Development of a clinical score to estimate pancreatitis-related hospital admissions in patients with a new diagnosis of chronic pancreatitis : the trinity score

Background: The clinical course of chronic pancreatitis is unpredictable and there is no globally accepted score to predict the disease course. We developed a clinical score to estimate pancreatitis-related hospitalisation in patients with newly diagnosed chronic pancreatitis. Methods: We conducted a retrospective cohort study using two clinical chronic pancreatitis databases held in tertiary referral centres in Dublin, Ireland, and in Tarragona, Spain. Individuals diagnosed with chronic pancreatitis between 2007 and 2014 were eligible for inclusion. Candidate predictors included aetiology, body mass index, exocrine dysfunction, smoking and alcohol history. We used multivariable logistic regression to develop the model. Results: We analysed data from 154 patients with newly diagnosed chronic pancreatitis. Of these, 105 patients (68%) had at least one hospital admission for pancreatitis-related reasons in the 6 years following diagnosis. Aetiology of chronic pancreatitis, body mass index, use of pain medications and gender were found to be predictive of more pancreatic-related hospital admissions. These predictors were used to develop a clinical score which showed acceptable discrimination (area under the ROC curve = 0.70). Discussion: We developed a clinical score based on easily accessible clinical parameters to predict pancreatitis-related hospitalisation in patients with newly diagnosed chronic pancreatitis.