National analysis of in-hospital resource utilization in choledocholithiasis management using propensity scores

Two treatment options exist for choledocholithiasis (CDL): endoscopic retrograde cholangiopancreatography (ERCP) and common bile duct exploration (CBDE). Resource utilization measured by total in-hospital charges (THC) and length of stay (LOS) was compared using the propensity score (PS). In this study, PS was the probability that a patient received CBDE based on comorbidities and demographics. The power of this method lies in balancing groups on variables by PS, resulting in 90% bias reduction and improved inferential validity compared to traditional analytic techniques.

Laparoscopic cholecystectomy (LC) patients with CDL who had ERCP or CBDE were identified in the 2002 U.S. Nationwide Inpatient Sample. Patients were ordered into five PS balanced strata. Mean THC, LOS, and estimated costs were compared. A linear regression model was used to estimate the contribution that LOS had on estimated costs. Monetary values were adjusted to 2004 dollars.

A total of 40,982 patients underwent LC with CDL in 2002; 27,739 had either ERCP (93%) or CBDE (7%). Mean age was 52.7 ± 0.4 years, with 74% women. Mean THC were less for CBDE ($25,200 ± $1,800) than for ERCP ($29,900 ± $800, p < 0.05). Mean LOS was less for CBDE (4.9 ± 0.2 days) than for ERCP (5.6 ± 0.1 days, p < 0.05). PS adjusted analysis revealed an estimated overall cost savings of $4,500 ± $1,600 and reduced LOS (0.6 ± 0.2 days) per hospitalization for CBDE. Mean THC, LOS, and estimated costs across PS score balanced strata were generally higher in the ERCP group compared to the CBDE group. LOS contributed 53% to increased THC and 62% of estimated costs. A higher cumulative incidence of complications was evident with CBDE (0.5–4.6%) compared to ERCP (0.3–3.6%).

Based on this PS analysis, CBDE incurs less THC, reduces LOS, and has less estimated costs for CDL compared to ERCP. Furthermore, CBDE appears to be dramatically underutilized.