Clinical and Economic Implications of Interventions in Pancreatic Fluid Collections

Objective To compare clinical and economic implications of percutaneous and endoscopic treatment approaches in patients with pancreatic fluid collections (PFCs). Materials and Methods This is a retrospective claims analysis of Medicare beneficiaries who underwent inpatient endoscopic or percutaneous PFC drainage procedures (2016–2020). We performed longitudinal analysis of claims for all-cause mortality and rehospitalization during 180-day follow-up. Main outcome was mortality. Other outcomes were rehospitalization and direct costs. Results A total of 1311 patients underwent endoscopic (n = 727) or percutaneous (n = 584) drainage. Percutaneous as compared with endoscopic approach was associated with higher mortality (23.08% vs 16.7%, P = 0.004), rehospitalization (58.9% vs 53.3%, P = 0.04), and mean direct hospital costs ($37,107 [SD = $67,833] vs $27,800 [SD = $43,854], P = 0.004). On multivariable analysis, percutaneous drainage (adjusted hazard ratio [HR], 1.38; 95% confidence interval [CI], 1.02–1.86; P = 0.039), older age (hazard ratio [HR], 1.04; 95% CI, 1.01–1.04; P < 0.001), intensive care unit stay (HR, 1.02; 95% CI, 1.01–1.03; P < 0.001), and multiple comorbidities (HR, 1.07; 95% CI, 1.05–1.09; P < 0.001) were significantly associated with mortality. Percutaneous drainage (adjusted odds ratio [OR], 1.30; 95% CI, 1.04–1.63; P = 0.027) and older age (OR, 0.98; 95% CI, 0.97–0.99; P < 0.001) were significantly associated with rehospitalizations. Conclusions As percutaneous drainage may be associated with higher mortality, rehospitalization, and costs, when requisite expertise is available, endoscopy should be preferred for treatment of PFC amenable to such an approach. Randomized trials are required to validate these findings.

][3] Treatment is indicated if the collection becomes infected or symptomatic.While surgery has historically been considered as the initial treatment option, given the minimal invasiveness, morbidity, and costs, there has been a growing interest in percutaneous and endoscopic approaches.These minimally invasive interventions have a high rate of technical and treatment success. 4,5When clinical outcome for either approach is suboptimal, surgery is performed as a rescue measure.This is particularly relevant in necrotizing pancreatitis where percutaneous drainage is an integral part of the surgical step-up approach. 6he recent American Gastroenterological Association clinical update recommends either approach, percutaneous or endoscopic, as initial intervention for management of necrotizing pancreatitis. 7While 4 randomized trials have compared endoscopic and minimally invasive surgical interventions in patients with necrotizing pancreatitis or pancreatic pseudocysts, 5,[8][9][10] there are no randomized controlled trials or large-scale real world data that have compared outcomes between endoscopic and percutaneous treatment approaches.This is important because pancreatitis is the third commonest reason for gastrointestinal disease-related hospitalizations in the United States and majority of PFC located in the lesser sac are accessible for drainage by either approach. 8,11n addition, in this era of cost containment, it is important to determine treatment that not only yields superior clinical outcomes but one that can also be delivered at lower costs.
To achieve this objective, we conducted a retrospective claims analysis of Medicare beneficiaries who underwent inpatient endoscopic or percutaneous PFC drainage procedures.Unlike single-center observational studies that are limited by sample size and institution-specific treatment bias, the Medicare administrative database yields real-world data of a national patient cohort that is representative of healthcare facilities across the United States.Our objective was to compare all-cause mortality, all-cause rehospitalization, and direct hospital costs between treatment groups.

Data Source
A retrospective claims analysis was conducted using the Medicare Standard Analytic Files (SAF) from July 1, 2016, to June 30, 2020, which provide data on individual claims submitted by healthcare institutions and providers for Medicare beneficiaries.These data include information on patient demographics, comorbidities, providers, healthcare facility characteristics such as bed strength and geography, procedures, Medicare severity diagnosis-related groups, length of stay (LOS), dates of service, charges, costs, and reimbursement amounts.

Patient Population
The study cohort consisted of all inpatients who were Medicare beneficiaries and underwent an endoscopic or percutaneous drainage procedure of the pancreas for acute or chronic pancreatitis in the hospital setting.The Current Procedural Terminology (CPT); International Classification of Diseases, Tenth Revision, Procedure Codes (ICD-10-PCS); and International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) diagnosis codes were used to identify the study cohort are shown in Table 1.Patients who underwent drainage using more than one treatment modality at index intervention were excluded.Also excluded were patients who had any pancreatic procedure performed within 180 days before the index intervention.To minimize bias the analysis was restricted only to those facilities that had claims for both endoscopic ultrasound and interventional radiology-guided percutaneous drainage procedures, patients treated at facilities having claims of only one procedure type-either only endoscopic or only percutaneous-were excluded.We censured counts and associated proportions when nationwide counts of specific claims were 10 or fewer in accordance with Medicare SAF100 data license agreement.

Outcome Measures
The primary outcome was all-cause mortality within 180 days after index intervention.Secondary outcomes were all-cause rehospitalization and reinterventions within 180 days after index procedure.Other outcomes included LOS and hospital costs associated with index intervention.

Statistical Analysis
All analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC).χ 2 test was used to conduct univariate analyses of count data and analysis of variance for univariate analyses of continuous data.For testing potential statistical differences in racial composition, due to censoring small numbers of many races, difference between cohorts was calculated as White versus non-White.Both unadjusted and adjusted regressions were performed.For performing adjusted multivariable analyses, generalized method of moments (GMM) was adopted and adjusted for potential confounders, including patient demographics, LOS, and comorbidity (Elixhauser score). 11We used linear GMM for continuous outcomes and logistic GMM for dichotomous outcomes.

Patient and Healthcare Facility Characteristics
A total of 1311 patient claims met criteria for inclusion in the study with 727 patients being treated endoscopically and 584 by percutaneous techniques.Figure 1 is a flow diagram of the derived study population.There was no significant difference in patient demographics or characteristics of healthcare facilities between groups as shown in Table 2.The median age of patients in both cohorts were 65 years or older, 60% male, >80% White, >75% procedures were performed in large hospitals (≥500 beds), and >60% procedures were performed in urban areas.Of 1311 patients, 212 were coded to have acute pancreatitis, 359 chronic pancreatitis, and 740 with both acute and chronic pancreatitis.There was no significant difference in top 10 comorbidities or overall Elixhauser comorbidity score between cohorts (Supplemental Table 1, http://links.lww.com/MPA/B105).

Clinical Outcomes
The primary outcome, all-cause mortality within 180 days of index intervention, was significantly higher at 23.1% for interventional radiology-guided percutaneous drainage as compared with 16.7% for the endoscopic cohort (P = 0.004) (Table 3).Likewise, all-cause rehospitalization at 180 days was significantly higher for the percutaneous approach at 58.9% versus 53.3% for the endoscopic cohort (P = 0.04).
Although no significant difference was observed in overall rates of reintervention at 180 days, patients treated by percutaneous drainage were significantly more likely to require reinterventions using endoscopic techniques at 33% versus only 16% of the endoscopic cohort who required adjuvant percutaneous drainage (P = 0.02) (Fig. 2).The mean LOS (19.7 vs 15.1 days, P = 0.0001) and proportion of patients requiring care in the intensive care unit (ICU) at index admission was significantly higher for patients treated by percutaneous drainage as compared with endoscopic approach (54.4% vs 40.5%, P < 0.0001) (Table 3).When assessed specifically for acute or chronic pancreatitis, while there was no difference in all-cause rehospitalization (P = 0.404) or reinterventions (P = 0.956), mortality at 180 days was higher for acute pancreatitis (P = 0.008).
Logistic regression analysis was performed to examine factors associated with mortality and rehospitalization at 180 days (Table 4).We found that for the primary endpoint of mortality at 180 days, there was significant association with percutaneous approach (P = 0.039), older age (P < 0.0001), higher Elixhauser comorbidity score (P < 0.001), ICU stay (P < 0.001), and chronic pancreatitis (P = 0.040).The adjusted hazard for death within 180 days for patients undergoing percutaneous drainage at index intervention was 38% higher compared with the endoscopic cohort.Likewise, for rehospitalization within 180 days, there was significant association with percutaneous drainage (P = 0.027) and older age (P < 0.001).The odds of rehospitalization within 180 days was 30% higher among patients treated by percutaneous as compared with endoscopic approach.

Economic Implications
The mean charges and total hospital costs were significantly higher for percutaneous than the endoscopic approach by a difference of $61,183 (P = 0.003) in charges and $16,810 in total costs (P = 0.002), respectively (Table 5).Direct treatment costs, derived by deduction of overhead costs, were significantly higher for percutaneous than the endoscopic approach by a difference of over  $10,000 (P = 0.004).However, total hospital payment for the index procedure was comparable between cohorts.Thus, with greater costs and comparable payments, performing percutaneous drainage incurred financial loss-$−5876 (P < 0.0001)and was not advantageous.

DISCUSSION
This study demonstrates that the interventional radiologyguided percutaneous treatment approach as compared with endoscopic drainage may be associated with increased mortality, higher rehospitalizations, longer length of hospital stay, and increased costs.
Most patients with symptomatic PFC due to acute or chronic pancreatitis are increasingly being referred for percutaneous or endoscopic drainage as first-line treatment option.Exceptions include acute illness where an intervention is delayed until the collection becomes encapsulated. 12In addition, in 15% of patients where the PFC may extend to the lower abdomen or pelvis and is not accessible to endoscopic drainage, only a percutaneous drain placement is feasible. 8In vast majority of other patients, both endoscopic and percutaneous approaches are viable treatment options.In clinical practice, given the lack of consensus, the choice of treatment should not be determined by the hospital services such as internal medicine or surgery under whom a patient is admitted but rather by evidence-based multidisciplinary consensus.
Our study demonstrates significant difference in important clinical outcomes between approaches.The 38% lower mortality and 30% less rehospitalization for the endoscopic cohort is likely related to technical advantages conferred by the procedure.Unlike percutaneous drainage catheters that are usually 8 mm or smaller in diameter, transluminal pathways created at endoscopy are larger (10-20 mm) and facilitate better drainage of inflammatory or infectious cyst contents.In persistently symptomatic patients, they can also act as conduit through which endoscopic necrosectomy can be performed at the same or subsequent session.These advantages confer rapid symptom relief and thereby shortens LOS. 5,8n the contrary, in addition to prolonged hospital course, percutaneous drain placement is associated with increased incidence of cutaneous pancreatic fistulae, precipitation of infection within a previously sterile collection due to suboptimal drainage, drain malfunction due to dislodgement or clogging by debris, and external skin infection. 7,10,13,14Consequently, patients undergo frequent rehospitalization and may require rescue therapy using alternate treatment modalities such as internal endoscopic drainage, percutaneous sinus tract necrosectomy, or video-assisted retroperitoneal debridement. 15,16From an economic standpoint, treating patients adopting an endoscopic approach translate to decreased  healthcare costs.As Medicare inpatient payment is bundled to diagnosis related groups, our data show that endoscopic treatment has a more favorable financial impact for healthcare systems than using a percutaneous approach.
What lessons can we learn from these observations that can advance our knowledge on treatment of PFC? One, given the superior clinical outcomes, when PFCs are amenable to drainage by transluminal techniques and when the requisite expertise and resources are available, endoscopy should be the preferred first-line treatment.Percutaneous approach should be used as treatment measure when the PFC is not amenable to endoscopic drainage, when a step-up surgical treatment approach is being contemplated, or when the requisite endoscopic expertise and resources are unavailable.[10] However, a meta-analysis of the 3 randomized controlled trials have shown a significant difference in new-onset multiple-organ failure, enterocutaneous fistula/perforation, pancreatic fistula, and overall LOS in favor of the endoscopic approach. 17Given that this analysis was derived from a claims database without details on baseline clinical or radiological characteristics of subjects, well-designed, multicenter, prospective studies involving a large cohort of patients are required to confirm the observed endpoint and to assess whether this negative outcome may apply to all or only a specific cohort of patients with PFC.Two, as endoscopic drainage is less costly and has a more favorable economic impact, large healthcare facilities should consider investing in requisite technology and manpower to offer a comprehensive and multidisciplinary continuum of care and smaller facilities should consider referring patients to larger, regional centers of excellence.Three, given the close interface between disciplines of interventional radiology, endoscopy, and gastrointestinal surgery, patients with complex PFCs may best be managed adopting a multidisciplinary collaborative approach.
The present study has several limitations.First, as the database used in analysis was administrative, we were unable to ascertain clinical attributes in individual patients, perform propensity score matching or subgroup analysis such as outcomes in pseudocysts versus necrotic collections, or performance of specific downstream interventions such as endoscopic or retroperitoneal debridement in the percutaneous cohort that may have contributed to suboptimal clinical outcomes.Second, although patient demographics and overall comorbidity scores were comparable between groups, treatment approaches were assigned and not randomized.It is also possible that patients may have received treatment based on underlying disease characteristics or severity, which may have precluded alternate treatment approaches.Third, because of the observational design of the study, residual unmeasured confounding may persist such as the possibility of the hospital inaccurately coding the procedures.Fourth, implications of the study pertain mainly to Medicare patients who are 65 years or older, with disability or chronic medical diseases.Consequently, the severity of disease in this study population is likely higher than those encountered in routine clinical practice (Supplemental Table 1, http://links.lww.com/MPA/B105).The strengths of the study are, first, by inclusion of only those centers that performed both procedures the possibility of treatment assignment bias was minimized.Second, as the data were derived from healthcare facilities across the entire United States encompassing both urban and rural areas, the study findings should serve as nidus for well-designed randomized trials comparing outcomes between endoscopic and percutaneous approaches in patients with symptomatic pancreatic fluid collections.

FIGURE 2 .
FIGURE 2. Procedural reinterventions in study subjects in endoscopic and percutaneous drainage cohorts.

TABLE 1 .
Current Procedural Terminology, ICD-10-PCS, and ICD-10-CM Diagnosis Codes ICD-10-CM K85.X.Acute pancreatitis ICD-10-CM K86.X.Alcohol-induced chronic pancreatitis, other chronic pancreatitis, cyst and pseudocyst, other diseases of the pancreas excluding steatorrhea, islet cell tumor, and fibrocystic disease FIGURE 1. Flow diagram of the derived study population based on inclusion or exclusion criteria.Pancreas • Volume 53, Number 5, May 2024 Pancreatic Fluid Collection Treatment Outcomes

TABLE 2 .
Comparison of Patient Demographics and Hospital Characteristics *Race P value represents White versus non-White.Full cohort data not presented to prevent identification of censored data fields.

TABLE 3 .
Comparison of Clinical Outcomes Between Study Cohort

TABLE 4 .
Multivariable Analysis Examining Factors Associated With Mortality and Rehospitalization Within 180 Days *Endoscopic approach is reference group.† Female reference group.‡ Incremental odds per day ICU length of stay.§ Incremental odds per point increase Elixhauser score.|| Cases coded as both acute and chronic pancreatitis as reference group.

TABLE 5 .
Economic Implications