High-Flow Nasal Oxygen and Desaturation During Deep Sedation for Colonoscopy in Patients at Increased Respiratory Risk: A Propensity Score-Matched Cohort Study

Background/Objectives: Hypoxaemia is the most common adverse event during procedural sedation. High-flow nasal oxygen (HFNO) may reduce desaturation through continuous positive airway pressure, anatomical dead space washout, and stable FiO2 delivery. This retrospective cohort study assessed the association between two oxygen delivery strategies—HFNO versus Venturi mask oxygen—and the incidence of desaturation (SpO2 <90% at any time) during deep sedation for colonoscopy in patients at increased respiratory risk. The two strategies differed simultaneously in flow rate, delivered FiO2 (0.50 vs. 0.40), device interface, and humidification; any observed difference cannot be attributed exclusively to the high-flow mechanism. Methods: A single-centre retrospective cohort study with propensity score matching (PSM) was performed, including 200 consecutive adult patients with established obstructive sleep apnoea (OSA) or high-risk STOP-BANG criteria undergoing deep sedation for colonoscopy, grouped by oxygen delivery strategy: HFNO via nasal cannula (50 L/min, FiO2 0.50) versus Venturi mask (12 L/min, FiO2 0.40). The estimand of interest was the average treatment effect on the treated (ATT). Results: After 1:1 nearest-neighbour PSM (calliper 0.2 SD of the logit propensity score), 71 matched pairs were obtained; 10 of 17 covariates achieved adequate balance (SMD < 0.10). In the matched cohort, desaturation occurred in 12.7% of HFNO patients versus 22.5% of controls (matched-pair OR 0.53, 95% CI 0.23–1.23; p = 0.211; McNemar test), an absolute risk difference of 9.8 percentage points (95% CI −1.5 to +21.1). The estimate was imprecise and the confidence interval included no association. A sensitivity analysis using conditional logistic regression yielded an adjusted OR of 0.52 (95% CI 0.22–1.21; p = 0.129). The sample provided only 30.8% statistical power; an estimated 474 total patients (237 per group, parallel group design) would be required for 80% power in a future confirmatory randomised trial. Conclusions: In this propensity-matched cohort, the observed desaturation rate was lower with HFNO than with Venturi mask oxygen, but the estimate was imprecise, the confidence interval included no association, and residual confounding precludes definitive conclusions. These findings are hypothesis-generating and should be evaluated in a prospectively designed, adequately powered randomised trial.