Presentation Authors: Diana K. Bowen*, CHICAGO, IL, Lihai Song, Gregory E. Tasian, Philadelphia, PA
Introduction: Ureteroscopy (URS) and shockwave lithotripsy (SWL) are the most commonly performed surgical interventions for kidney and ureteral stones. However, the comparative effectiveness of these interventions at the population level is unclear as prior studies have been performed in younger, commercially insured populations and have estimated retreatment for the marginal rather than the average patient. We sought to determine the risk of retreatment after SWL and URS among all patients undergoing these surgeries in South Carolina.
Methods: Using all payer billing data from 74 hospitals in South Carolina, we performed a series of pseudo-randomized trials to determine the risk of retreatment within 6 months of URS or SWL between 1997 and 2016. To account for confounding between surgical interventions, we first fit a propensity score model for each year&[prime]s data to predict the probability of undergoing SWL conditional on hospital-level and patient-level covariates. Next, a discrete time failure model was fit using inverse probability weighted logistic regression which accounted for repeated observations of the same patient across multiple trials and balanced measured confounders across treatment groups. Odds Ratios (OR), 95% confidence intervals, and probabilities of retreatment were estimated.
Results: From 1997-2016, 123,970 children and adults underwent SWL (n=74,235; 59.9%) or URS (n=49,735; 40.1%), of whom 12,208 patients (9.9%) underwent retreatment with SWL and/or URS within 6 months. 73% of retreatments occurred within 2 months of the original surgery. The probability of retreatment was 7.5% for URS and 10.4% for SWL (number of SWL needed to retreat, 33). Compared to initial URS, initial SWL was associated with a 44% increased odds of retreatment (95% CI 1.36, 1.51). SWL (vs. URS) was always associated with a greater odds of retreatment over the 6-month period, but the magnitude of the association varied significantly across the follow-up time. SWL had the greatest risk for retreatment compared to URS at months 2 (OR 2.09, 95% CI 1.86, 2.35) and 3 (OR 1.98, 95% CI 1.70, 2.31). Of those patients who had retreatment, patients who had initial SWL were more likely to have SWL for retreatment (84.6%) than were patients who had initial URS to have URS for retreatment (29.3%).
Conclusions: Compared to URS, SWL was associated with a substantially increased odds of retreatment among the population of South Carolina over a 20 year period. However, the probability of retreatment for the average patient was modest for both URS (7.5%) and SWL (10.4%).