Longer Sleep Latency Time Observed With Recent Definition of Sleep Onset, Meta-Analysis Shows
A recent systematic review and meta-analysis showed that normal mean sleep latency in adults was 11.7 minutes using the earlier definition of sleep onset and 11.8 minutes using the later definition of sleep onset.
Mark I. Boulos, MD, BSc, FRCP, CSCN, MSc
In a recent systematic review and meta-analysis published in Sleep Medicine, findings showed a longer average of mean sleep latency among healthy adults using a more updated later definition of sleep onset compared. Investigators concluded that establishing updated ranges for mean sleep latency may guide decision-making surrounding sleep pathologies and better inform research in the future of the associations between patient variables, daytime sleepiness, and sleep pathologies.1
Among 110 cohorts involving 4058 healthy adults, the average mean sleep latency was 11.7 min (95% CI, 10.8–12.6; 95% PI, 5.2–18.2) for the studies assessed with the earlier definition of sleep onset and 11.8 min (95% CI, 10.7–12.8; 95% PI, 7.2–16.3) for those evaluated using the later definition. Despite no significant associations between mean sleep latency and demographic or methodological variables, a negative association of −0.29 per one unit increase (95% CI, −0.55 to −0.04) was observed between mean sleep latency and apnea-hypopnea index on prior night polysomnography.
Senior author Mark I. Boulos, MD, BSc, FRCP, CSCN, MSc, associate professor, department of medicine, division of neurology, Institute of Medical Science, University of Toronto, and colleagues, provided an updated review of normative mean sleep latency values on the multiple sleep latency test (MSLT) in an analysis of healthy adult cohorts that ranged from 1968 to 2016. Researchers also investigated the impact of demographic variables including age, sex, body mass index, sleep architecture and sleep-disordered breathing as well as methodological variables like sleep onset definitions and multiple sleep latency test protocols.
Researchers evaluated the control cohorts on 5 Likert (0–2) rating scales, rating how well the patients were screened for medical diseases, sleep-related illnesses, psychiatric illnesses, drug use, and representativeness of recruitment from the general population. The data from the studies, as well as the associated Likert scores of the control groups, were reviewed by 2 study authors. Estimates of the mean sleep latency from the studies were pooled using a random effects generic inverse meta-analysis with the means and standard errors used as input parameters. Authors then further reviewed reported MSL values from data in the pathological ranges seen in narcolepsy, since such values may reflect unmeasured prior sleep restriction or familiarity with sleep environment.
In the analysis, 60% (n = 66) of healthy control groups completed a 4-trial MSLT and 40.0% (n = 44) completed a 5-trial MSLT. Notably, 48% (n = 53) of healthy control groups had their MSLT scored using the earlier definition of sleep onset2 and 19% (n = 21) using the later definition3. The definition was not specified for the remaining healthy control groups, (28.2%, n = 31), or they used an alternative definition (4.5%, n = 5).
In the single-variable mixed effects models, there was a significant inverse association observed between the mean sleep latency and the apnea-hypopnea index recorded on polysomnography performed the night prior for 19 healthy control groups. In the multivariable model, there were no significant associations observed with the mean sleep latency. In a sensitivity analysis which excluded cohorts that used an unknown or modified definition of sleep onset, the mean apnea-hypopnea index was not significantly associated with the mean sleep latency, although this subgroup included only 8 healthy control groups.
Study limitations authors noted include the possibility that patients had undiagnosed or unreported comorbidities that led to the sample being more heterogeneous than expected. Also, 32% (n = 36) of included cohorts did not specify their definition of sleep onset. The older adult patients were notably underrepresented in the sample with only 4.5% (n = 5) of the included cohorts having a known mean age of greater than or equal to 65. In addition, the infrequent reports of the mean body mass index (BMI), which were available for 32.7% (n = 36) of included control groups, and incomplete range of possible values (range, 20.9-32.0) may have made it difficult to reveal a BMI effect. The search also was restricted to 2016 which may have limited the number of included studies.
