Impact of Clinical Assessment on Use of Data from Unattended Limited Monitoring as Opposed to Full-in lab PSG in Sleep Disordered Breathing

Abstract

Study objectives We examined agreement among multiple sleep clinicians when presented with clinical data plus the full tracings and data obtained from unattended limited monitoring (ULM) or a full polysomnography (PSG). Methods Subjects included 66 patients with complaints of sleep disordered breathing (SDB) and 19 volunteers willing to undergo 2 nights of ULM followed by PSG. Two assessment packages were created for each subject with identical clinical history (Hx) and ARES Symptom Questionnaire, plus the electronic record of signals collected on the ARES Unicorder (Hx+ULM) or on the PSG (Hx+PSG). Data were presented to 4 sleep-trained clinicians for diagnosis and treatment recommendation. For agreement on diagnosis and treatment, comparisons were made between clinicians using ULM or PSG, and within clinicians comparing both techniques. Results For diagnosis, agreement between pairs of clinicians using Hx+PSG ranged from 74% to 86% and 66% to 85% when using Hx+ULM. For treatment, agreement using Hx+PSG ranged from 74% to 86% and 58% to 77% when using Hx+ULM. Agreement between clinicians was highest in the subjects with the highest RDI and fell off markedly at the lowest RDI, irrespective of whether the clinicians used the Hx+PSG or Hx+ULM. This pattern was also seen for the decisions made by an individual clinician using Hx+ULM vs. Hx+PSG. Conclusion Our data show that sleep clinicians have significant disagreements for diagnosis even when presented with the "gold standard" of a PSG and clinical data. Agreement was high when the SDB index was elevated and lower when the SDB index was in the mild-to-moderate range, regardless of the technique used to obtain it.