Predictors of Time-Dependent Ascites and Hydrothorax Response after Transjugular Intrahepatic Portosystemic Shunt Creation: A Claims Analysis.

Abstract

Purpose: To identify patient factors influencing the time-dependent response of ascites and hydrothorax after TIPS creation using a large claims dataset.

Materials and methods: Adult patients from a nationwide claims database with at least 4 paracenteses or thoracenteses within 60 days before TIPS and at least 30 days follow-up were included (paracentesis cohort, n=3,742; diuretic subset, n=1,110; thoracentesis cohort, n=238). Continuous-time hidden Markov models were used to classify patients into underlying response states-non-response, partial response, and complete response (CR)-based on observed paracentesis, diuretic prescription, and thoracentesis claims. Transition intensities between response states were used to quantify how quickly patients responded to TIPS. Covariate effects influencing response time were estimated.

Results: In the paracentesis cohort, CR probabilities were 48%, 70%, and 88% at 90, 180, and 365 days post-TIPS, respectively. Male gender, metabolic dysfunction-associated steatohepatitis cirrhosis, and higher pre-TIPS paracentesis rates predicted slower freedom from paracenteses, whereas alcoholic cirrhosis predicted faster response. Among diuretic users, 38% achieved freedom from diuretics at 1 year; younger age and lower pre-TIPS dose predicted faster weaning. In the hydrothorax cohort, CR rates at 90, 180, and 365 days were 54%, 81%, and 94%. Higher pre-TIPS thoracentesis rates predicted slower response.

Conclusion: Modeling of nationwide claims data provided quantitative, time-dependent estimates of the changes in paracentesis, diuretic, and thoracentesis requirements after TIPS and identified predictors of slower versus faster response. These findings may help physicians to better counsel patients regarding outcomes after TIPS.