Cost-Effectiveness of Modalities of Palliative Biliary Drainage in Malignant Extrahepatic Biliary Obstruction

Abstract

INTRODUCTION:

Biliary drainage in malignant extrahepatic distal bile duct obstruction can be achieved with endoscopic retrograde cholangiopancreatography (ERCP), percutaneous transhepatic biliary drainage (PTBD), or endoscopic ultrasound-assisted choledochoduodenostomy (EUS-CDS). The aim of our study was to evaluate all these modalities in a cost-effective model.

METHODS:

A patient level discrete time cycle Markov model was developed using United Kingdom's healthcare resource group tariff costs. Health-related quality of life data were used to develop quality-adjusted life years. Monte-Carlo deterministic analysis was performed. Probabilistic sensitivity analysis and deterministic analysis were performed.

RESULTS:

EUS-CDS was the most cost-effective modality for biliary drainage in malignant extrahepatic distal bile duct obstruction (cost £2,116 [$2,853.99]/effectiveness 1.35) compared with ERCP (£2,349 [$3,168.25]/1.33) and PTBD (£3,161 [$4,263.45]/1.31). Microsimulation over 1,000 patient iterations confirmed that EUS-CDS had the highest net monetary benefit.

DISCUSSION:

EUS-CDS seems to be a cost-effective strategy for biliary drainage in malignant obstruction under a range of modeling assumptions, particularly in settings with adequate availability and expertise.

INTRODUCTION

Biliary drainage in malignant extrahepatic distal biliary obstruction (MBO) is essential to facilitate palliative therapy and improve quality of life (1,2). This has been traditionally achieved by endoscopic retrograde cholangiopancreatography (ERCP) and biliary stenting or percutaneous transhepatic biliary drainage (PTBD). Endoscopic ultrasound-assisted choledochoduodenostomy (EUS-CDS) has been demonstrated to be an effective biliary drainage modality, particularly when ERCP-assisted drainage is unsuccessful (3–5). Since the first reported case in 2001, EUS-CDS has been achieved using lumen apposing metal stents (LAMS) (6). There have been significant developments in the design and application with current LAMS platforms incorporating a cautery system that enables direct duct-puncture and deployment (7,8). More recently, EUS-CDS has been proposed as a primary drainage modality in MBO (9), with the added advantage of enabling tissue acquisition and biliary drainage in a single session (10). Moreover, fluoroscopy is generally not required in EUS-CDS. Current guidelines continue to recommend ERCP as the first-line modality for biliary decompression, with EUS-guided approaches typically reserved as second-line options ahead of PTBD (11).

Although debate continues regarding the optimal modality, there remains a paucity of data on the cost-effectiveness of these strategies. We therefore sought to evaluate the cost-effectiveness of EUS-CDS, ERCP, and PTBD.

METHODS

Cost-effectiveness model

A patient level discrete time cycle Markov model was developed to model biliary drainage modalities in MBO (12) (Figure 1). A literature search was performed to identify variables to populate model inputs (Table 1). Each Markov cycle was 3 months in duration and crossover salvage biliary drainage strategies were included. After biliary decompression, patients remained in health states (postdrainage, complications, or death) until the end of the time horizon or death. Procedure-related complications were modeled as occurring within the first cycle, whereas subsequent cycles captured longer term outcomes including survival and reintervention, with associated complications incorporated. Clinical outcomes were included, including survival data for palliative extrahepatic pancreaticobiliary malignancy. UK healthcare resource group tariff costs were used (13). Health-related quality of life data were used to develop quality-adjusted life years (QALYs) (14). Monte-Carlo microsimulation was performed over 1,000 iterations. A probabilistic sensitivity analysis (PSA) was performed using distributions (Table 1) to model uncertainty. Sensitivity analyses were additionally performed.

Figure 1.

Study model. ERCP, endoscopic retrograde cholangiopancreatography; EUS-CDS, endoscopic ultrasound-assisted choledochoduodenostomy; MBO, malignant extrahepatic distal biliary obstruction; PTBD, percutaneous transhepatic biliary drainage.

Table 1.

Model inputs (study variables and distributions)

Variable	Point estimate	Minimum value	Maximum value	Reference
Costs (£)
Cost of ERCP	3,062	2,500	5,000	(13)
Cost of EUS	2,748	2,000	3,000	(13)
Cost of PTBD	1,041	500	3,000	(13)
Cost of biliary SEMS	800	700	1,000	Retail costs UK
Cost of LAMS	2,000	1,500	2,500	Retail costs UK
Cost of therapeutic OGD	522	500	600	(13)
Cost inpatient bed day	330	300	400	(13)
Cost of critical care bed day	2,200	2,000	3,000	(13)
Discount	0.035	0.025	0.04
Clinical efficacy of ERCP	0.965	0.9	0.99	(15)
Clinical efficacy of EUS-CDS	0.948	0.9	0.99	(15–17)
PTBD efficacy	0.945	0.9	0.99	(15,17)
Frequency of post ERCP complications	0.106	0.05	0.249	(18)
Frequency of severe EUS-CDS associated complications	0.006	0.0001	0.01	(15)
Frequency of post PTBD complications	0.208	0.059	0.215	(15)
Frequency of post ERCP pancreatitis	0.035	0.01	0.1	(19)
Frequency of post ERCP perforation	0.001	0.0005	0.003	(19)
Frequency of significant post ERCP bleeding	0.003	0.001	0.011	(19)
Frequency of post PTBD cholangitis	0.263	0.15	0.35	(20)
Mortality because of post ERCP bleeding	0.0004	0.0001	0.001	(19,21)
Mortality because of post ERCP pancreatitis	0.007	0.0005	0.009	(22)
Frequency of severe post ERCP pancreatitis	0.05	0.01	0.1	(23)
Mortality because of post ERCP perforation	0.0006	0.0001	0.001	(24)
Mortality associated with therapeutic EUS	0.001	0.0001	0.01	(25)
Mortality associated with PTBD	0.052	0.01	0.1	(20)
Utility value — pancreatic cancer treatment	0.749	0.5	0.8	(18)
Utility — palliation	0.507	0.4	0.7	(18)
Utility — pancreatitis state	0.7	0.6	0.8	(26)

Distribution	Variable modeled	Type of distribution	Alpha	Beta
Distribution 1	ERCP efficacy	Beta	101.63	3.69
Distribution 2	PTBD efficacy	Beta	46.51	2.71
Distribution 3	EUS efficacy	Beta	46.66	2.56
Distribution 4	ERCP complications	Beta	68,382.89	576,738.75
Distribution 5	PTBD complications	Beta	134,185.30	510,936.34
Distribution 6	EUS-CDS complications	Beta	9.68	1,603.13

ERCP, endoscopic retrograde cholangiopancreatography; EUS-CDS, endoscopic ultrasound-assisted choledochoduodenostomy; LAMS, lumen apposing metal stents; OGD, esophagogastroduodenoscopy; PTBD, percutaneous transhepatic biliary drainage; SEMS, self-expanding metal stents.

Biliary drainage

Patients with MBO underwent ERCP, PTBD, or EUS-CDS for drainage. It was assumed that all patients were clinically suitable to undergo ERCP, EUS-CDS, and PTBD. To compare EUS-CDS with ERCP, it was assumed that EUS-CDS would be performed as a primary EUS-CDS procedure and not in the setting of duodenal obstruction where ERCP would not be feasible. ERCP and EUS-CDS were modeled as outpatient/ambulatory procedures, and PTBD was modeled as an inpatient procedure. The duration of inpatient stay with PTBD can vary, with a median length of stay quoted as 13 days in a large retrospective series (27). All modalities were evaluated as standalone primary drainage modalities with technical failure leading to crossover into alternate strategies to facilitate biliary decompression. Costs specific to self-expanding metal stents, for PTBD and ERCP use and LAMS (for EUS-CDS) were used in the model. Reintervention costs were included along with costs related to complications, and a discounting date of 3.5% was applied to the model.

Definitions

Base-case analysis.

The most likely or best-estimate values were used for all model inputs and provided the primary estimate of cost-effectiveness against which all sensitivity analyses were compared.

Deterministic sensitivity analysis.

Deterministic sensitivity analysis assessed the impact of uncertainty in individual model parameters by varying inputs over a plausible range while holding others constant, to evaluate their influence on the results.

Probabilistic sensitivity analysis.

PSA simultaneously accounts for uncertainty across all model parameters by assigning probability distributions to inputs and repeatedly sampling from these distributions, generating a range of possible outcomes and estimates of decision uncertainty.

RESULTS

Base-case analysis

On base-case analysis, EUS-CDS was the most cost-effective modality for biliary drainage in MBO (cost £2,116 [$2,853.99]/effectiveness 1.35) compared with ERCP (£2,349 [$3,168.25]/1.33) and PTBD (£3,161 [$4,263.45]/1.31) with the EUS-CDS strategy absolutely dominating the other strategies (Table 1). The net monetary benefit for EUS-CDS was £38,442 ($51,849.22) compared with ERCP (£37,426 [$50,478.88]) and PTBD (£36,114 [$48,709.30]).

Deterministic analysis

Deterministic analysis using point estimates was performed over 1,000 patient journeys, and the EUS-CDS strategy remained undominated (£2,133 [$2,876.92]/1.37) compared with ERCP (£2,386 [$3,218.15]/1.33) and PTBD (£3,166 [$4,270.19]/1.31) (Figure 2). The net monetary benefit for EUS-CDS was the highest at £38,833 ($52,376.59) in comparison with £37,851 ($51,052.10) for ERCP and £36,102 ($48,693.11) for PTBD.

Figure 2.

Cost-effectiveness deterministic analysis of ERCP, EUS-CDS, and PTBD. ERCP, endoscopic retrograde cholangiopancreatography; EUS-CDS, endoscopic ultrasound-assisted choledochoduodenostomy; QALYs, quality-adjusted life years; PTBD, percutaneous transhepatic biliary drainage.

Procedure-related complications

Although severe EUS-CDS complications were used as a point estimate in the deterministic analysis, on sensitivity analyses, all EUS complications (5%–20%) were tested in the model with no change in the results (EUS-CDS was more cost-effective compared with the other drainage modalities).

Probabilistic sensitivity analysis

On PSA, EUS-CDS remained the most cost-effective strategy (mean cost £2,115 [$2,852.64]; effectiveness 1.35), compared with ERCP (mean £2,351 [$3,170.95]; 1.33) and PTBD (mean £3,165 [$4,268.84]; 1.30). Confidence intervals (CIs) for costs and QALYs overlapped but did not alter the overall conclusion (Figure 3). Acceptability curves (Figure 4) demonstrated that the most cost-effectiveness iterations were those of EUS-CDS for willingness-to-pay (WTP) £30,000 ($40,462.95). Tornado diagrams examining incremental cost-effectiveness ratio (ICER) values for each variable used in the model as part of sensitivity analyses demonstrated that the expected value from a healthcare payer perspective was well below a WTP of £30,000 ($40,462.95) for EUS-CDS when compared with ERCP (Figure 5) and PTBD (Figure 6). Threshold analysis did not identify a threshold cost or utility variable affecting on the model. However, the incremental QALY differences between strategies were small, and therefore results are sensitive to assumptions regarding utilities and complication rates.

Figure 3.

Cost-effectiveness scatter plot. ERCP, endoscopic retrograde cholangiopancreatography; EUS, endoscopic ultrasound; QALYs, quality adjusted life years; PTBD, percutaneous transhepatic biliary drainage.

Figure 4.

Acceptability curves (probabilistic sensitivity analysis). ERCP, endoscopic retrograde cholangiopancreatography; EUS, endoscopic ultrasound; PTBD, percutaneous transhepatic biliary drainage.

Figure 5.

Tornado diagram (ICER)—ERCP vs EUS. ERCP, endoscopic retrograde cholangiopancreatography; EUS, endoscopic ultrasound; ICER, incremental cost-effectiveness ratio; PTBD, percutaneous transhepatic biliary drainage; WTP, willingness-to-pay.

Figure 6.

Tornado diagram (ICER)—PTBD vs EUS. ERCP, endoscopic retrograde cholangiopancreatography; EUS, endoscopic ultrasound; ICER, incremental cost-effectiveness ratio; PTBD, percutaneous transhepatic biliary drainage; WTP, willingness-to-pay.

Center access to EUS-CDS

EUS-CDS is an advanced endoscopy technique requiring significant training and technical abilities in therapeutic EUS. Our model assumes equal availability of EUS-CDS, ERCP, and PTBD. We modeled a scenario analysis reflecting limited EUS access. Because there are no population-level parameters for tertiary EUS availability, we modeled 3 EUS access parameters based on geography and access to tertiary EUS services (0%–20% access, representing limited availability in tertiary centers, 50% access [model for regional networks] and 80% or greater access [as a future projected model]). Sensitivity analyses were subsequently performed based on these access parameters. Modeling identified that EUS-CDS was the most cost-effective strategy when availability exceeded 25% (Supplementary Table 1, http://links.lww.com/CTG/B571). ERCP was the most cost-effective biliary drainage strategy for EUS access up to 25%. This represents one of the first attempts to explicitly model service availability constraints in EUS-guided biliary drainage.

PTBD as a purely salvage strategy

To compare EUS-CDS head to head with ERCP, we examined a cost-effectiveness model where PTBD was modeled as a purely salvage strategy for ERCP or EUS-CDS failure. EUS-CDS remained cost-effective compared with ERCP for biliary drainage, with an ICER of 6,058 (Supplementary Table 2, http://links.lww.com/CTG/B571), with EUS-CDS availability modeled from 20% and greater access.

DISCUSSION

There have been several studies comparing the cost of various modalities for biliary obstruction. Using the United States Nationwide Inpatient Sample database, Inamdar et al (28) compared ERCP with PTBD for malignant biliary obstruction. In patients with pancreatic cancer, the ERCP group had a statistically significant lower cost of hospitalization compared with PTBD (ERCP: $53,881.10 vs PTBD: $73,151.40; P < 0.01). In a subgroup analysis of patients with cholangiocarcinoma, PTBD was associated with higher costs in both low and high-volume centers (low-volume: ERCP $50,139.70 vs PTBD $88,428.80; high-volume: ERCP $60,098.40 vs PTBD $74,748.50; P = 0.07).

In a comparative study of ERCP vs EUS-guided CDS, Téllez-Ávila et al (29) reported that EUS-guided CDS was associated with lower costs compared with ERCP with metal stents or plastic stents (PS) (Median [SD]: EUS-CDS $1,950 vs ERCP with PS $3,640 vs ERCP with metal stents $4,750; P = 0.0001). The EUS-CDS group also required fewer reinterventions compared with patients in the ERCP groups with metal or PS. In a Korean study, Yoon WJ et al compared EUS-CDS with PTBD as a second-line modality in patients with ERCP failure (30). The median cost of biliary treatment was lower in the EUS-CDS group compared with PTBD ($1,203.36 vs $1,517.83; P = 0.0015). The median total cost of the entire treatment was also lower in the EUS-CDS group, attributed to the reduced need for reintervention ($4,175.53 vs $5,391.87; P = 0.0496). The authors further analyzed cost minimization for the US Medicare population and found that EUS-CDS remained cost-effective (EUS-CDS: $9,497.03 vs PTBD: $13,878.44; average cost-savings: $4,381.41).

In a US-based study, a semi-Markov microsimulation model was used to compare the effectiveness of PTBD vs EUS-biliary drainage in patients with pancreatic adenocarcinoma after failed index ERCP (31). EUS-CDS was found to be cost-effective compared with PTBD, with an ICER of $55,421/QALY. EUS-CDS remained cost-effective at a WTP threshold of $100,000/QALY if the cost was less than $13,693. The PTBD group had a higher reintervention rate compared with EUS-CDS, consistent with findings from previously reported studies. Another study compared EUS-CDS with LAMS vs PTBD and found that EUS-CDS was cost-effective, with an ICER of $15,008/QALY at a WTP threshold of $100,000/QALY (32).

Although current guidelines still support ERCP as the primary modality of biliary decompression in patients with extrahepatic biliary obstruction, the trend is gradually shifting toward EUS as a primary option (11,33). A meta-analysis of 6 RCTs including 577 patients showed that EUS-CDS was comparable to ERCP regarding technical success (risk ratio [RR] 1.04; 95% CI, 0.96–1.13) and clinical success (RR 1.02; 95% CI, 0.96–1.08) (9). However, hospital stay was significantly shorter with EUS-CDS (mean difference −1.03 days; 95% CI, −1.53 to −0.53), with a lower risk of reintervention (RR 0.57; 95% CI, 0.37–0.88). An additional advantage of EUS-CDS is the lower risk of postprocedure pancreatitis, which was demonstrated in this study as well (RR 0.15; 95% CI, 0.03–0.66).

To the best of our knowledge, this is the first study to compare all 3 techniques to understand the comparative cost-effectiveness of biliary drainage in MBO. We found that EUS-CDS was the most cost-effective strategy at a WTP of £30,000 per QALY compared with ERCP and PTBD and EUS-CDS remained cost-effective when examined in a PSA model. Our study has several limitations. We assumed that each modality would be available in a center. EUS-guided biliary drainage is not available in many centers and is a niche technique with inherent risks. We assumed that patients would be suitable for crossover to alternate drainage modalities in case of failure of a particular technique (i.e., PTBD for failed ERCP or EUS-CDS). For the purposes of the study and to compare drainage modalities in a cost-model, we also assumed that patients could undergo PTBD as a primary drainage modality for MBO. Although this is slightly artificial, as MBO would generally be approached by ERCP, it offers a direct head-to-head comparative perspective. In the absence of disease-specific utility values for cholangiocarcinoma, we used pancreatic cancer utilities, which are commonly adopted in health economic models of hepatopancreatobiliary malignancy because of similar symptom burden and prognosis. This introduces structural uncertainty, although sensitivity analyses across a plausible range (0.5–0.8) did not alter conclusions. Finally, the simplified crossover structure may favor EUS-CDS, as real-world pathways typically use ERCP as first-line therapy, reserving EUS-CDS for failures. This may overestimate its effectiveness as a primary strategy.

The strengths of our study are that we adopted a clinical and pragmatic model of managing MBO and modeled procedural complications including death. We stress-tested each model as part of broader sensitivity analyses and modeled in follow-up based on survival data for inoperable MBO and found that EUS-CDS seems to be a cost-effective strategy for biliary drainage in malignant obstruction. In addition, from a mechanistic perspective, EUS-CDS may be more cost-effective because it provides durable internal drainage with high technical success, minimizing the need for repeat interventions and avoiding external drainage catheters, which are often associated with longer hospital stays, complications, and increased resource utilization. Biliary drainage techniques in MBO continue to evolve, and our study suggests that EUS-CDS should be part of a set of tools available for the management of MBO.

CONFLICTS OF INTEREST

Guarantor of the article: Abdullah Abbasi, MD.

Specific author contributions: Both authors were involved in idea of the article, analysis with writing and editing of the final manuscript.

Financial support: None to report.

Potential competing interests: None to report.

Study Highlights.

WHAT IS KNOWN

• ✓ Malignant distal extrahepatic biliary obstruction can be managed using endoscopic retrograde cholangiopancreatography (ERCP), percutaneous transhepatic biliary drainage (PTBD), or endoscopic ultrasound (EUS)-guided biliary drainage.

• ✓ Currently, ERCP is the most commonly used first-line modality, with PTBD or EUS-guided therapy as alternative options.

• ✓ Comparative cost-effectiveness data among these drainage strategies are limited, particularly from a payer perspective.

WHAT IS NEW HERE

• ✓ This study provides a comparative cost-effectiveness analysis of ERCP, PTBD, and EUS-guided choledochoduodenostomy (CDS) using a UK NHS payer perspective using a patient-level Markov microsimulation model.

• ✓ EUS-CDS was dominant, offering lower costs and higher quality adjusted life years compared with ERCP and PTBD.

• ✓ The cost-effectiveness of EUS-CDS remained robust across deterministic and probabilistic sensitivity analyses at a £30,000/quality adjusted life year willingness-to-pay threshold.

Supplementary Material

ct9-17-e01062-s001.pdf ^{(777.6KB, pdf)}

ct9-17-e01062-s002.pdf ^{(441.1KB, pdf)}

ABBREVIATIONS:

CDS

choledochoduodenostomy

CI

confidence interval

ERCP

endoscopic retrograde cholangiopancreatography

EUS

endoscopic ultrasound

EUS-CDS

endoscopic ultrasound-guided choledochoduodenostomy

ICER

incremental cost-effectiveness ratio

LAMS

lumen-apposing metal stent

MBO

malignant biliary obstruction

OGD

esophagogastroduodenoscopy

PSA

probabilistic sensitivity analysis

PTBD

percutaneous transhepatic biliary drainage

QALY

quality-adjusted life year

SEMS

self-expanding metal stent

WTP

willingness-to-pay