Mean sleep latency is determined from the MWT by calculating the average sleep-onset latency from all four trials. Like the MSLT and overnight polysomnogram, sleep onset is defined as the start of the first thirty-second epoch scored as sleep [19]. Normative data guide interpretation of mean sleep latency. In 64 healthy controls, when defining sleep onset as at least ten seconds of sleep or the first epoch scored as sleep, mean sleep latency was 32.6 ± 9.9 min and 42 % of subjects remained awake for the entirety of all four 40-min trials of the MWT [20]. When a more stringent definition of sleep onset was used (three continuous epochs of N1 sleep, or one epoch of another sleep stage), mean sleep latency was 35.2 ± 7.9 min and 59 % of subjects remained awake for the entirety of all four trials [20]. Additionally, 97.5 % of normal subjects had a mean sleep latency ≥8 min [10]. Based on these data and a subsequent investigation in normal controls, the AASM recommends interpretation of mean sleep latency less than 8 min as abnormal and mean sleep latency between 8 and 40 min as a value of uncertain significance [21, 10]. Remaining awake for the full 40 min on each trial provides the greatest objective evidence to support an individual’s ability to maintain alertness [10].

Mean sleep latency may be affected by multiple factors. A significant, positive correlation between age and mean sleep latency is seen with the 40-min but not 20-min MWT protocol. This suggests that the ability to remain awake after the first 20 min of an MWT trial may be age-dependent [20]. Furthermore, medications, caffeine, prior total sleep time, sleep fragmentation, circadian phase, anxiety, depression, and motivation have all been shown to affect MWT results [22–35]. Even playing background music can increase mean sleep latency on the MWT [36]. Contrary to what may be expected, napping between MWT trials did not significantly impact the results in one investigation [37].

At this time, the MWT is not recommended in the pediatric population secondary to lack of data in this group including undefined normative values. However, a recent study did demonstrate that the MWT may be beneficial in assessing treatment responses of pediatric patients with narcolepsy [38]. The MWT has been evaluated in other specific patient populations, for example, patients with retinitis pigmentosa, myotonic dystrophy, and more extensively individuals with Parkinson’s disease [22, 39, 40, 41]. The MWT may be a valuable assessment tool in patients with Parkinson’s disease, particularly to assess the ability to maintain alertness in the setting of dopaminergic medication use [22, 41].

The MWT is recommended as a tool to assess treatment response in individuals receiving therapy for hypersomnolence. The magnitude of change that represents adequate treatment is not defined, but the direction of change may be valuable when combined with clinical assessment [10]. Studies of patients with narcolepsy, idiopathic hypersomnia, obstructive sleep apnea, and shift work disorder have shown varying degrees of increased mean sleep latency on the MWT after treatment.

Additionally, the MWT may be used to evaluate individuals in whom impaired alertness could create risk to personal or public safety [10]. Studies have used driving simulators to investigate the relationship of mean sleep latency with performance, and their results are as follows. Subjects with disorders that cause hypersomnolence demonstrate significant negative correlations between mean sleep latency and the following measures of driving impairment: lane-position variability, crash frequency, inappropriate line crossing, and standard deviation from the center of the road [42–44]. Some studies have shown the greatest driving impairments in those with mean sleep latencies below 19 min [42]. Conversely, untreated obstructive sleep apnea patients with mean sleep latencies above 30 minutes perform similar to healthy controls on measures of vehicle control and vigilance [43]. Notably, in one investigation, driving simulator performance showed significant improvements when obstructive sleep apnea was treated with continuous positive airway pressure (CPAP) [45]. However, this did not correlate with increases in mean sleep latency [45]. Furthermore, in a group of subjects undergoing partial sleep loss, only those who were also impaired by alcohol demonstrated a significant inverse correlation between simulator car crashes and mean sleep latency on the MWT [46]. Mean sleep latencies derived from the MWT show correlations with simulated driving performance that are superior to the MSLT [43, 44].

Significant limitations arise in evaluation of sleepiness in real drivers on actual roadways. A study that did examine actual highway driving showed a significant increase in inappropriate line crossings among very sleepy (mean sleep latency 0–19 min) and sleepy (mean sleep latency 20–33 min) untreated obstructive sleep apnea patients, as compared to alert patients and controls (mean sleep latency > 33 min) (Fig. 23.1, [47]. However, line crossings may not necessarily translate directly to automobile crashes. Another study found that drivers who had been in a motor vehicle accident, in comparison with those who had not, did not show significantly different MWT results [48]. Due to the paucity of data that address real-life fitness for operating a motor vehicle, the MWT is not generally required for commercial vehicle drivers with sleep disorders [49]. Furthermore, in contrast to the MWT findings on simulated driving performance, mean sleep latency derived from the MSLT has been associated with motor vehicle accidents, as demonstrated by an increased risk of car crashes in individuals with mean sleep latencies less than or equal to 5 min on MSLT [50].

Data on performance in aviators are scarce, but two pilots with hypersomnolence and short sleep latencies on MSLTs were successfully returned to duty after they demonstrated the ability to stay awake on all naps of the MWT [51]. Currently, the Federal Aviation Administration (FAA) does employ the MWT to ensure alertness in pilots with obstructive sleep apnea (www.​faa.​gov).

In addition to limitations in assessment of safety and performance, the MWT has other shortcomings. Compared to MSLT, less validity data exist for the MWT and clinicians tend to have less experience with this test. Although highly sensitive to sleepiness produced by acute, severe sleep deprivation, the linear relationship of mean sleep latency with sleep loss is attenuated in the setting of less severe, more chronic sleep deprivation [24]. Also, microsleep episodes, which are associated with performance decrements, are not typically reported in the interpretation of the MWT [52]. The MWT is not designed to aid in the diagnosis of narcolepsy. Sleep-onset REM periods are significantly decreased on the MWT compared to the MSLT, and use of the MWT in patients with suspected narcolepsy would lead to false-negative results [53]. The range of mean sleep latencies in the category of unclear clinical significance (between 8 and 40 min) is large, with the potential for overlap of normal individuals, those with sleep disorders, and patients with treated and untreated conditions. Furthermore, the ability to remain awake for all trials on the MWT may not reflect alertness in real-life situations where confounding factors such as prior sleep duration and circadian phase contribute to sleepiness, and motivation to stay awake may differ from that experienced during a laboratory test. For the MWT, as with the MSLT, the lack of prospective data to show prediction of future crashes and morbidity remains a significant limitation and area for further research.

The maintenance of wakefulness test is a tool designed to assess the ability to sustain alertness rather than the tendency to the fall asleep. This test may be beneficial to evaluate response to therapy in individuals with hypersomnolence. In addition, the maintenance of wakefulness test may provide valuable information when performed in patients whose occupations require constant alertness to ensure safety. However, like all diagnostic modalities, careful clinical correlation is required with interpretation of this test as results that fall into normal ranges may not guarantee alertness under other conditions.