Validity of the Pupillographic Sleepiness Test for the diagnosis of daytime sleepiness in children and adolescents and its relationship to sleepiness-associated outcomes.

OBJECTIVES

To report validation data for the Pupillographic Sleepiness Test (PST) in children and adolescents, evaluate its applicability for diagnosing excessive daytime sleepiness and its relationship to sleepiness-associated outcomes.

METHODS

A cross-sectional diagnostic test accuracy study was performed. Patients underwent three PST at 9 a.m. (T1), 11 a.m. (T2) and 1 p.m. (T3) plus the Multiple Sleep Latency Test (MSLT) on a single day. Additionally, two neurocognitive tests were performed and three questionnaires about quality of life, sleep-related self-efficacy and behavioural aspects completed. Gender-stratified z-values of the natural logarithm of the Pupillary Unrest Index (z-lnPUI) were correlated to Sleep Latency (SL) and Mean Sleep Latency (MSL) and to variables of neurocognitive tests and questionnaires using Spearman's rank correlation. Cut-off values were determined by receiver operating characteristic (ROC) analysis.

RESULTS

47 patients were recruited (median 10.6 years, range 6-18). Correlation between z-lnPUI and SL was rT1 = -0.373 (p = 0.011); rT2 = -0.320 (p = 0.028) and rT3 = -0.336 (p = 0.022). Correlation between z-lnPUI and MSL was rT1 = -0.338 (p = 0.020); rT2 = -0.202 (p = 0.173); rT3 = -0.117 (p = 0.433). ROC analysis showed an area under the curve of 90.7% and PUI cut-off values of 12.6 mm/min (boys) and 11.6 mm/min (girls). There were moderate correlations between z-lnPUIT1 and reaction time and omission errors in neurocognitive tests (r = 0.394, p = 0.007 and 0.391, p = 0.008).

CONCLUSIONS

We found satisfactory correlations between PST and MSLT results. The z-lnPUIT1 was related to MSL and objective measures of attention ability. Given this accuracy, the PST may be used as a screening tool for evaluating daytime sleepiness in children and adolescents. Corresponding gender-related reference values are now available.