Peripheral oxygen saturation (SpO2) measured by pulse oximetry is an unreliable surrogate marker for arterial oxygenation (SaO2) in critically ill patients. We hypothesized that a higher perfusion index (PFI) would be associated with better accuracy of SpO2 measurement. We retrospectively collected SaO2, SpO2, and PFI data for each arterial blood gas (ABG) analysis in a cohort of intensive care unit patients. PFI was categorised as low (PFI < 1.0), intermediate (1.0 ≤ PFI ≤ 2.5), or high (PFI > 2.5). The correlation between SpO2 and SaO2 was studied using Pearson's correlation. The Bland-Altman plot was used to analyse the agreement between SpO2 and SaO2. Furthermore, the correlation between the (SpO2-SaO2) difference and PFI was assessed. The level of (dis)agreement was calculated for the three PFI categories separately. Overall, 281 patients and 1281 data points were analysed. There was a significant correlation between SaO2 and SpO2 (r = 0.69, p < 0.01). The Bland-Altman analysis revealed a mean difference between SaO2 and SpO2 of 0.2% with limits of agreement of ± 6% (SD ± 2%). The correlation between the PFI and the (SpO2-SaO2) difference was low; the (SpO2-SaO2) difference improved only marginally with higher PFI values. The accuracy of pulse oximetry for estimating arterial oxygenation was moderate and improved little with increasing PFI values. Thus, the additive value of PFI in clinical decision making is limited. Therefore, we advise performing an ABG before adjusting fraction of inspired oxygen (FiO2) settings.
Keywords: Critical illness; Intensive care unit; Oxygen saturation; Perfusion index; Pulse oximetry.