Predictors of ERCP Procedure Time: A Multivariate Analysis of 10,000 ERCPs

Abstract

Background: ERCP procedure time (PT) is influenced by many procedural and patient factors. The magnitude of their effects has not been studied, but is needed to help create PT benchmarks in different clinical settings. Aim: To measure the independent effects of procedural and patient factors on ERCP PT. Methods: 10,120 patients entered in the 10-year MUSC GI Trac database (1994-2004) with PT between 5 and 180 minutes were used for the analysis. PT was recorded prospectively, and was defined as the time elapsed between scope insertion and removal. Univariate predictors were sought and clinically relevant potential confounders were identified and studied. A multivariate linear regression model was constructed to assess for predictors of PT. Results: Mean PT was 42.4 min (SD 24.7 min). Several covariates and potential confounders were considered: age, sex, inpatient/outpatient status, ASA class, clinical context (suspected sphincter dysfunction, tumors, benign biliary disease, pancreatitis and other), general anaesthesia (GA), sphincterotomy (ES) performed, previous ES, ERCP complexity grade (1-3), and year that the ERCP was performed (surrogate for ERCP “era”). Each increase in the complexity grade corresponded to a 3.9 [3.3-4.4] min (9%) average increase in PT (p<0.001). ES increased PT by 12.5 [12-14] min (30%); prior ES decreased PT by 7.5 [6.4-8.5] min (18%), (both p<0.001). Insertion of a pancreatic duct (PD) stent, e.g. for pancreatitis prevention, added 3.2 [1.6-4.8] min (8%) to the PT (p<0.001). More recent ERCP “era” was an independent predictor of decreasing PT (p<0.001). In the univariate analysis, GA appeared to decrease PT, but when corrected for confounding with “era” seen using nested models, GA's effect was no longer significant. Advancing age, male sex, tumor indication, and ASA class were significant independent predictors of longer PT (all p<0.001); inpatient status and phase of fellow training (e.g. time of year July-December) were not significant. Conclusions: ERCP era is an important confounder of PT, even when trends in other procedural and patient details appear to be accounted for. Quantifying, with relatively narrow confidence intervals, the average independent effect on PT of different clinical contexts and of therapeutics such as sphincterotomy and PD stents is feasible for ERCP, and may be important for cost-effectiveness analysis, reimbursement, and case-mix adjusted resource planning.