If ERCP was unsuccessful in a precholecystectomypatient, EUS results were compared with intraoperativecholangiography. If ERCP was not successful in a post-cholecystectomy patient, ERCP was repeated or the pa-tient was followed to January 1996. A ﬁnal diagnosis of choledocholithiasis was made if stones were removed from the bile duct by perioperativeendoscopic sphincterotomy or intraoperative stone extrac-tion. The absence of choledocholithiasis was conﬁrmed if no stones were seen in patients undergoing ERCP and/orintraoperative cholangiography, another diagnosis wasmade, or there was no relapse of symptoms and cholesta-sis during a minimum follow-up of 12 months. All minorandmajorpost-procedurecomplications(i.e.,allundesiredoutcomes) of both diagnostic and therapeutic ERCP werenoted, including those that did not require hospitalization.
Becausepatientsenrolledinthestudywerenotactuallycharged for the EUS, the cost analysis is based on costestimates and not actual costs. However, using the actualoutcomes of each diagnostic test, the relative direct costsof evaluating and treating patients with suspected chole-docholithiasis using EUS- and ERCP-based strategieswere identiﬁed and compared after stratifying by riskgroup. The former strategy used EUS as a diagnostic testfollowed by endoscopic sphincterotomy and stone extrac-tion at ERCP only if stones were found at EUS. TheERCP-based strategy used ERCP as both a diagnostic andtherapeutic modality.The cost analysis was based on the existing approach tothe management of bile duct stones at our institution. Inprecholecystectomy patients, intraoperative cholangiogra-phy was not routinely performed at laparoscopic cholecys-tectomy. Expertise in therapeutic ERCP was available atour institution; hence, bile duct stones identiﬁed preoper-atively were preferentially extracted at ERCP. If ERCPwas unsuccessful and multiple/large/proximal stones wereidentiﬁed by EUS and/or cholangiography, open commonbile duct exploration was performed. Bile duct stonesdiagnosed by intraoperative cholangiography were prefer-entially removed by ERCP after laparoscopic cholecystec-tomy. In post-cholecystectomy patients, ERCP was reat-tempted at least once if the initial procedure wasunsuccessful.The direct costs associated with each type of diagnosticapproach were calculated from the perspective of thethird-party payer. Hospital charges and professional feesdo not necessarily reﬂect the cost to the payer in Ohio.Hence, 1994 Medicare reimbursement ﬁgures for perti-nent CPT codes for outpatient procedures were used(instead of charges) for the cost analysis. These wereobtained from the University Hospitals of Cleveland andthe faculty practice billing ofﬁces (University Physicians,Incorporated). The total hospital facility and professionalreimbursement for EUS (CPT 43259 and 76975), diagnos-tic ERCP (CPT 43260), endoscopic sphincterotomy (CPT43262), and stone extraction (CPT 43264) were used in thecost analysis. The radiology costs related to ERCP (use of ﬂuoroscopy and radiology technicians) were not includedbecause all ERCPs were performed in the Digestive Dis-ease Center’s ﬂuoroscopy room without a technician. Thecost of diagnosis, treatment, and hospitalization forprocedure-related complications were also included, if applicable.In the cost analysis, the cost of an additional ERCP wasadded to the cost of an EUS if the latter was unsuccessful(n
1), falsely negative or positive (n
4), or haduncertain/indeterminate results (i.e., endosonographerunsure about the presence or absence of a bile duct stone,n
6). The cost of another diagnostic ERCP was includedin the cost of an ERCP-based strategy if the initial ERCPwas unsuccessful (n
4) or falsely positive or negative(n
2). The cost of endoscopic sphincterotomy was addedto the cost of an ERCP-based strategy when it wasactually performed because of diagnostic uncertainty (n
5) or a false-positive diagnosis (n
1). The total andaverage costs of an EUS- and ERCP-based strategy of managing choledocholithiasis were computed. Averagecost was deﬁned as the total cost of performing a diagnos-tic strategy divided by the total number of patients stud-ied. The marginal cost or the difference between each typeof diagnostic strategy was calculated for each risk groupand compared.
McNemar’s test was used to compare paired categoricaldata.
Spearman’s rank correlation was used to deter-mine the degree of agreement in EUS and ERCP stonesize.
Kappa scores were calculated to determine theconcordance between EUS and ERCP results.
Continu-ous paired data (such as the estimated total and marginalcosts) were analyzed using nonparametric methods, in-cluding the Wilcoxon rank-sum test.
value of 0.05was considered signiﬁcant. All statistical analyses wereperformed using SAS 6.10 (SAS PC for Windows; SASInstitute, Inc., Cary, N.C.).
We studied 64 patients; 42 were women and 56%were awaiting cholecystectomy. The mean age was53 years (range 22 to 92 years). All patients under-went both EUS and ERCP. Seventy seven percent of pre-cholecystectomy patients also had intraopera-tive cholangiography. The overall prevalence of cho-ledocholithiasis by stone extraction at ERCP orcholecystectomy was 31%. The distribution of pa-tients into risk groups is given in Table 1. Asexpected, the actual prevalence of bile duct stoneswas signiﬁcantly different among the risk groups(
0.001, Fisher’s exact test). Seventy percent of high-risk patients had bile duct stones comparedwith only 28% of moderate risk, none of the low risk,and4%ofindeterminateriskpatients.USand/orCTdiagnosed only 36% of high-risk patients with cho-ledocholithiasis and none of the moderate and inde-terminate risk patients with stones.
EUS vs ERCP for diagnosis of choledocholithiasis M Canto, A Chak, T Stellato, et al.
VOLUME 47, NO. 6, 1998 GASTROINTESTINAL ENDOSCOPY