Objectives To investigate discordance in oxy-hemoglobin saturation measured both by pulse oximetry (SpO2) and arterial blood gas (ABG, SaO2) among critically ill coronavirus disease 2019 (COVID-19(+)) patients compared to COVID-19(-) patients. Methods Paired SpO2 and SaO2 readings were collected retrospectively from consecutive adult admissions to four critical care units in the United States between March and May 2020. The primary outcome was the rate of discordance (|SaO2-SpO2|>4%) in COVID-19(+) versus COVID-19(-) patients. The odds each cohort could have been incorrectly categorized as having a PaO2/FiO2 above or below 150 by their SpO2: Fractional inhaled oxygen ratio (pulse oximetry-derived oxyhemoglobin saturation:fraction of inspired oxygen ratio [SF]) was examined. A multivariate regression analysis assessed confounding by clinical differences between cohorts including pH, body temperature, renal replacement therapy at time of blood draw, and self-identified race. Results There were 263 patients (173 COVID-19(+)) included. The rate of saturation discordance between SaO2 and SpO2 in COVID-19(+) patients was higher than in COVID-19(-) patients (27.9% vs 16.7%, odds ratio [OR] 1.94, 95% confidence interval [CI]: 1.11 to 2.27). The average difference between SaO2 and SpO2 for COVID-19(+) patients was −1.24% (limits of agreement, −13.6 to 11.1) versus −0.11 [–10.3 to 10.1] for COVID-19(–) patients. COVID-19(+) patients had higher odds (OR: 2.61, 95% CI: 1.14-5.98) of having an SF that misclassified that patient as having a PaO2:FiO2 ratio above or below 150. There was not an association between discordance and the confounders of pH, body temperature, or renal replacement therapy at time of blood draw. After controlling for self-identified race, the association between COVID-19 status and discordance was lost. Conclusions Pulse oximetry was discordant with ABG more often in critically ill COVID-19(+) than COVID-19(–) patients. However, these findings appear to be driven by racial differences between cohorts.

BackgroundInterpretation of lung ultrasound artifacts by clinicians can be inconsistent. Artificial intelligence (AI) may perform this task more consistently.Research QuestionCan AI characterize lung ultrasound artifacts similarly to humans, and can AI interpretation be corroborated by clinical data?Study Design and MethodsLung sonograms (n=665) from a convenience sample of 172 subjects were prospectively obtained using a pre-specified protocol and matched to clinical and radiographic data. Three investigators scored sonograms for A-lines and B-lines. AI was trained using 142 subjects and then tested on a separate dataset of 30 patients. Three radiologists scored similar anatomic regions of contemporary radiographs for interstitial and alveolar infiltrates to corroborate sonographic findings. The ratio of oxyhemoglobin saturation:fraction of inspired oxygen (S/F) was also used for comparison. The primary outcome was the intraclass correlation coefficient (ICC) between the median investigator scoring of artifacts and AI interpretation.ResultsIn the test set, the correlation between the median investigator score and the AI score was moderate to good for A lines (ICC 0.73, 95% CI [0.53-0.89]), and moderate for B lines (ICC 0.66, 95% CI [0.55-0.75]). The degree of variability between the AI score and the median investigator score for each video was similar to the variability between each investigator’s score and the median score. The correlation among radiologists was moderate (ICC 0.59, 95% CI [0.52-0.82]) for interstitial infiltrates and poor for alveolar infiltrates (ICC 0.33, 95% CI [0.07-0.58]). There was a statistically significant correlation between AI scored B-lines and the degree of interstitial opacities for five of six lung zones. Neither AI nor human-scored artifacts were consistently associated with S/F.InterpretationUsing a limited dataset, we showed that AI can interpret lung ultrasound A-lines and B-lines in a fashion that could be clinically useful.

No abstract