MANAGEMENT OF ACUTE UPPER GASTROINTESTINAL BLEEDING IN THE INTENSIVE CARE UNIT: AN INVERSE PROBABILITY-WEIGHTED ANALYSIS OF GENERAL ANAESTHESIA DURING ENDOSCOPY

Background: Acute upper gastrointestinal bleeding frequently requires intensive care, haemodynamic resuscitation andurgent endoscopic assessment. General anaesthesia may facilitate endoscopy in patients at risk of aspiration or proceduralintolerance, but its effect on endoscopic performance and clinical outcomes remains uncertain. Materials and Methods:Patients undergoing upper gastrointestinal endoscopy for acute upper gastrointestinal bleeding were classified according towhether the procedure was performed under general anaesthesia or without general anaesthesia. Twenty-five patients whohad already been intubated because of critical illness were excluded. The final analysis included 200 patients: 40 in thegeneral anaesthesia group and 160 in the non-general anaesthesia group. Unadjusted comparisons were supplemented byinverse probability of treatment weighting to account for baseline imbalances. Results: Patients receiving generalanaesthesia had a higher prevalence of chronic liver disease, higher Simplified Acute Physiology Score II values and morefrequent haemorrhagic shock before endoscopy. General anaesthesia was not associated with a significant difference inerroneous diagnosis or complete haemostasis at the first endoscopy. The adjusted time to complete haemostasis was 7.0hours shorter with general anaesthesia (95% confidence interval −11.5 to −2.6; p=0.002). General anaesthesia was stronglyassociated with adequate procedural tolerance (odds ratio 43.18, 95% confidence interval 5.91–5501.94; p<0.001), a lowerlikelihood of premature interruption (odds ratio 0.11, 95% confidence interval 0.00–0.82; p=0.026) and successfulmanagement with a single endoscopy (odds ratio 4.64, 95% confidence interval 1.38–15.63; p=0.013). After adjustment,general anaesthesia was not associated with acute respiratory failure, invasive mechanical ventilation or in-hospitalmortality. Conclusion: General anaesthesia improved procedural tolerance, reduced premature interruption and shortenedthe time to complete endoscopic haemostasis without evidence of inferior adjusted respiratory or mortality outcomes. Itsprincipal benefit appeared to be improved procedural conditions rather than increased haemostatic efficacy at the firstendoscopy. Selective use based on airway risk, physiological instability and anticipated procedural tolerance is moreappropriate than routine administration.