.