A randomised controlled trial comparing patient outcomes following full polysomnography versus limited sleep study testing for suspected obstructive sleep apnea.

Abstract

RATIONALE: Although studies have shown reasonable agreement between full polysomnography (PSG) and limited-channel sleep studies in the number of disordered breathing events identified, the clinical utility of simplified testing and resulting patient outcomes remain unclear. The aim of this study was to compare treatment outcomes for patients managed using PSG versus limited sleep studies for suspected obstructive sleep apnea (OSA). METHOD(S): Patients aged 25 to 80 years referred to one of seven sleep centres with suspected OSA who underwent full laboratory PSG were randomly assigned to one of three groups which determined the level of sleep study information to be disclosed to treating sleep physicians: 1) full PSG; 2) level 3 (L3), i.e. airflow, thoraco-abdominal bands, body position, ECG and oxygen saturation; or 3) level 4 (L4), i.e. oxygen saturation and heart rate, for 4 months follow-up. Outcome measures included the change in Functional Outcomes of Sleep Questionnaire (FOSQ), Epworth Sleepiness Scale (ESS), and Sleep Apnea Symptoms Questionnaire (SASQ), continuous positive airway pressure (CPAP) adherence, and physician diagnostic confidence and decision making. RESULT(S): 406 patients were randomised. For the primary outcome, change in FOSQ, L3 and L4 were not inferior to PSG using an a priori non-inferiority margin of -1.0 (adjusted mean difference [AMD]: 0.01, lower bound 1-sided 95%CI -0.43, p=0.97 [level 3]; and -0.46, lower bound 95%CI -0.90, p=0.08 [level 4]). For the change in ESS, L3 was not inferior to PSG using a non-inferiority margin of 2.0 (AMD -0.02, upper bound 95%CI 0.9, p=0.97), however L4 was inferior to PSG (AMD 1.4, upper bound 95%CI 2.3, p=0.01). No difference was observed between PSG and L3 for change in SASQ and mean nightly CPAP use. However, compared to PSG, there was less improvement in SASQ (-24.3 vs -15.8, p=0.01) and reduced CPAP use (5.3 vs 4.5 hours/night, p=0.04) with L4. Although a similar proportion of diagnoses and treatment recommendations were made by physicians across the 3 arms, diagnostic confidence was lower for L4 (Pearson chi2=23.0, p=0.003, see Table). CONCLUSION(S): Management of suspected OSA by a sleep physician using level 3 data was not clinically inferior to PSG. Although level 4 testing was not inferior to PSG for the primary outcome (change in FOSQ), symptom improvement (by ESS and SASQ) and CPAP adherence were worse. The reasons for this are unclear, however, it is possible that reduced physician confidence when presented with only level 4 data may have impacted on patient outcomes. (Table Presented).