Conscious sedation (CS) failure in diagnostic endoscopic ultrasound (dEUS) remains poorly characterised. Understanding the demographic and clinical characteristics of patients who experience sedation failure with CS during dEUS is useful for risk stratification and the development of pre-procedural screening tools.  

This study aims to characterise patients experiencing CS failure during dEUS procedures at a tertiary referral centre. 

A retrospective analysis was conducted on all patients undergoing dEUS, between October 2024 and September 2025. CS failure was defined as an incomplete procedure attributed to inadequate patient tolerance of sedation (agitation, paradoxical reaction, or intolerance to standard sedation regimens). Demographic data, comorbidities, sedation dosages, and clinical outcomes were extracted and described.

Thirteen patients (1.7% of 765 procedures) experienced CS failure. The cohort was predominantly female (61.5%, n=8) with median age 60 ± 13.9 years (range 33–78). 23.1% were current smokers (n=3) and 15.4% ex-smokers (n=2). 23.1% reported active alcohol use (n=3). Complex comorbidities were prevalent, including chronic obstructive pulmonary disease (n=2), chronic kidney disease/end-stage renal disease (n=2), type 2 diabetes mellitus (n=2), and psychiatric conditions including depression and prior intravenous drug use (n=2). Despite high sedation doses - mean fentanyl 104.2 ± 30.3 mcg and mean midazolam 5.8 ± 1.7 mg - these patients failed to achieve adequate sedation. Procedural indications were predominantly pancreatic (n=8) and biliary pathology (n=2). Nine of these patients went on to have an EUS under general anaesthesia (GA). Of the remaining patients, one was referred for CT guided biopsy while the others were managed with non-invasive surveillance after multidisciplinary team discussion. There were no post-procedure adverse events at 30-day follow-up. 

In general, CS appears to be effective and well-tolerated for dEUS.  Patients experiencing CS failure demonstrate a distinct clinical phenotype characterised by female predominance, and alcohol and smoking use history. Pre-procedural identification of these "hard-to-sedate" characteristics may enable risk-stratified sedation planning, with consideration for alternative approaches including deep sedation or general anaesthesia to improve procedural success and patient safety.