Introduction

Laparoscopic pancreatoduodenectomy (LPD) is emerging as a safe and effective option compared to open pancreatoduodenectomy (OPD) but still remains an area of significant controversy.1–4 Several retrospective studies have shown LPD to be non-inferior in regard to morbidity and mortality when compared to OPD.5–12 Collectively, these studies have identified potential benefits to LPD including decreased blood loss, delayed gastric emptying rates, hospital length of stay, wound infection rates, and utilization of intensive care services while facilitating larger percentages of patients proceeding on to adjuvant therapy.6–11 These benefits even appear to be realized during the institutions’ early experience with LPD.6,13,14 Further, there does not appear to be a clinically meaningful difference between the approaches with respect to surrogate markers of oncologic adequacy including positive margin rates6,8–10 and number of resected lymph nodes.6,8–10,13 Recently, we and others have reported similar long-term oncologic outcomes between the approaches including disease-free and overall survival.10,15 Taken together, these data suggest that LPD is equivalent to OPD with respect to most clinical outcome measures and may have some small benefits over OPD.

While LPD might have small benefits compared to OPD, what are the financial implications of a LPD? Only a limited number of small studies have published comparative cost data between LPD and OPD.6,7,13,16 These studies evaluated hospital costs associated with the surgical procedure and index hospitalization, showing increased operative costs are recouped by reduction in other costs (i.e., length of stay), but they did not evaluate the potential value of LPD in the context of the total episode of care. Thus, we hypothesized that episode of care costs favor LPD over OPD for patients with periampullary malignancy. To gain a better representation of the total costs associated with these operations, we analyzed healthcare needs related to the surgical procedure, evaluating costs associated with readmissions and utilization rates for skilled nursing facilities (SNF), long-term acute care (LTAC), physical rehabilitation, and home healthcare (HHC).

Methods

This study was approved by the University of Florida Institutional Review Board (IRB). A prospectively maintained clinical database identified all patients who underwent a PD from November 2010 to February 2014. This allowed for a minimum post-operative follow-up time of 12 months, enabling evaluation of itemized hospital cost data as well as post-discharge healthcare needs for periampullary cancer patient cohorts following LPD or OPD. The LPD was performed by a single surgeon who was already experienced with LPD. The OPD was performed by a group of experienced pancreatic surgeons.

The inclusion and exclusion criteria and clinical outcomes for these cohorts were previously described and reported in Delitto et al.8 Briefly, the cohorts represented consecutive patients with the diagnosis of a periampullary malignancy from November 2010 to February 2014 (excluding neuroendocrine neoplasms) that met the National Comprehensive Cancer Network (NCCN) criteria for resectable disease.8,17 Contraindications for enrollment included borderline resectable or locally advanced disease, morbid obesity (BMI > 40), rare aberrant organ anatomy, and prior surgeries with documentation stating an extensive amount of adhesive disease (Supplementary Fig. 1). The LPD and OPD cohorts were previously reported to be clinically similar with two exceptions; the median primary tumor size in greatest diameter was significantly larger in the OPD group (median (IQR) 2.5 (1.9–3.0) vs 2.7 cm (2.0–3.5), P = 0.046) and the R1 resection rate in the OPD group was significantly higher (26% versus 9.6%, p = 0.030)8 (Supplementary Tables 1 and 2). The laparoscopic techniques employed have also been previously described.14,18

All patients in the study received care in a single intensive care unit, a single regular care medical/surgical ward, and by a single discharge coordinator, social worker, and physician assistant, and a rotating team of residents. Post-operative care was highly proscribed by an Early Recovery After Surgery (ERAS) protocol that aimed for discharge on post-operative day 5. Items that are regulated by this and other ICU-specific protocols include pulmonary toilet, fluid resuscitation, blood transfusion triggers, nasogastric tube management, initiation and progression of oral intake, initiation of tube feeds or total parental nutrition, physical activity and use of physical therapy consultation, glucose monitoring and management, and diabetic education.

The cost data was abstracted from University of Florida Health’s data repository that includes direct and indirect costs. Total costs were calculated from the sum of the direct and indirect costs for specified categories including hospital salaries, supplies, and overhead. Physician professional billing was not included in the costs as these are not part of the institution’s data repository. All costs were abstracted and applied to each patient in identical fashion. Readmissions, lengths of stay, and discharge dispositions were also abstracted from University of Florida Health’s data repository. Readmissions were evaluated for attribution to surgical care versus other causes (i.e., neutropenia or electrolyte abnormalities associated with chemotherapy) at 30, 60, 90, 180, and 360 days.

All statistical analyses were performed using the SPSS version 22.0 statistical software package (IBM SPSS statistics for Windows; IBM Corp). Continuous variables were analyzed using the independent samples t test. Differences between categorical variables were analyzed using χ2 coefficients and Fisher’s exact test, as appropriate. Significance was considered for P < 0.05.

Results

During the study dates from November 2010 to February 2014, our institution performed 236 PDs of which 91 (39%) were LPD. Of these patients, 138 met the inclusion criteria of a periampullary malignancy; 7 patients were excluded as not being candidates for LPD for reasons not related to the neoplasm.8 Patients were further excluded if preoperative imaging or intraoperative findings identified disease burden beyond NCCN criteria for “resectable” disease. Of the 61 OPD patients and 70 LPD patients with periampullary adenocarcinoma, 11 and 18 patients, respectively, were excluded due to the presence of borderline resectable tumors. Thus, the OPD and LPD cohorts represented 50 and 52 patients, respectively. Of the 52 LPD patients, 2 required conversion to OPD but were included in the LPD arm as an intention to treat analysis.

The comparison of operating room (OR) costs between the cohorts found the LPD to have significantly higher total costs (median (IQR) of US$12,290 (US$11,387–13,475) for LPD vs US$11,299 (US$9053–12,968) for OPD; P = 0.05). When itemized, LPD had greater costs of laparoscopic equipment and regional nerve blocks (Table 1). The differences in total OR cost can be attributed to these two categories. No significant differences in costs associated with surgical staplers, electrosurgical devices, sutures, or room charges were observed between the two cohorts.

Table 1 Itemized OR costs
Full size table

While the hospital length of stay was significantly shorter for LPD (median (IQR) of 7 (6–11.5) vs 8 days (7–12); P = 0.025)8 (Supplementary Table 3), this was not associated with statistically significant savings in any non-OR category of care-related costs including room or ICU costs. However, trends in savings in the LPD cohort in these and several other categories led to the observation that total costs of the initial hospital stay were not significantly different between LPD and OPD (Table 2). Thus, non-specific savings in postoperative care costs negated the increased costs of the OR in the LPD cohort.

Table 2 Itemized hospital costs
Full size table

Another source of variability in the overall cost of PD that has not been previously evaluated is postdischarge costs directly attributable to recovery from the surgical procedure. These include expenses associated with home healthcare (HHC), skilled nursing facilities (SNFs), long-term acute care hospitals (LTACs), and rehabilitation facilities. We found that LPD patients were more likely to be discharged home compared to the OPD patients (P = 0.047) (Table 3). When this was subcategorized between those who went home with HHC and those who went home without HHC, there was also a trend for more LPD patients going home without HHC (P = 0.064). Further, more patients were discharged to a healthcare facility in the OPD group compared to the LPD group, respectively, 26 vs 10% (P = 0.038). No patients in the LPD cohort were discharged to rehabilitation facilities (P = 0.054) or to an LTAC (P = 0.24).

Table 3 Disposition at initial discharge
Full size table

Readmissions were evaluated for attribution to surgical care versus other etiologies (i.e., neutropenia or electrolyte abnormalities associated with chemotherapy) at 30, 60, 90, 180, and 360 days. No readmissions occurred due to surgery-related complications after 60 days, but readmissions related to chemotherapy toxicity were identified between 60 and 90 days (data not shown). Thus, all readmissions within 60 days of the surgical procedure were analyzed for potential cost differences. No differences in the total numbers of readmissions between LPD and OPD were found, 23 vs 30%, respectively (P = 0.43) (Table 4). No patient was readmitted more than one time in the 60 days from discharge. The total number of readmission hospital days in the OPD group was 186 vs 90 in the LPD group. However, the average length of stay per patient was not statistically significant between the groups. Ultimately, total costs associated with readmissions were not significantly different between the groups. Even when costs were itemized into categories (Table 5), no specific differences in costs were observed between the groups.

Table 4 60-day readmissions
Full size table
Table 5 Itemized hospital costs of readmissions
Full size table

Discussion

To determine if the LPD is an effective cost alternative to OPD, we retrospectively evaluated total episode of care costs for patients undergoing PD. This entailed compiling the costs of the initial hospitalization, readmissions related to the surgery, and disposition from the hospital. We show there is no difference between costs of the index admission or readmissions, except for a statistically significant, but only marginally relevant procedure-specific OR cost. However, we observed a significant difference in disposition from the hospital following the index admission. Although we could not obtain costs of the rehabilitation facilities, LTACs, and SNFs, or length of stay data, it is reasonable to assume the overall cost burden would be decreased in the LPD group as a result of these disposition differences.

Our data regarding costs of the index hospitalization are consistent with the reports of others. However, one report did identify that OPD surgical costs are significantly less than the LPD.6,7,16 Our data expand upon these earlier studies, identifying that LPD results in a significant number of patients going directly home rather than to an LTAC, rehabilitation facility, or SNF as compared to OPD. We also analyzed the total number of readmissions attributable to the surgical procedure and the associated hospital days and costs of those readmissions for both the LPD and OPD cohorts. The data show no significant difference in the number of readmissions and no differences in the length of stay per patient. The total costs of the readmissions are also not statistically different between the groups, nor are there any differences in the subcategories, suggesting that LPD does not confer a benefit via reduction in readmission rates or the associated costs of those readmissions.

As expected, when comparing the OR costs, we identified a statistically significant total cost difference favoring OPD. There was no difference between groups in costs associated with staplers, energy devices, or overall room charges. Rather, this difference was attributable to significant differences in the regional nerve block and laparoscopic equipment categories. The LPD group received bilateral paravertebral catheters while the OPD group received an epidural catheter at the time of surgery, and this doubled the LPD cost in that category. Thus, should the LPD group receive epidural catheter analgesia moving forward, the total cost of the LPD is decreased by US$630 per patient and this would eliminate the OR cost benefit of OPD (P = 0.40). Given that the total index admission cost of care was equitable between the two groups, for bundled payment models, these discrete categorical differences may not be relevant in the future.

Of note, all patients in this study were subjected to the same ERAS protocol, a practice that has become standard in most tertiary pancreatic surgery centers. The rationale for these protocols is to standardize care and thus drive reductions in complications and length of stay.19,20 This consistency between cohorts strengthens the validity of our observations but prohibits our ability to assess the impact of ERAS protocols upon costs or the potential decrease in cost that can be realized by implementing ERAS protocols.

This study is subject to a number of limitations. It reflects the results of a single LPD surgeon compared to multiple surgeons performing OPD within the experience of a single institution, and like all retrospective analyses, it is at risk for selection bias. Although the selection criteria was for resectable disease only, the cohorts did in fact have a difference in R1 resection with 26% of patients in the OPD having an R1 resection while only 9.6% in the LPD had an R1 resection. But, the median tumor size in the OPD group was also larger (2.5 (IQR 1.9–3.0) vs 2.7 cm (IQR 2.0–3.5), p = 0.046); thus, this certainly could be a confounding factor. Studies aimed at assessing the potential value of LPD with respect to obtaining negative retroperitoneal margin may be warranted. Further, this study does not account for the impact of a learning curve when implementing LPD that associates with increased OR time but not necessarily increased complications.6,13,21,22 Specific to increased OR time that would impact cost, this learning curve is realized somewhere between 10 and 50 cases.6 Thus, the time frame analyzed represents a period significantly beyond our institution’s LPD learning curve. As OR times are longer during this learning curve, associated cost increases should be expected. Finally, our study is also limited by our lack of data regarding robotic pancreatoduodenectomy (RPD) and we cannot provide cost data to support the increasing use of RPD2,5,11 or the impact of costs during the associated learning curve of RPD.23,24 Similar to LPD, the literature does suggest that RPD may offer some clinical outcome advantages over OPD.2,5,25,26 Thus, our finding of reduced healthcare needs upon discharge following LPD may suggest this benefit will also be realized with RPD. Taken together, the translation of our findings to other high-volume pancreatic surgery programs should be limited.

In summary, the study suggests that the cost of the index hospitalization for LPD is equivalent to OPD. However, total episode of care costs may favor LPD via reduced post-hospital needs for skilled nursing and rehabilitation.