Validity and Sensitivity of a Brief Psychomotor Vigilance Test (PVT-B) to Total and Partial Sleep Deprivation

Abstract

The Psychomotor Vigilance Test (PVT) objectively assesses fatigue-related changes in alertness associated with sleep loss, extended wakefulness, circadian misalignment, and time on task. The standard 10-min PVT is often considered impractical in applied contexts. To address this limitation, we developed a modified brief 3-min version of the PVT (PVT-B). The PVT-B was validated in controlled laboratory studies with 74 healthy subjects (34 female, aged 22-45 years) that participated either in a total sleep deprivation (TSD) study involving 33 hours awake (N=31 subjects) or in a partial sleep deprivation (PSD) protocol involving 5 consecutive nights of 4 hours time in bed (N=43 subjects). PVT and PVT-B were performed regularly during wakefulness. Effect sizes of 5 key PVT outcomes were larger for TSD than PSD and larger for PVT than for PVT-B for all outcomes. Effect size was largest for response speed (reciprocal response time) for both the PVT-B and the PVT in both TSD and PSD. According to Cohen's criteria, effect sizes for the PVT-B were still large (TSD) or medium to large (PSD, except for fastest 10% RT). Compared to the 70% decrease in test duration the 22.7% (range 6.9%-67.8%) average decrease in effect size was deemed an acceptable trade-off between duration and sensitivity. Overall, PVT-B performance had faster response times, more false starts and fewer lapses than PVT performance (all p<0.01). After reducing the lapse threshold from 500 ms to 355 ms for PVT-B, mixed model ANOVAs indicated no differential sensitivity to sleep loss between PVT-B and PVT for all outcome variables (all P>0.15) but the fastest 10% response times during PSD (P<0.001), and effect sizes increased from 1.38 to 1.49 (TSD) and 0.65 to 0.76 (PSD), respectively. In conclusion, PVT-B tracked standard 10-min PVT performance throughout both TSD and PSD, and yielded medium to large effect sizes. PVT-B may be a useful tool for assessing behavioral alertness in settings where the duration of the 10-min PVT is considered impractical, although further validation in applied settings is needed.

Figure 1

Relative frequency distributions of PVT response time are shown for alert (bouts 1 to 7 during total sleep deprivation and baseline 2 during partial sleep deprivation) and sleep deprived states (bouts 8 to 17 during total sleep deprivation and restriction nights 1 to 5 during partial sleep deprivation) for both the modified 3-min (PVT-B) and the 10-min version of the PVT. The insert shows the frequency of false starts (errors of commission) including 95% confidence intervals.

Figure 2

Effect sizes ± 95% bootstrap confidence intervals are compared between the standard 10-min PVT (PVT) and the modified 3-min version of the PVT (PVT-B) for 5 outcome metrics and for both total (left) and partial (right) sleep deprivation. For PVT-B, both standard 500 ms and modified 355 ms lapse thresholds were applied. Ranges representing small (S, >0.2 and <0.5), medium (M, >0.5 and <0.8), and large (L, >0.8) effect sizes according to Cohen [41] are indicated by black horizontal lines. The relative decrease in effect size from PVT to PVT-B is indicated as percentages above each outcome metric.

Figure 3

For each of the 6 outcome variables, between-subject averages (N=31 subjects) are shown for each of the 17 tests performed during a 33 h period of total sleep deprivation for both the 10-min (black circles) and the 3-min (open circles) PVT. Error bars represent 95% BCa confidence intervals based on a bootstrap sample with 1,000,000 replications. In the right column of the figure, the 5 outcome variables of the 3-min and the 10-min PVT were centered around alert performance (average of test bouts 1 to 7). Paired t-tests were performed on each of test bouts 8 to 17 during sleep deprivation to test whether the modified 3-min (PVT-B) and the 10-min PVT differed statistically significantly. * p<0.05, ** p<0.01, *** p<0.001 (adjusted for multiple testing)

Figure 4

For each of the 6 outcome variables, between-subject averages (N=43 subjects) are shown for each of the 7 conditions of the partial sleep deprivation protocol (BL = baseline, R = restriction) for both the modified 3-min PVT-B (open circles) and the 10-min PVT (black circles). Error bars represent 95% BCa confidence intervals based on a bootstrap sample with 1,000,000 replications. In the right column of the figure, the 5 outcome variables of the 3-min PVT-B and the 10-min PVT were centered around alert performance (BL2). Paired t-tests were performed on each of the five sleep restriction conditions (R1 – R5) to test whether the 3-min and the 10-min PVT differed statistically significantly. * p<0.05, ** p<0.01, *** p<0.001 (adjusted for multiple testing)