Abstract
INTRODUCTION: Malignant gastric outlet obstruction (GOO) has a significant impact on quality of life. Endoscopic ultrasound-guided gastroenterostomy (EUS-GE) has shown promising results. Traditional isolated outcome measures do not sufficiently address critical considerations for end-of-life patients like oral intake tolerance. This study aimed to determine via a probabilistic approach, the optimal management strategy for GOO patients that maximizes their oral intake tolerance. METHODS: A Markov decision model was developed, with input variables based on a systematic review and meta-analysis of randomized controlled trials (RCT) comparing duodenal stenting (DS), EUS-GE and surgical gastrojejunostomy (GJ). A prospective cohort study with a comparator group was also included for EUS-GE model given the scarcity of RCTs. Model assumption was a patient with malignant GOO, with equal probabilities of being allocated to 1 of 3 treatment options. Each data point was evaluated using pooled probabilities from the meta-analysis of clinical outcomes. Primary outcome was successful oral intake tolerance at various time points of 1-6 months post-intervention. RESULTS: Fifteen studies were included into the Markov model. Based on 10 000 simulations in each arm, at a survival of 1-month, DS and EUS-GE had the highest likelihood of oral intake (81.2% and 80.4%) compared to GJ (75.5%). However, at a survival of 6-month, EUS-GE and GJ were better at palliating GOO, with likelihood of oral intake at 23.8% and 25.2%, compared to 21.3% for DS. CONCLUSION: For patients with a prognosis of more than 1-month, a surgical GJ, or EUS-GE if technical expertise is available, is preferred for GOO palliation.