O59 Predicting use of opiates in patients with inoperable pancreatic cancer: a retrospective cohort study. (19th June 2022)

• Record Type: Journal Article
• Authors: Koulouris, Andreas; Clark, Allan; Baio, Gabriela; Alexandre, Leo
• Is Part Of: Gut. Volume 71(2022)Supplement 1
• Page Start: A33
• Page End: A33
• Publication Date: 2022-06-19
• Subjects: Gastroenterology -- Periodicals; 616.33
• DOI: 10.1136/gutjnl-2022-BSG.59
• Languages: English
• ISSNs: 0017-5749

Abstract

Abstract : Introduction: Opioids are frequently prescribed for patients with inoperable pancreatic cancer for abdominal pain. Their doses usually have to be up-titrated leading to adverse effects. A clinical model predictive of the need for opioids may enable the early administration of Endoscopic Ultrasound-guided Coeliac Plexus Neurolysis (EUS-CPN) to treat pain and prevent opioid dose escalation. We sought to (1) estimate the prevalence of opioid use; (2) investigate the associations of demographic, clinical, radiological and cytological parameters with opioid use; and (3) derive a model predictive of opioid use. Methods: This was a single-centre retrospective cohort study of patients with unresectable pancreatic adenocarcinoma with a minimum survival of three months. Prevalence of opiate use was estimated at three monthly intervals following diagnosis. Logistic regression estimated the associations between clinical, biochemical, histo-cytological and radiological variables for the outcome of opiate use by three months. Model performance was assessed with Receiver Operator Characteristic (ROC) analysis, a calibration plot and the Hosmer-Lameshow test. The model was internally validated with bootstrap resampling. Results: Overall, 383 patients were included in the analysis, covering the period from January 2010 to January 2020. Of those 146 (38%) had available radiological data. Prevalence of pain ranged between 37 to 47% at three monthly intervals in the first year of the illness. Age (odds ratio [OR] per year 0.97, 95% CI, 0.94 to 0.99, p<0.001), pain at presentation (OR 9.57, 95% CI, 5.78 to 15.85, p<0.001), performance status of 3 (OR 2.57, 95% CI, 1.32 to 5.00, p = 0.006), tumours located at the body of the pancreas (OR 0.20, 95% CI, 0.05 to 0.75, p=0.018) and the distance from the right coeliac ganglion (OR per mm 0.96, 95% CI, 0.93 to 0.98, p=0.001) were independent risk factors for opioid use at three months. The C-statistic for the prediction model including clinical and radiological characteristics was 0.84. Sensitivity 78.9%, specificity 78.5%, positive and negative predictive values were 80.0% and 77.3%, respectively. Conclusions: In total, 47% of patients with inoperable pancreatic cancer use opiates by three months. Age, pain at presentation, health performance status, tumours of the body and distance of the tumour from the right coeliac ganglion are associated with opioid use at three months. The predictive model demonstrates good performance. External validation of this predictive model is required.