Hypoxia predicts COVID-19 mortality: Evidence to inform low-resource care

Abstract

Background: Hypoxia is a critical yet under-recognised driver of poor outcomes in coronavirus disease 2019 (COVID-19). Early detection with cheap pulse oximetry is feasible in resource-limited settings.
Aim: This study estimated the prevalence of hypoxia at admission and its role in predicting mortality in three facilities in diverse resource settings in Kenya.
Setting: The study was conducted in three Kenyan hospitals.
Methods: We retrospectively analysed 1124 COVID-19 patient hospitalisation records (October 2020 – December 2021). Hypoxia was defined as the saturation of peripheral oxygen (SpO₂) ≤ 94% at admission. Differences in categorical variables were assessed using the χ² test. We used a multivariable Cox proportional hazards model to identify mortality predictors and Kaplan–Meier methods to estimate survival probabilities, with or without oxygen supplementation.
Results: Hypoxia was present in 81.4% of patients; 39.9% had no dyspnoea. Hypoxic patients compared to non-hypoxic patients were older (≥ 60 years: 44.6% vs. 24.4%) and had higher rates of dyspnoea (60.1% vs. 36.9%), hypertension (40.4% vs. 25.8%), and tachycardia (38.2% vs. 24.6%) (all p < 0.001). Only 68.6% of hypoxic patients received oxygen. Mortality was higher among hypoxic (38.0%) vs. non-hypoxic patients (13.6%, p < 0.001). Hypoxia independently predicted mortality (adjusted hazard ratio [aHR]: 1.9; 95% confidence interval [CI]: 1.2–2.8), particularly in older adults (aHR: 1.8) and those with dyspnoea (aHR: 1.5). Survival probabilities were worse for hypoxic patients regardless of dyspnoea or oxygen supplementation (p < 0.001).
Conclusion: Hypoxia was prevalent and significantly increased the mortality risk among hospitalised COVID-19 patients.
Contribution: Routine SpO₂ monitoring and targeted hypoxia management are critical in low-resource settings, particularly for vulnerable patients.