Sleep Scientists Defend Vigilance Test Cutoff in Treated Sleep Apnea Debate

A technical dispute over how to measure lingering daytime sleepiness in patients with treated obstructive sleep apnea has erupted into a revealing exchange about the limits of clinical measurement. In a formal reply published in the Journal of Clinical Sleep Medicine, researchers from the University of Miami have responded to a critique of their MAGNETO study, defending their use of a widely cited threshold on the psychomotor vigilance task while conceding that the field still lacks a properly validated cutoff for identifying residual excessive daytime sleepiness. The exchange, though narrow in scope, touches on one of the most consequential questions in sleep medicine: how to detect the patients who remain dangerously sleepy even after their breathing is successfully treated.

Obstructive sleep apnea affects hundreds of millions of people worldwide, causing the upper airway to collapse repeatedly during sleep and fragmenting rest night after night. The standard treatment, positive airway pressure therapy, keeps the airway open and, for most patients, resolves the hallmark symptom of excessive daytime sleepiness. Yet a stubborn minority continue to struggle with profound drowsiness despite excellent adherence to therapy, a condition known as residual excessive daytime sleepiness. These patients face elevated risks of motor vehicle accidents, impaired workplace performance, and, as emerging neuroimaging research suggests, possible structural changes in the brain’s white matter. Identifying them reliably is therefore a pressing clinical priority.

The psychomotor vigilance task, or PVT, has become the field’s most trusted laboratory instrument for quantifying the behavioral consequences of sleep loss. In its standard form, participants sit before a screen for roughly ten minutes and press a button the instant a counter appears at an unpredictable interval. The test captures lapses in attention, defined as reaction times exceeding a specified duration, along with mean reaction time and the variability of responses over time. Because the task depends minimally on learned skill or education, it is considered a relatively pure index of the brain’s underlying capacity for sustained attention, making it attractive for comparing sleepiness across diverse populations.

In their original MAGNETO study, Junco, Ramos, Hernandez-Cardenache and colleagues examined cognition and psychomotor vigilance in treated sleep apnea patients with and without daytime sleepiness. The sample was notable for its composition: primarily Hispanic and Latino adults, a group that has been chronically underrepresented in sleep research despite carrying a substantial burden of sleep-disordered breathing. The researchers adopted a threshold of more than five lapses on the vigilance task to define residual sleepiness, drawing on prior studies of positive airway pressure-adherent populations. That choice allowed their findings on the neurological correlates of residual sleepiness to be compared directly with earlier work that had documented white matter differences in persistently sleepy patients.

The critique from Fei and colleagues, published as a letter in the same journal, questioned whether that cutoff had ever been properly validated. Their concern was methodologically sound: a threshold adopted from prior literature is not the same as one demonstrated, through formal receiver operating characteristic analysis, to accurately separate truly sleepy patients from alert ones against a cognitive or clinical gold standard. Receiver operating characteristic analysis, a statistical technique borrowed from signal detection theory, evaluates how well a continuous measure discriminates between two states across all possible cutoffs, yielding a curve whose area summarizes discriminative power. Without such validation, any single lapse threshold risks being arbitrary, and conclusions tied to it may not generalize.

In their reply, the Miami team addressed the criticism on three fronts. First, they clarified the intent of the original study, emphasizing that MAGNETO was designed as an exploratory, hypothesis-generating analysis of an underrepresented sample and was never presented as a validation study for the vigilance task cutoff. The threshold, they explained, was adopted specifically to enable comparison with extant findings on residual sleepiness and its neurological correlates, including prior magnetic resonance imaging studies that had linked persistent sleepiness in treated apnea patients to white matter structural differences. Framing matters here: an exploratory analysis that borrows an established convention is playing a fundamentally different scientific role than a diagnostic validation study.

Second, the authors pointed out that the robustness of their findings across alternative thresholds had already been tested. Sensitivity analyses using different lapse cutoffs appear in Supplemental Table S2 of the published article, and those analyses demonstrated that the primary results held regardless of which specific threshold was applied. This is an important rebuttal, because a finding that depends entirely on one arbitrary cutoff is fragile, whereas one that survives across a range of reasonable thresholds is far more credible. The exchange thus illustrates a practice that rigorous researchers increasingly demand: showing that conclusions are not artifacts of analytic choices, a principle sometimes described as multiverse thinking in statistical methodology.

Third, and perhaps most significantly, the authors revealed that the categorical grouping of patients by sleepiness status was never the centerpiece of their analysis. The primary analytic approach was continuous: total lapses, mean reaction time, and the coefficient of variation of reaction time were all modeled as continuous predictors in regression analyses adjusted for sex, time since diagnosis, and self-reported sleep duration. Treating vigilance measures as continuous variables preserves the full richness of the data and avoids the information loss and arbitrary boundary problems inherent in dichotomizing patients into sleepy and non-sleepy groups. The categorical definitions, they noted, served descriptive and supplementary purposes only, which substantially blunts the force of the critique.

Despite their defense, the authors extended an olive branch. They agreed with Fei and colleagues that receiver operating characteristic-based cutoff validation targeting cognitive endpoints, together with standardized reporting of continuous vigilance metrics, represents a worthwhile direction for future research on residual sleepiness, particularly as larger samples become available. That concession reflects a genuine tension in the field. Clinicians need categorical decisions: does this patient warrant additional evaluation or wakefulness-promoting therapy? Researchers, meanwhile, increasingly recognize that sleepiness exists on a continuum and that forcing it into binary categories can obscure dose-response relationships and inflate apparent effects. Reconciling these two needs will require large, well-characterized cohorts in which vigilance performance can be anchored to meaningful outcomes such as crash risk, cognitive decline, or objective sleepiness measures.

The broader lesson of this scholarly exchange extends well beyond one cutoff on one test. Sleep medicine is moving toward precision approaches in which residual symptoms in treated patients are taken seriously as signals of ongoing brain vulnerability rather than dismissed as treatment failures. Studies like MAGNETO, conducted in populations that have historically been left out of sleep research, are essential for ensuring that the resulting diagnostic standards work across demographic groups, since reaction time distributions and symptom reporting can vary with language, education, and cultural context. The Miami authors’ willingness to defend their methodology transparently while endorsing stricter validation standards for the future models the kind of constructive scientific dialogue on which reliable clinical progress depends. For the millions of patients who remain sleepy despite successful treatment of their breathing, the outcome of this methodological debate is anything but academic: it will determine who gets recognized, and treated, before the consequences of unrecognized drowsiness strike on the road or in the operating room.

Subject of Research: Psychomotor vigilance task cutoffs for residual excessive daytime sleepiness in treated obstructive sleep apnea

Article Title: Reply to “Validating the psychomotor vigilance task cutoff for residual excessive daytime sleepiness in treated obstructive sleep apnea”

Article References: Junco, B., Ramos, A., & McIntosh, R. (2026). Reply to “Validating the psychomotor vigilance task cutoff for residual excessive daytime sleepiness in treated obstructive sleep apnea”. Journal of Clinical Sleep Medicine, 22(1), Article 101. https://doi.org/10.1007/s44470-026-00115-6

Image Credits: AI Generated

DOI: 10.1007/s44470-026-00115-6

Keywords: obstructive sleep apnea, excessive daytime sleepiness, psychomotor vigilance task, PVT cutoff, MAGNETO study, CPAP adherence, continuous positive airway pressure, reaction time variability, ROC analysis, white matter changes, sleep medicine, Hispanic and Latino health