Background: Diagnosis of complications is important for patient care, diagnostic quality, and safety. Paracentesis is a common procedure among hospitalized patients. The use of ultrasound decreases complications and improves safety. We aimed to characterize the complication rate observed among paracenteses performed by a hospital medicine bedside procedures service (MPS) before and after a transition to a live-guided approach and develop a standard trigger tool for identifying complications for manual review.
Methods: The University of Minnesota established a MPS in September 2016. Initially, static ultrasound technique was used. By May of 2019, live guidance was standardized. All paracenteses performed by the MPS between 09/01/2016-12/31/2018 were compared to those performed between 05/01/2019-03/21/2022. Complications were identified using ICD codes. Linear mixed models and generalized linear mixed models (logistic for binary outcome) were used with a time-varying covariate for treatment group, and a random effect for patients undergoing multiple procedures. Complications were manually chart-reviewed and validated using an a-priori protocol to determine the validity of the ICD code.
Results: A total of 2964 paracenteses were performed on 1095 persons. Baseline characteristics were similar. Out of 73 complications identified, chart review determined that 14/73 (positive predictive value 19.2%) were attributable to the MPS. All complications were related to bleeding. There was no significant difference in the complication rate with 8 (0.6%) occurring with static guidance and 6 (0.4%) occurring with live-guidance (OR 0.85 (0.02, 34.29), P-value 0.932). Our procedural efficiency decreased with a rate ratio of 0.77 (p< 0.001).
Conclusions: The MPS observed a low complication rate for paracenteses performed. Periodic monitoring using ICD codes as a trigger for case review is feasible to ensure appropriate diagnosis of procedural complications. We did not observe a difference in the complication rate following the transition to live-guidance. Real-time guidance lowered procedural efficiency but the effect was small.