Relative accuracy of algorithm-based prescription of nasal CPAP in OSA

Respir Med. 2004 Feb;98(2):152-4. doi: 10.1016/j.rmed.2003.09.009.


Background: Patients with OSA on nasal continuous positive airway pressure (CPAP) have considerable night-to-night variation in their pressure requirements, suggesting that a one-night titration might not be very precise. This study investigates the likely error incurred using a one-night titration, and explores whether an algorithm-based approach to determine the pressure is as accurate.

Methods: Thirty patients with OSA used an autotitrating CPAP device for 28 nights and the average was regarded as the 'reference' pressure for that patient. Using estimates of precision and bias, this 'reference' pressure was compared with (1) an algorithm-derived pressure (based on neck circumference and OSA severity), (2) a one-night titration (using four alternative nights), and (3) a fixed pressure of 10 cmH2O.

Results: The mean 'reference' pressure for the group was 9.83 (SD 2.12) cmH2O. There was little bias from any of the alternatives. However, the precision varied between 1.65 and 2.45 cmH2O for the four one-night titrations, was 2.00 for the algorithm, and was 2.12 using a fixed pressure of 10 cmH2O.

Conclusions: Considerable night-to-night variation means that a one-night titration is not very precise and is subject to random variation. A one-night titration has a similar inaccuracy to that resulting from using an algorithm, based on OSA severity and neck circumference. Setting all patients with OSA at 10 cmH2O is little worse.

MeSH terms

  • Algorithms*
  • Bias
  • Continuous Positive Airway Pressure / methods*
  • Humans
  • Regression Analysis
  • Sensitivity and Specificity
  • Sleep Apnea, Obstructive / therapy*