Abstract Background In patients with chronic obstructive pulmonary disease (COPD), pulse oximetry (SpO2) is routinely used to assess oxygenation status. However, in cases of moderate hypoxemia, SpO2 may overestimate arterial oxygen saturation (SaO2), potentially delaying recognition of clinically significant desaturation. This raises the question of whether a similar discrepancy occurs in patients with severe hypoxemia, in whom accurate assessment is crucial for determining eligibility for long-term oxygen therapy (LTOT). We aimed to evaluate the diagnostic accuracy of pulse oximetry for identifying severe resting hypoxemia (PaO2 55 mmHg) in COPD patients undergoing LTOT, and to determine the optimal SpO2 cutoff for this purpose. Methods A retrospective analysis was conducted using medical records from a specialized LTOT outpatient clinic between January 2022 and December 2023. Severe hypoxemia was defined as PaO2 55 mmHg measured by arterial blood gas (ABG) analysis. All measurements were obtained simultaneously, in room air and at rest, to ensure standardized assessment conditions. Demographic and clinical data, including smoking status and Charlson Comorbidity Index (CCI), were recorded. Agreement between SpO2 and SaO2 was assessed using Bland-Altman analysis, and the diagnostic performance of SpO2 for detecting severe hypoxemia was evaluated by receiver operating characteristic (ROC) curve analysis. Results A total of 85 COPD patients were included, 64.7% female, with a median age of 70 years (IQR 64-79). Median PaO2 was 55.5 mmHg (IQR 50.2-65.4) and median CCI was 4 (IQR 3-6). Fourteen percent were current smokers and 86% were former smokers. Bland-Altman analysis demonstrated a mean bias of 0.05 percentage points (95% CI − 1.01 to 1.10) between SaO2 and SpO2, with no significant systematic difference (t = 0.089; p = 0.90). The area under the ROC curve was 0.979 (95% CI 0.949-1.00), indicating excellent discriminative ability. An SpO2 threshold of 89.5% yielded 100.0% sensitivity (95% CI 91.1-100.0) and 91.1% specificity (95% CI 78.8-97.5), with a positive predictive value of 90.1%, a negative predictive value of 100.0%, and a Youden’s index of 0.911. Conclusions Among COPD patients under LTOT follow-up, pulse oximetry demonstrated excellent diagnostic performance in detecting severe resting hypoxemia. An SpO2 cutoff of 89% (corresponding to the optimal ROC-derived threshold of 89.5%) provided excellent sensitivity and specificity, supporting its use as an initial screening tool in this population. This abstract is funded by: grant #2024/07508-4, São Paulo Research Foundation (FAPESP)
Prudente et al. (Fri,) studied this question.