The American Journal of Surgery (2016) -, 1–7

An economic analysis of pancreaticoduodenectomy: should costs drive consumer decisions? Thuy B. Tran, M.D., Monica M. Dua, M.D., David J. Worhunsky, M.D., George A. Poultsides, M.D., Jeffrey A. Norton, M.D., Brendan C. Visser, M.D.* Division of Surgical Oncology, Department of Surgery, Stanford University School of Medicine, 300 Pasteur Drive, Suite H3680D, Stanford, CA 94305, USA KEYWORDS: Cost analysis; Hospital volume; Pancreatico duodenectomy; Hospital charges; Whipple; Complication

Abstract BACKGROUND: Consumer groups campaign for cost transparency believing that patients will select hospitals accordingly. We sought to determine whether the cost of pancreaticoduodenectomy (PD) should be considered in choosing a hospital. METHODS: Using the Nationwide Inpatient Sample database, we analyzed charges for patients who underwent PD from 2000 to 2010. Outcomes were stratified by hospital volume. RESULTS: A total of 15,599 PDs were performed in 1,186 hospitals. The median cost was $87,444 (interquartile range $16,015 to $144,869). High volume hospitals (HVH) had shorter hospital stay (11 vs 15 days, P , .001) and mortality (3% vs 7.6%, P , .001). PD performed at low volume hospitals had higher charges compared with HVH ($97,923 vs $81,581, P , .001). On multivariate analysis, HVH was associated with a lower risk of mortality, while extremes in hospital costs, cardiac comorbidity, and any complication were significant predictors of mortality. CONCLUSION: Although PDs performed at HVH are associated with better outcomes and lower hospital charges, costs should not be the primary determinant when selecting a hospital. Ó 2016 Elsevier Inc. All rights reserved.

Price transparency, as part of the growing consumerism movement in health care, is still in its infancy. In the era of hospital risk–based reimbursements and pay for There were no relevant financial relationships or any sources of support in the form of grants, equipment, or drugs. The authors declare no conflicts of interest. Long Oral Presentation at the Americas Hepato-Pancreato-Biliary Association 2015 Annual Meeting, Miami Beach, Florida. * Corresponding author. Tel.: 11-650-721-1693; fax: 11-650-7249806. E-mail address: [email protected] Manuscript received May 28, 2015; revised manuscript October 10, 2015 0002-9610/$ - see front matter Ó 2016 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.amjsurg.2015.10.028

performance, there has been an increasing interest in cost transparency. However, the impact of cost-transparency initiatives on patient outcomes remains limited. The notion of cost-conscious ‘‘shopping,’’ while perhaps acceptable for basic medical goods such as prescription drugs, has not been investigated for complex procedures and services. Although proponents of cost transparency suggest that the wide variation in total hospital charges provide an opportunity for consumers to engage in more cost-conscious shopping, thereby selecting hospitals with lower prices for medical services, the relationship between quality of medical care (ie, outcomes) and hospital charges remains unknown.

2

The American Journal of Surgery, Vol -, No -, - 2016

For complex procedures such as pancreaticoduodenectomy (PD), referral to high volume centers has been strongly recommended. This is based on the Leapfrog Group initiative in 2000 that defined an annual institutional resection volume for several complex surgical procedures.1 For pancreatic resections, institutions performing a minimum of 11 pancreatic resections per year are classified as ‘‘high volume institutions.’’2 However, the minimum number of pancreatic resections set by the Leapfrog initiative has not been universally accepted. In fact, the number of pancreatic resections that defines ‘‘high volume hospital (HVHs)’’ for pancreatic surgery remains variable, ranging from 11 to 25 pancreatic resections per year.2–5 Despite the variable cut offs used to define high volume centers for pancreatic resection, the inverse relationship between hospital volume of complex surgical procedures and complications/mortality has been demonstrated in numerous retrospective studies.2,6–13 Although it is well established that surgeries at HVH are associated with better quality of care and outcomes, what remains unclear is the relationship between hospital costs and outcomes. Therefore, the aim of this study is to evaluate the interaction between hospital charges and hospital volume on postoperative short-term outcomes following PD using a national, multi-institutional administrative database. We also sought to investigate whether the cost of a PD should be a potential determinant when selecting a hospital. We hypothesized that hospital charges should not be the primary determinant when selecting a hospital for PD.

Department of Labor Consumer Price Index calculator in 2014 for each patient. To further understand the relationship between the presence of complications and hospital costs, hospitals were subdivided into 4 groups based on mean hospital charges: Quartile no. 1, ,$85,600; Quartile no. 2, $85,600 to $103,000; Quartile no. 3, $103,001 to $140,600; and Quartile no. 4, .$140,600. Patients who underwent PD were identified using the International Classification of Disease, Ninth Revision, Clinical Modification (ICD-9CM) procedure code 52.7 (radical PD). Elixhauser et al14 previously identified and defined comorbidities using ICD9-CM codes from administrative databasesdthese codes were used to classify important preoperative comorbidities. ICD-9 codes that are used to identify complications are shown in Supplemental Table 1. Subgroup analysis of complicated (defined as length of stay [LOS] . 14 days) vs uncomplicated PD (defined as LOS , 14 days) was performed to determine differences in hospital costs based on hospital type. We dichotomized PD into uncomplicated vs complicated based on LOS less than or greater than 14 days, respectively, because of the recent body of literature showing that the median LOS is generally less than 14 days in the hands of experienced pancreatic surgeons.15,16 Although it may appear arbitrary to define a complicated course as greater than 14-day stay based on reports from experienced pancreatic surgeons, one would also expect a hospital stay longer than 14 days for complicated PD performed at an LVH without experienced pancreatic surgeons. Categorical variables were presented as percentages and compared using Pearson chi-square test. Continuous variables were expressed as median and interquartile range and compared using Student t test for parametric variables or Wilcoxon rank-sum test for nonparametric variables. Multivariable logistic regression was performed to determine predictors of morbidity and in-hospital death. Statistical analyses were performed using SPSS version 22.0 (Chicago, IL). Statistical significance was set at P value less than .05.

Methods A retrospective analysis was performed using patient data collected from the Nationwide Inpatient Sample (NIS) file between 2000 and 2010. NIS is maintained by the Agency for Healthcare Research and Quality as part of the Healthcare Cost and Utilization Project and comprised hospital discharge records from more than 1,000 hospitals in the United States. NIS database represents approximating 20% sample of all hospital discharges in the United States. Data available within the NIS database include patient demographics, hospital characteristics, insurance information, diagnoses, inpatient procedures, inpatient mortality, and unadjusted hospital charges. Hospital volume was chosen based on prior studies investigating the volume–outcome relationship, including a recent NIS study that defined HVH as those that performed 19 and more PD per year while low volume hospitals (LVHs) performed fewer than 19 PD per year.4 We defined HVH as those centers that performed 20 and more PD per year (choosing a round number that reflected a reasonable compromise within the spectrum of reported cut offs). Individual hospital procedure volume was determined using the NIS-assigned hospital identification number to calculate the annual number of PD performed. Inflation-adjusted hospital costs were calculated using the United States

Results Patient demographics and hospital characteristics During the study period from January 2000 to December 2010, a total of 15,614 patients who underwent PD were identified in the database. Charity cases (defined as PD that is less than $10,000 in a recent study using the same database)17 were excluded from our analysis. After excluding charity cases, a total of 15,599 patients were analyzed. Stratifying by hospital volume, a total of 94 (8%) of 1,186 hospitals were categorized as HVH. Clinical characteristics and outcomes are detailed in Table 1. The median age of diagnosis was 65 years. Patients were predominantly elderly men. Minimally invasive PDs were more commonly performed in HVH (5.3% vs 3.5%,

T.B. Tran et al. Table 1

Economic analysis of pancreaticoduodenectomy

3

Clinical characteristics of the entire cohort

Hospital volume

Low volume (n 5 6,693)

High volume (n 5 8,906)

Entire cohort (n 5 15,599)

P value

Age (median, IQR) Female (%) Comorbidities (%) Cardiac Pulmonary Diabetes Obesity HTN CKD MIS (%) In-hospital mortality (%) Hospital costs (mean, SD) Postoperative outcomes (%) Prolong ventilation (.96 hours) Bleeding complication Cardiac complication Pulmonary complication Renal complication DVT Sepsis Wound infection Transfusion LOS

66 (56–74) 3,216 (48)

65 (55–73) 4,291 (48)

65 (56–73) 7,507 (48)

,.001 .842

444 (6.6) 1,764 (26.4) 366 (5.5) 1,456 (21.8) 2,744 (41) 98 (1.5) 235 (3.5) 512 (7.6) $138,362 ($140,738)

513 (5.8) 938 (10.5) 1,849 (20.8) 3,912 (42.8) 3,891 (43.7) 150 (1.7) 471 (5.3) 266 (3) $120,991 ($137,449)

960 (6.2) 1,838 (11.8) 3,305 (21.2) 6,472 (41.5) 6,635 (42.5) 248 (1.6) 706 (4.5) 788 (5) $128,259 ($139,094)

.018 ,.001 .133 ,.001 ,.001 .277 ,.001 ,.001 ,.001

325 102 444 1,764 366 105 1,010 159 1,986 15

216 116 466 1,602 297 122 1,050 226 1,884 11

541 218 910 3,366 633 227 2,060 385 3,870 12

,.001 .243 ,.001 ,.001 ,.001 .304 ,.001 .519 ,.001 ,.001

(4.9) (1.5) (6.6) (26.4) (5.5) (1.6) (15.1) (2.4) (29.7) (10–23)

(2.4) (1.3) (5.2) (18) (3.3) (1.4) (11.8) (2.5) (21.1) (8–16)

(3.5) (1.4) (5.8) (21.6) (4.3) (1.5) (13.2) (2.5) (21.2) (9–19)

CKD 5 chronic kidney disease; DVT 5 deep venous thrombosis; HTN 5 hypertension; IQR 5 interquartile range; LOS 5 length of stay; MIS 5 minimally invasive surgery; SD 5 standard deviation.

P , .001). A greater proportion of patients treated at HVH had significant cardiac comorbidities and obesity, whereas more patients treated at LVHs had underlying pulmonary comorbidities. The cost of a PD varied greatly, ranging from $10,170 to $3,602,252. The mean cost of PD was $128,259, with a standard deviation of $139,094. Fig. 1 illustrates the frequency of cases by total hospital charges. PDs performed at HVH were less costly than those performed at LVHs. The figure demonstrates a wider distribution of costs in LVHs than HVH. The mean cost of PDs was approximately $17,000 less expensive in HVH compared with LVHs ($120,991 vs $138,088, P , .001).

Patient outcomes Analysis of the entire cohort demonstrated that complications were less frequently observed in HVH compared with LVHs: renal failure requiring dialysis (.7% vs 1.3%, P , .001), cardiac complication (5.2% vs 6.6%), pulmonary complication (18% vs 26.4%, P , .001), sepsis (11.8% vs 15.1%, P , .001), blood transfusion (21.1% vs 29.7%, P , .001), need for blood transfusion (21.1 vs 29.7, P , .001), LOS (11 vs 15 days, P , .001), and hospital mortality (3% vs. 1.7%). Intraoperative complication rates were higher in LVHs than HVH (7.5% vs 5.7%, P , .001). When outcomes of uncomplicated PD (defined as LOS , 14 days) were evaluated, there were no significant differences in overall morbidity between LVHs and HVH (21% vs 19%). Along the same lines, the mean charges of uncomplicated PD were similar between LVHs and HVH ($78,466 vs $77,766). Outcomes of PD with LOS less than 14 days are shown in Supplemental Table 2.

Average hospital charges and multivariate analysis

Figure 1 Histogram of pancreaticoduodenectomy charges in high and low volume hospitals (entire cohort).

When patients are stratified into 4 quartiles based on average hospital costs, then there is an increased morbidity (47%) and a 2-fold increase in mortality (8%) in patients treated at hospitals in the highest quartile of charges for PD (.$140,000). The majority of hospitals in the highest

The American Journal of Surgery, Vol -, No -, - 2016

4 Table 2

Clinical characteristics of patients stratified by average hospital cost

Hospital volume Age (median, IQR) Female (%) Comorbidities (%) Cardiac Pulmonary Diabetes HTN Obesity CKD MIS (%) Hospital costs (mean, SD) Number of patients treated at high volume hospital (%) In-hospital mortality (%) Any postoperative complication (%) Postoperative outcomes (%) Prolong ventilation (.96 hours) Bleeding complication Cardiac complication Pulmonary complication Renal complication DVT Sepsis Wound infection Transfusion LOS

Quartile no. 1

Quartile no. 2

Quartile no. 3

Quartile no. 4

,$85k (n 5 3,801)

$85k–$103k (n 5 3,801)

$103k–140k (n 5 3,805)

.$140k (n 5 3,800)

65 (56–73) 1,845 (48.5) 203 479 786 1,560 124 35 201 68,672 2,028

(5.3) (12.6) (20.7) (41) (3.3) (.9) (5.3) (49,247) (53.4)

152 (4) 1,107 (29.1) 66 37 174 629 107 36 396 76 106 11

66 (56–73) 1,837 (48.3) 254 442 842 1,699 147 75 173 93,580 2,846

(6.7) (11.6) (22.2) (44.7) (3.9) (2) (4.6) (76,134) (74.9)

142 (3.7) 1,193 (31.4)

(1.7) (1.0) (4.6) (16.5) (2.8) (.9) (10.4) (2) (29.5) (8–17)

108 56 198 666 136 50 441 81 49 11

(2.8) (1.5) (5.2) (17.5) (3.6) (1.3) (11.6) (2.1) (29.5) (8–18)

65 (56–73) 1,809 (47.6) 204 420 777 1,619 124 53 138 119,441 2,449

(5.4) (11) (20.4) (42.5) (3.3) (1.4) (4.6) (102,880) (64.4)

165 (4.3) 1,296 (34.1) 119 51 211 808 152 62 492 102 74 12

(3.1) (1.3) (5.5) (21.2) (4) (1.6) (12.9) (2.7) (35.6) (9–18)

65 (56–73) 1,831 (48.3) 273 448 815 1,568 115 67 171 230,151 1,503

(7.2) (11.8) (21.4) (41.3) (3) (1.8) (4.5) (205,988) (39.6)

P value .555 .880 .001 .206 .241 .004 .208 .001 .006 ,.001 ,.001

299 (7.9) 1,748 (46)

.004 ,.001

227 62 298 1,194 255 73 682 112 152 16

,.001 .080 ,.001 ,.001 ,.001 .003 ,.001 .220 .138 ,.001

(6) (1.6) (7.8) (31.4) (6.7) (1.9) (17.9) (2.9) (36.3) (11–24)

CKD 5 chronic kidney disease; DVT 5 deep venous thrombosis; HTN 5 hypertension; IQR 5 interquartile range; LOS 5 length of stay; MIS 5 minimally invasive surgery; SD 5 standard deviation.

quartile of average hospital charges for PD tended to be LVHs (Table 2). When HVH was ranked by average cost of PDs, there was a weakly linear relationship between average hospital cost and in-hospital mortality (Fig. 2). However, Fig. 2 also

Figure 2 Mortality related to average hospital cost among high volume hospitals. There is a weakly linear relationship between average hospital cost and mortality.

illustrates the tremendous variation in average hospital charges for PD even among high volume institutions. If the low volume patients were treated in HVHs, theoretically the cost savings would be in the magnitude of $116,264,103 in the 10-year study period (calculated using the equation mean LVH charges 2 mean HVH charges ! number cases treated at LVHs), or $116 million dollars saved in the 20% of all hospital discharges represented in this analysis. Univariate analysis demonstrated a greater risk of mortality with blood transfusion (hazard ratio [HR] 1.696, P , .001), cardiac comorbidity (HR 2.098, P , .001), any postoperative complication (HR 14.020, P , .001), and care in hospitals in which the average cost of PD was in the highest quartile (4th quartile . $140,600; odds ratio 2.054, P , .001). Costs in the second and third quartile were not associated with a greater risk of mortality (compared with lowest quartile hospitals). On multivariate analysis, extreme of hospital costs (fourth quartile hospitals), any postoperative complication, and blood transfusion were significant predictors of in-hospital mortality, while female sex and HVH were associated with better outcomes (Table 3).

T.B. Tran et al. Table 3

Economic analysis of pancreaticoduodenectomy

5

Univariate and multivariate analysis of mortality Univariate

Multivariate

Hospital volume

Odds ratio

95% CI

P value

Odds ratio

95% CI

P value

Average hospital cost Quartile no. 1 Quartile no. 2 Quartile no. 3 Quartile no. 4 High volume hospital Female Transfusion MIS Cardiac comorbidity Any complications

Reference .932 1.089 2.054 .371 .745 1.696 .778 2.098 14.020

.738–1.176 .869–1.364 1.680–2.510 .319–.432 .644–.862 1.457–1.974 .529–1.144 1.667–2.641 11.343–17.329

.552 .459 ,.001 ,.001 ,.001 ,.001 .202 ,.001 ,.001

Reference 1.050 1.022 1.296 .454 .845 1.177 d 1.561 11.938

.821–1.342 .808–1.292 1.050–1.599 .386–.536 .724–.868 1.000–1.384 d 1.220–1.999 9.633–14.795

.698 .859 .016 ,.001 .033 .050 d ,.001 ,.001

CI 5 confidence interval; MIS 5 minimally invasive surgery.

Comments There is growing interest among payers in shifting a greater portion of the burden of healthcare costs to patients and hospitals via higher patient copays, risk-based reimbursements, capitation, and bundled payments. This study examines the impact of hospital charges on patient outcomes for PDs performed between 2000 and 2010. PDs performed at HVH are associated with lower morbidity and mortality, a shorter median LOS, and ultimately lower charges. Adjusting for hospital and patient factors, high charges associated with PD remain an independent predictor of mortality. Although hospital volume is deemed a benchmark of quality of care, our study demonstrates that, under the current fee-for-service paradigm, average hospital charges associated with complex procedures may serve as a rough surrogate of outcomes. More specifically, we determined that the presence of complications leads to prolonged LOS and ultimately higher costs. However, among HVH the cost–outcome relationship largely evaporates. Over the last 2 decades, health advocacy organizations have encouraged patients to consider hospital volume when deciding where to undergo complex procedures. Although this has resulted in modest centralization of complex cases,18 healthcare expenditures have continued to rise. Thus policies are emerging to contain costs by shifting a greater portion of the bill away from the payers first to hospitals and then ultimately to patients. Of particular relevance to our discussion is the notion of ‘‘bundled payment’’dreimbursement of healthcare providers (both hospitals and physicians) on the basis of expected costs for clinically defined episodes of care (eg, a PD). These rates are negotiated by hospitals based on average costs of services to avoid underpayment. In an emerging era of ‘‘bundling,’’ it is therefore important to understand whether the average charges for PDs have been (historically) under the fee-for-service model. The second emerging trend in healthcare billing that is relevant to our discussion is the

effort by employer healthcare plans to contain costs by making employees pay more through copayments, deductibles, and coinsurance. Although copayments and deductibles provide modest negative incentive for patients to seek care by transferring upfront cost, it is coinsurance that has the most potential to drive consumer choices. Coinsurance means, simply put, that the patient pays a percentage of the final bill. It is believed that all these strategies will make the patient (read: ‘‘consumer’’) more aware of the price of health care because they will ‘‘share’’ a greater percentage of the cost. We found that, under the current fee-for-service system, there is a tremendous variation in the charges associated with PD. Among all hospitals, the median cost ranged from $10,781 to $1,343,557. The standard deviation almost equals the mean, further emphasizing the wide variation in charges. A sizable portion of the variation does appear to be explained by outcomes. PD is an operation that is notoriously vulnerable to postoperative complications. Complications are costly; therefore, on both univariate and multivariate analyses, worse outcomes were associated with the highest charges. Our data reaffirm the strong relationship between volume and outcome for PD.9–11 LVHs had more complications and charged more (longer stay, more care, more costly). Thus, not surprisingly, ‘‘uncomplicated’’ PDs were much cheaper. And, moreover, the difference in price between HVH and LVHs for these uncomplicated cases evaporated (both HVH and LVHs charged, on average, just about $78,000, which is about 60% of the overall average price of a PD). Therefore, in the fee-for-service model, charges are an inverse (and perhaps somewhat perverse) predictor of quality. The tremendous variation in charges for the PD cannot be fully explained by differences in outcomes. Among HVH, the relationship between charges and outcomes appears very weak. In the analysis of PD charges by quartiles, only the fourth quartile charges demonstrate a significantly greater likelihood of morbidity and mortality. Although some cost drivers have been defined elsewhere

6

The American Journal of Surgery, Vol -, No -, - 2016

for hospital charges more broadly (geographic regional differences, payer-mix differences, etc.),19–25 these will require more investigation to see if they apply specifically to the PD operation. But, ultimately, it seems likely that a good portion of the cost variation for the PD will not be explained by all these elements, and the patient/consumer will be left wondering how the average price for the ‘‘same’’ operation at hospital X can be two-fold that of hospital Y. The charges data for PD presented here have important implications. If we are to move to bundled payments (ie, negotiated rates based on average historic costs of services), those costs cannot be based on idealized uncomplicated cases alone, especially for an operation like the PD with a high rate of morbidity. But higher negotiated rates based on higher average charges at LVHs cannot be justified, especially given the higher than expected morbidity associated with PD performed at these centers. The median price and the variation in price are clearly lower for HVH. Even in the current reimbursement model, there would be tremendous cost saving if payers urged patients to seek care at HVH. For our cohort of roughly 16,000 patients, that savings (ie, reduction in healthcare expenditure) might be on the order of magnitude of 116 million dollars. If we multiply that by 5 (because NIS is a 20% sample), we get $580 million in potential savings over the study period on just one, relatively uncommon, operation. Finally, as payers begin to increasingly shift the burden to patients by means of coinsurance, the high out-of-pocket expense at LVHs may not be justified. This argues for cost transparency for complex surgical procedures, which will allow patients to select experienced hospitals that can perform the same services with lower morbidity, mortality, and costs. There are several important limitations in this study. First, although the NIS database provides generalizable short-term results on aggregate costs and outcomes after PD, it does not provide an itemized breakdown of charges or control for the effects of market forces that can influence hospital charges. In other words, contracts between insurers and providers may dictate costs of PD such that some providers may charge more than others for the same procedure, regardless of hospital or surgeon volume. Second, we could not determine if high hospital costs have an adverse impact on long-term overall survival after PD as NIS only provides short-term outcomes. For example, for a complex operation like PD, outpatient costs (ie, skilled nursing facility costs, home health) and readmissions may not be insignificant, but we cannot these specific parameters using this database. Third, we only analyzed hospital volume rather than account for the effect of individual surgeon volume on operative mortality. Fourth, as in all administrative databases, there is potential miscoding of robotic cases in administrative data. Finally, retrospective analysis of an administrative database limits our ability to assess that the impact of procedure-related outcomes estimated blood loss and operative time. However, using the NIS provides insight on differences in

outcomes with respect to hospital volume and total hospital charges that may be consider in future healthcare policies on cost transparency. Cost transparency may be part of the solution to the rising healthcare expenditures by allowing market forces to encourage competition among hospitals and providers that provide similar services. Especially with the introduction of the Affordable Care Act in 2010, price transparency will continue to be an important component in the financial landscape of medicine. The informed patient requiring PD should use volume as the primary determinant when selecting a hospitaldthe superior outcomes of HVH compared with LVHs are unequivocally established. However, among HVH, there seems to be little relationship between charges and outcomes, so cost could reasonably be a secondary consideration where one chooses to seek care.

Supplementary Data Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.amjsurg.2015.10.028

References 1. Milstein A, Galvin RS, Delbanco SF, et al. Improving the safety of health care: the leapfrog initiative. Eff Clin Pract 2000;3:313–6. 2. Birkmeyer JD, Finlayson EV, Birkmeyer CM. Volume standards for high-risk surgical procedures: potential benefits of the leapfrog initiative. Surgery 2001;130:415–22. 3. Fong Y, Gonen M, Rubin D, et al. Long-term survival is superior after resection for cancer in high-volume centers. Ann Surg 2005;242: 540–4; discussion 544–7. 4. Meguid RA, Ahuja N, Chang DC. What constitutes a ‘‘high-volume’’ hospital for pancreatic resection? J Am Coll Surg 2008;206:622.e1–9. 5. McPhee JT, Hill JS, Whalen GF, et al. Perioperative mortality for pancreatectomy: a national perspective. Ann Surg 2007;246:246–53. 6. Choti MA, Bowman HM, Pitt HA, et al. Should hepatic resections be performed at high-volume referral centers? J Gastrointest Surg 1998;2: 11–20. 7. Finlayson EV, Goodney PP, Birkmeyer JD. Hospital volume and operative mortality in cancer surgery: a national study. Arch Surg 2003; 138:721–5; discussion 726. 8. Lieberman MD, Kilburn H, Lindsey M, et al. Relation of perioperative deaths to hospital volume among patients undergoing pancreatic resection for malignancy. Ann Surg 1995;222:638–45. 9. Sosa JA, Bowman HM, Gordon TA, et al. Importance of hospital volume in the overall management of pancreatic cancer. Ann Surg 1998; 228:429–38. 10. Birkmeyer JD, Finlayson SR, Tosteson AN, et al. Effect of hospital volume on in-hospital mortality with pancreaticoduodenectomy. Surgery 1999;125:250–6. 11. Birkmeyer JD, Warshaw AL, Finlayson SR, et al. Relationship between hospital volume and late survival after pancreaticoduodenectomy. Surgery 1999;126:178–83. 12. Hannan EL, Radzyner M, Rubin D, et al. The influence of hospital and surgeon volume on in-hospital mortality for colectomy, gastrectomy, and lung lobectomy in patients with cancer. Surgery 2002;131:6–15. 13. Hollenbeck BK, Wei Y, Birkmeyer JD. Volume, process of care, and operative mortality for cystectomy for bladder cancer. Urology 2007; 69:871–5.

T.B. Tran et al.

Economic analysis of pancreaticoduodenectomy

14. Elixhauser A, Steiner C, Harris DR, et al. Comorbidity measures for use with administrative data. Med Care 1998;36:8–27. 15. Schmidt CM, Turrini O, Parikh P, et al. Effect of hospital volume, surgeon experience, and surgeon volume on patient outcomes after pancreaticoduodenectomy: a single-institution experience. Arch Surg 2010;145:634–40. 16. Tseng JF, Pisters PW, Lee JE, et al. The learning curve in pancreatic surgery. Surgery 2007;141:694–701. 17. Broderick RC, Fuchs HF, Harnsberger CR, et al. The price of decreased mortality in the operative management of diverticulitis. Surg Endosc 2015;29:1185–91. 18. Gasper WJ, Glidden DV, Jin C, et al. Has recognition of the relationship between mortality rates and hospital volume for major cancer surgery in California made a difference?: a follow-up analysis of another decade. Ann Surg 2009;250:472–83. 19. Barker AR, McBride TD, Kemper LM, et al. Geographic variation in premiums in health insurance marketplaces. Rural Policy Brief 2014; 2014 10:1–4.

7

20. Franzini L, White C, Taychakhoonavudh S, et al. Variation in inpatient hospital prices and outpatient service quantities drive geographic differences in private spending in Texas. Health Serv Res 2014;49: 1944–63. 21. Franzini L, Mikhail OI, Zezza M, et al. Comparing variation in Medicare and private insurance spending in Texas. Am J Manag Care 2011; 17:e488–95. 22. Zuckerman S, Waidmann T, Berenson R, et al. Clarifying sources of geographic differences in Medicare spending. N Engl J Med 2010; 363:54–62. 23. Cooper RA. Geographic variation in health care and the affluencepoverty nexus. Adv Surg 2011;45:63–82. 24. LaPar DJ, Bhamidipati CM, Mery CM, et al. Primary payer status affects mortality for major surgical operations. Ann Surg 2010;252: 544–50; discussion 550–1. 25. Boxer LK, Dimick JB, Wainess RM, et al. Payer status is related to differences in access and outcomes of abdominal aortic aneurysm repair in the United States. Surgery 2003;134:142–5.

The American Journal of Surgery, Vol -, No -, - 2016

7.e1

Supplemental Table 1

ICD-9-CM coding for procedures and complications

Procedure Pancreaticoduodenectomy MIS pancreaticoduodenectomy Complications Prolonged mechanical ventilation (.96 hours) Bleeding complication Cardiac complication Pulmonary complication Renal complication Deep vein thrombosis Sepsis Wound infection Transfusion

52.7 52.7 1 54.21 96.72 568.81, 998.11, 998.12 427.0–427.32, 427.5, 427.41–427.42, 428.21, 428.31, 997.1, 785.51 507.0, 512.2, 518.5, 518.81–518.82, 997.30–997.39 584.6–584.9 453.40–453.42, 453.8–453.89 995.90–995.94 998.12, 998.13, 998.31, 998.32, 998.34, 998.6, 998.83, 997.5 99.0–99.05

ICD-9-CM 5 International Classification of Disease, Ninth Revision, Clinical Modification; MIS 5 minimally invasive surgery.

Supplemental Table 2

Clinical outcomes of patients with hospital length of stay less than 14 days (‘‘uncomplicated PD’’)

Hospital volume

Low volume (n 5 2,686)

High volume (n 5 5,627)

Entire cohort (n 5 8,313)

P value

Postoperative outcomes (%) Bleeding complication Cardiac complication Pulmonary complication Renal complication DVT Wound infection

14 84 386 33 10 17

20 190 624 56 26 54

34 274 1,010 89 36 71

.268 .552 ,.001 .334 .560 .130

(.5) (3.1) (14.4) (1.2) (.4) (.6)

DVT 5 deep venous thrombosis; PD 5 pancreaticoduodenectomy.

(.4) (3.4) (11.1) (1) (.5) (1)

(.4) (3.3) (12.1) (1.1) (.4) (.9)

An economic analysis of pancreaticoduodenectomy: should costs drive consumer decisions?

Consumer groups campaign for cost transparency believing that patients will select hospitals accordingly. We sought to determine whether the cost of p...
360KB Sizes 4 Downloads 11 Views