Rationale: Despite oxygen's classification as an essential medication by the World Health Organization, it is inconsistently available in many resource-constrained settings. Hypoxemia is associated with increased mortality, and mounting evidence suggests that hyperoxia may also be associated with adverse outcomes.Objectives: To determine if overuse of oxygen for some patients in a Rwandan tertiary care hospital emergency department might coexist with oxygen shortages and underuse of oxygen for other patients, and whether an educational intervention coupled with provision of pulse oximeters could improve the distribution of limited oxygen resources.Methods: We screened all patients in the adult emergency department (ED) of the University Teaching Hospital of Kigali for hypoxemia and receipt of oxygen therapy for 5 weeks. After completing baseline data collection, we provided pulse oximeters and conducted a didactic training with pre- and posttests on oxygen titration, with a chosen target oxygen saturation (SpO2) of 90% to 95%. Four and 12 weeks after the intervention, we evaluated all patients in the ED again for SpO2 and receipt of oxygen therapy for 4 weeks each period. We also recorded ED oxygen use and availability of reserve oxygen for the hospital during the three study periods.Results: During all data collection periods, 214 of 1,765 (12.1%) unique patients screened were hypoxemic. The proportion of patient-days with appropriately titrated oxygen therapy (SpO2, 90-95%) increased from 18.7% at baseline to 38.5% and 42.0% at 4 and 12 weeks postintervention (P < 0.001). On a multiple-choice examination testing knowledge of appropriate oxygen titration, clinicians' scores improved from average 60% (interquartile range [IQR], 40-80%) correct to 80% (IQR, 60-80%) correct immediately after the educational intervention (P < 0.001). Oxygen use in the ED decreased from a median of 32.0 (IQR, 28.0-35.0) tanks per day to 25.5 (IQR, 24.0-29.0) and 16.0 (IQR, 12.5-21.0) tanks per day at Weeks 4 and 12, respectively (P < 0.001), and the median daily number of tanks in reserve for the hospital appeared to increase, although this did not reach statistical significance (30.0 [IQR, 9.0-46.0], 86.5 [IQR, 74.0-92.0], and 75.5 [IQR, 8.5-88.5], respectively; P = 0.07).Conclusions: Among patients in a Rwandan adult ED, 12.1% of patients were hypoxemic and 81.3% of patient-days were either under- or overtreated with oxygen during baseline data collection on the basis of our defined target of SpO2 90% to 95%. Follow-up results at 4 and 12 weeks postintervention demonstrated sustained improvement in oxygen titration and likely increased availability of oxygen resources.
Keywords: hyperoxia; hypoxemia; low-income country; quality improvement; resource poor settings.