Keywords
Sleep apnea - obstructive - hypoxia - pulse oximetry
Introduction: Oxyhemoglobin desaturation events (ODE) are major contributors to the detection and
classification of obstructive sleep apnea (OSA). However, variations in pulse oximeter
properties, such as resolution, may affect ODE detection and, consequently, the oxygen
desaturation index (ODI). We hypothesized that traditional low-resolution pulse oximetry
(1%), which rounds decimal points in the readout, overestimates ODE detected by high-resolution
pulse oximetry (0.1%).
Objective: To compare the ODI determined by high-resolution pulse oximetry data with the ODI
obtained from the same data, mathematically rounded to simulated traditional pulse
oximetry.
Methods: Consecutive deidentified diagnostic sleep studies of patients with suspected OSA,
recorded between February and June 2023, were extracted from the Biologix database
and analyzed. High-resolution pulse oximetry data from the Biologix oximeter (0.1%)
was compared with mathematically generated traditional low-resolution pulse oximetry
data (1%). ODI differences between high- and traditional low-resolution oximetry were
compared using both 3% and 4% oxyhemoglobin desaturation criteria. A confusion matrix
and Bland-Altman plot were used to assess the agreement on ODI classifications between
high- and traditional low- resolution oximetry.
Results: A total of 9,551 diagnostic sleep studies were analyzed. Compared to the high-resolution
oximeter, the traditional low-resolution oximeter overestimated the mean ODI for both
3% (14.3 (6.9–25.3) vs. 21.1 (12.0–34.0)) and 4% (9.1 (3.7–18.0) vs. 13.0 (6.0–23.3))
oxyhemoglobin desaturation criteria (p<0.0001 for both comparisons). Traditional low-resolution
oximeter misclassified detection or OSA severity in 40% and 31% of the patients using
3% and 4% oxygen desaturation criteria, respectively (p<0.0001). This overestimation
of ODI was not homogeneous across OSA severities and was amplified among patients
with no or with mild forms of OSA. Median accuracy was observed between high- and
traditional low-resolution oximetry, considering both 3% and 4% oxygen desaturation
criteria (60% and 69%, respectively).
Conclusion: Traditional low-resolution pulse oximetry overestimates the number of ODE detected
by the high-resolution pulse oximetry due to the embedded rounding algorithm. Since
assignment of respiratory events is influenced by the magnitude of oxyhemoglobin desaturation,
traditional low-resolution pulse oximetry may overestimate detection and severity
of OSA.
