National trends in resection of cystic lesions of the pancreas

HPB

http://dx.doi.org/10.1016/j.hpb.2015.11.002

ORIGINAL ARTICLE

National trends in resection of cystic lesions of the pancreas Bradley N. Reames, Christopher P. Scally, Timothy L. Frankel, Justin B. Dimick & Hari Nathan Department of Surgery, University of Michigan, Ann Arbor, MI, USA

Abstract Background: Management of cystic lesions of the pancreas (CLP) is controversial. In this study, we sought to evaluate national changes in the resection of CLP over time, to better understand the impact of evolving guidelines on CLP management. Methods: We used Medicare data to examine CLP resection among patients undergoing pancreatic resection between 2001 and 2012. Patients with a diagnosis of CLP were identified and compared to patients with non-CLP indications. We then examined changes over time in patient and hospital characteristics and outcomes among patients with a CLP diagnosis. Results: We identified 56,419 Medicare patients undergoing pancreatic resection, of which 2129 had a CLP diagnosis. The annual number of CLP resections, and proportion of all resections performed for CLP increased significantly during the period, from 2.1% (65/3072) resections in 2001, to 4.5% (286/6348) in 2012 (p < 0.001). The proportion of CLP resections with a malignant diagnosis did not change (15.5% in 2001–2003 vs. 13.1% in 2010–2012, p = 0.4). Overall rates of 30-day mortality decreased significantly during the period (9.6% in 2001–2003 vs. 5.5% in 2010–2012, p < 0.001). Discussion: CLP resections were performed with increasing frequency in Medicare patients between 2001 and 2012, but this did not correspond to increased diagnosis of malignancy. Additional research is needed to understand the influence of recent guidelines on management of CLP. Received 6 October 2015; accepted 9 November 2015

Correspondence: Hari Nathan, University of Michigan, 2210D Taubman Center, 1500 E Medical Center Dr, Ann Arbor, MI 48109-5343, USA. Tel: +1 734 936 7607. Fax: +1 734 232 6188. E-mail: [email protected]

Introduction Cystic lesions of the pancreas (CLP) are being identified with increasing frequency, but the natural history of certain CLP lesions, such as intraductal papillary mucinous neoplasms (IPMN), remains poorly understood. Studies of abdominal imaging suggest as many as 44.7% of patients have an incidentally found CLP,1,2 while 24.3% of patients were discovered to have a CLP at autopsy.3 Though certain cyst features are widely recognized to be associated with increased malignancy risk,4 studies of CLP meeting criteria for surveillance have reported varying rates of malignant progression.4–6 As a result,

National Presentation: This work was presented as an oral presentation at the Society of Surgical Oncology Annual Cancer Symposium in Houston, TX, on March 28th, 2015.

HPB 2016, 18, 375–382

uncertainty exists regarding the optimal management strategy: efforts to halt progression to pancreatic cancer must be balanced with the morbidity of resection and the increasing incidence of diagnosis. The management of CLP has evolved over time: though routine resection was previous considered the standard by many,7,8 advances in radiographic and endoscopic techniques have led to increased support for a more selective approach.1,9,10 Consensus guidelines affirming this approach were released in 2006, and were recently updated in 2012.11,12 Despite these guidelines, the optimal management of CLP remains controversial.4,6,13–17 Given this controversy, the influence of advancing technology and changing recommendations on management of CLP in the real world is unclear. Previous literature examining current practice is limited to the experiences of high volume, singleinstitution centers.4,6,18,19 Because resection criteria likely vary

© 2016 International Hepato-Pancreato-Biliary Association Inc. Published by Elsevier Ltd. All rights reserved.

HPB

376

according to surgeon and institutional preferences,20,21 it is unclear whether these experiences represent practice nationally. To date, no studies have evaluated national trends in resection for CLP. A better understanding of how recommendations for management of CLP are applied nationally could have important implications for the development and dissemination of future guidelines. In this study we sought to examine trends in resection of CLP in the United States over a recent twelve-year period. To do this, we used national Medicare data to identify all patients undergoing pancreatic resection between the years 2001 and 2012. We examined resection rates among patients with and without a diagnosis of CLP, and we assessed changes over time in the patient and hospital characteristics, and outcomes, of patients with a CLP diagnosis.

Methods Data source and study population To complete this study, we used analytic files for the years 2001–2012 from the Center for Medicare and Medicaid Services. The Medicare Provider Analysis and Review (MEDPAR) file was used to create the primary study dataset, while the Medicare Denominator file was used to determine the vital status of all patients 30 days after surgery. The Institutional Review Board of the University of Michigan and the CMS approved this protocol and waived the requirement for informed consent. Using appropriate procedure codes from the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM), we identified Medicare patients between the ages of 65 and 99 years old who underwent a pancreatic resection (52.51, 52.52, 52.53, 52.59, 52.6, 52.7). Patients undergoing pancreaticogastrostomy or pancreaticoenterostomy (52.96) alone, and those undergoing transplant-related procedures (52.8) were excluded. Indications for resection The indication for pancreatic resection in Medicare data was identified by appropriate ICD-9-CM codes listed in any of the ten MEDPAR diagnosis fields. A patient was considered to have a CLP diagnosis if the ICD-9-CM code 577.2 was present in any of the diagnosis fields, regardless of other diagnoses present. Patients with pancreatic cancer were identified by the presence of diagnosis codes 157.0–157.9, while patients with duodenal, biliary, or ampullary cancer were identified by diagnosis codes 152.0 and 156.0–156.9, and patients with a neuroendocrine tumor were identified by diagnosis codes 209.0–209.39. To increase the homogeneity of our sample and minimize confounding, patients with an ICD-9-CM diagnosis code for pancreatitis (577.0 or 577.1) or transplant-related complications (996.59–996.89) in any field were excluded from the final analysis.

HPB 2016, 18, 375–382

Patient and hospital characteristics We evaluated patient characteristics including age, sex, race, and comorbidities. Comorbidities were identified by the appropriate ICD-9-CM diagnosis codes and defined using the Elixhauser method.22 A patient was considered to have malignancy if diagnosis codes for pancreatic, duodenal, biliary, or ampullary cancers, or neuroendocrine tumor, were present. We obtained characteristics of hospitals from the American Hospital Association (AHA) Annual Survey for the years 2008–2012. These data included bed size, teaching status, hospital ownership, urban or rural location, cancer center designation, critical access designation, availability of endoscopic ultrasound (EUS) and endoscopic retrograde cholangiopancreatography (ERCP), and level of technology. A hospital was considered to have “high technology” if it performed cardiac surgery and/or solid organ transplantation, in accordance with previous literature.23,24 Medicare hospital volume was determined by calculating the total number of pancreatic resections in the final study cohort performed by each hospital. Hospitals were then ranked, and quintiles were created by defining whole number cutoffs that would separate all patients into five categories of equal size, similar to prior studies.25,26 Patient outcomes evaluated in this study included mortality, post-operative complications, and failure to rescue. Operative mortality was defined as death within 30 days of the index procedure or before hospital discharge. Post-operative complications were identified by ICD-9-CM codes using previously validated methods.27,28 Failure to rescue was defined as a mortality among patients with one or more of the defined major complications.29 Statistical analysis We calculated descriptive statistics and rates of unadjusted surgical outcomes for patients stratified by procedure type and indication for resection. Baseline patient and hospital characteristics were compared using student’s t-test or Wilcoxon’s rank sum test for normally or non-normally distributed continuous data, and chi-squared tests for categorical data. We used Cuzick’s test for trend to assess differences over time in patient and hospital characteristics, surgical outcomes, and resection rates stratified by procedure type and indication for resection. All statistical analyses were performed using STATA version 12.1 (StataCorp, College Station, TX), with two-sided tests and alpha set at 0.05.

Results The final cohort included 56,419 Medicare patients undergoing pancreatic resection during the 12-year period. Details regarding patient characteristics and type of resection, stratified by the indication for resection, are shown in Table 1. The median age of all patients was 74 years old, with a slight female majority

© 2016 International Hepato-Pancreato-Biliary Association Inc. Published by Elsevier Ltd. All rights reserved.

HPB

377

Table 1 Patient characteristics stratified by indication for pancreatic resection, and significance of changes in characteristics in CLP pa-

tients during the 12-year period (CLP: cystic lesion of the pancreas) Patient characteristics

Entire cohort

CLP patients

Cystic lesion of the pancreas

Other surgical indications

2129

54,290

72

74

65–69

712 (33.5)

14,177 (26.1)

70–74

626 (29.4)

15,749 (29.0)

75–79

504 (23.7)

13,950 (25.7)

80–84

223 (10.5)

7763 (14.3)

85+

63 (3.0)

2651 (4.9)

Number of patients, N

p-Value

p-Value for trend over time

Demographics Age, years, median Age categories

Female

1267 (59.5)

27,969 (51.5)

African American race

159 (7.5)

4405 (8.1)

Number of comorbidities, mean

2.0

3 comorbidities Hypertension

<0.001

0.589

<0.001

0.924

<0.001

0.609

0.284

0.414

2.2

<0.001

<0.001

652 (30.6)

20,954 (38.6)

<0.001

<0.001

1216 (57.1)

27,514 (50.7)

<0.001

<0.001

Diabetes mellitus

455 (21.4)

12,332 (22.7)

0.146

0.075

Chronic lung disease

343 (16.1)

7008 (12.9)

<0.001

0.449

Obesity

122 (5.7)

2126 (3.9)

<0.001

<0.001

Congestive heart failure

103 (4.8)

2773 (5.1)

0.579

0.982

Liver disease

57 (2.7)

1157 (2.1)

280 (13.2)

38,062 (70.1)

Comorbidities

Malignant diagnosis Type of pancreatectomy Pancreaticoduodenectomy

558 (26.2)

33,597 (61.9)

Distal and subtotal

1258 (59.1)

15,609 (28.8)

Other partial

200 (9.4)

2594 (4.8)

Total

84 (4.0)

1906 (3.5)

Proximal partial

29 (1.4)

584 (1.1)

(51.8%). Compared to patients undergoing resection for other indications, patients with a diagnosis of CLP were younger (33.5% <70 years old vs. 26.1%, p < 0.001) and more likely to be female (59.5 vs. 51.5%, p < 0.001), but less likely to have malignancy (13.2% vs. 70.1%, p < 0.001), and healthier (30.6% with 3 Comorbidities vs. 38.6%, p < 0.001). The types of resection performed varied significantly according to indication. Pancreaticoduodenectomy, for example, was performed in a minority of CLP patients (26.2%), but a majority of patients with other indications for resection (61.9%, p < 0.001). Table 1 also illustrates changes over time in the characteristics of CLP patients undergoing resection. There were no significant trends in patient age, gender, race, presence of malignancy, and type of resection performed. The mean number of comorbidities increased from 1.7 in 2001 to 2.4 in 2012 (p < 0.001), as did rates of hypertension (48.9% vs. 65.0%, p < 0.001) and obesity (4.6% vs. 11.2%, p < 0.001).

HPB 2016, 18, 375–382

0.088

0.478

<0.001

0.786

<0.001

0.248

Hospital characteristics, stratified by indication for resection, are shown in Table 2, along with the significance of changes over time in the CLP cohort. A majority of patients underwent resection at hospitals that are teaching centers (85.2%), have high technology (89.3%), have access to EUS & ERCP (87.9%), are located in urban areas (99.2%), are not-for-profit (80.6%), and have >500 beds (52.6%). Less than 1% of patients received operations at a critical access hospital, while 77.8% received their procedure at a designated cancer center. Compared to patients with other indications for resection, patients with a diagnosis of CLP were slightly more likely to undergo resection at high and very high volume hospitals (43.5% vs. 39.7%, p < 0.001), teaching hospitals (87.2% vs. 85.2%, p = 0.03), high technology hospitals (91.0% vs. 89.2%, p = 0.01), and centers with EUS and ERCP (90.6% vs. 87.8%, p < 0.001). During the 12-year period, there were numerous significant trends in the characteristics of hospitals performing pancreatic

© 2016 International Hepato-Pancreato-Biliary Association Inc. Published by Elsevier Ltd. All rights reserved.

HPB

378

Table 2 Hospital characteristics stratified by indication for pancreatic resection, and significance of changes in characteristics in CLP

patients during the 12-year period (CLP: cystic lesion of the pancreas; EUS: Endoscopic Ultrasound; ERCP Endoscopic Retrograde Cholangiopancreatography) Hospital characteristics

Entire cohort

CLP patients

Cystic lesion of the pancreas

Other surgical indications

Number of patients, N

2129

54,290

Number of hospitals, N

642

2331

Very low & low volume

772 (36.3)

22,016 (40.1)

High & very high volume

926 (43.5)

21,555 (39.7)

1 (0.05)

42 (0.08)

Pancreatectomy volume

Critical access hospital Hospital size <200 beds

145 (6.8)

4309 (7.9)

200–500 beds

863 (40.5)

21,085 (38.8)

>500 beds

1108 (52.0)

28,564 (52.6)

Government

240 (11.3)

6666 (12.3)

Investor owned

121 (5.7)

3478 (6.4)

Not-for-profit

1751 (82.2)

43,720 (80.5)

Hospital ownership

Hospital teaching status Teaching

1857 (87.2)

46,237 (85.2)

Non-teaching

259 (12.2)

7721 (14.2)

p-Value

p-Value for trend over time

<0.001

0.001

0.883

0.532

0.176

0.032

0.248

0.667

0.028

<0.001

Urban location

2113 (99.3)

53,829 (99.2)

0.663

0.594

High technology hospital

1937 (91.0)

48,447 (89.2)

0.011

0.004

Cancer center

1694 (79.6)

42,206 (77.7)

0.133

0.194

Availability of EUS & ERCP

1929 (90.6)

47,652 (87.8)

<0.001

0.286

resection on CLP patients. The proportion of CLP resections performed in very low and low volume hospitals (60.0% in 2001 vs. 36.4% in 2012, p = 0.001) and hospitals with <200 beds (10.9% vs. 5.3%, p = 0.03) decreased substantially over time, while the proportion in teaching hospitals (79.7% vs. 91.6%, p < 0.001), and high technology hospitals (80.0% vs. 90.9%, p = 0.004) increased significantly. Overall, rates of any complication (17.7%, vs. 27.4%, p < 0.001), failure to rescue (13.3%, vs. 19.4%, p = 0.003), and operative mortality (3.1%, vs. 7.0%, p < 0.001) were significantly lower in CLP patients. Table 3 reports changes over time in unadjusted surgical outcomes, stratified by procedure type and indication for resection. Although the sample sizes of CLP patients were small, significant improvements over time were seen in the operative mortality of CLP patients undergoing distal pancreatectomy (4.1% vs. 1.6%, p = 0.020). In patients with other indications for resection, rates of both failure to rescue and operative mortality decreased significantly over time in both pancreaticoduodenectomy and distal pancreatectomy. Changes over time in the annual number and proportion of CLP resections reported in Medicare data are demonstrated in Fig. 1. The number of CLP resections increased 4-fold during the

HPB 2016, 18, 375–382

study period, from 65 resections in 2001, to 286 CLP resections in 2012. Similarly, the proportion of all resections performed for CLP doubled, from 2.1% in 2001, to 4.5% in 2012 (p < 0.001 for trend). To investigate the relationship between CLP resection and malignancy, we evaluated changes over time in the proportion of CLP patients with and without a concurrent diagnosis of malignancy. As shown in Fig. 2a, the proportion of CLP resections with a malignant diagnosis did not change over time during the 12-year period (15.5% in 2001–2003 vs. 13.1% in 2010–2012, p = 0.4 for trend), while Fig. 2b shows nearly identical rates of malignancy in CLP patients before and after release of international consensus guidelines (15.6% before, vs. 15.7% after, p = 0.9). In patients 80 years and older, the trend over time was not significant, although the overall rate of malignancy was higher in this subgroup (19.4% in 2001–2003, vs. 27.2% in 2010–2012, p = 0.6 for trend).

Discussion In a large, nationwide cohort of Medicare patients, we found that the annual number of pancreatic resections performed for

© 2016 International Hepato-Pancreato-Biliary Association Inc. Published by Elsevier Ltd. All rights reserved.

HPB

379

Table 3 Changes in unadjusted rates of adverse outcomes during the years 2001–2012, in both CLP and non-CLP cohorts, stratified by the

2 most common procedures (CLP: cystic lesion of the pancreas) Adverse outcomes

2001–2003

2004–2006

2007–2009

2010–2012

p-Value*

Pancreaticoduodenectomy Number of patients, N CLP indication Non-CLP indication

63 (1.0)

97 (1.3)

184 (2.0)

6240 (99.0)

7273 (98.7)

9223 (98.0)

214 (1.9) 10,861 (98.1)

Any complication CLP indication Non-CLP indication

12 (19.0)

24 (24.7)

44 (23.9)

49 (22.9)

0.758

1819 (29.2)

2128 (29.6)

2879 (31.2)

3413 (31.4)

<0.001

Failure to rescue CLP indication Non-CLP indication

5 (41.7)

3 (12.5)

5 (11.3)

8 (16.3)

0.205

475 (26.1)

443 (20.8)

523 (18.2)

566 (16.6)

<0.001

Operative mortality CLP indication Non-CLP indication

7 (11.1) 618 (9.9)

5 (5.2)

7 (3.8)

9 (4.2)

0.065

578 (8.0)

637 (6.9)

684 (6.3)

<0.001

Distal pancreatectomy Number of patients, N CLP indication Non-CLP indication

147 (5.6)

236 (6.7)

365 (7.7)

510 (8.5)

2464 (94.4)

3265 (93.3)

4366 (92.3)

5514 (91.5)

Any complication CLP indication Non-CLP indication

15 (10.2)

37 (15.7)

54 (14.8)

72 (14.1)

0.539

523 (21.2)

676 (20.7)

939 (21.5)

1136 (20.6)

0.661

Failure to rescue CLP indication Non-CLP indication

3 (20.0)

8 (21.6)

118 (22.6)

134 (19.8)

3 (5.6) 145 (15.4)

8 (11.1)

0.123

157 (13.8)

<0.001

Operative mortality CLP indication Non-CLP indication

6 (4.1)

9 (3.8)

5 (1.4)

8 (1.6)

0.020

200 (8.1)

203 (6.2)

202 (4.6)

223 (4.0)

<0.001

* p-value for the trend over time in that group, for the specified complication.

diagnosis of CLP has increased four-fold over a recent 12-year period, while the proportion of all resections performed for CLP doubled. Though patients undergoing resection today have greater comorbid disease (obesity and hypertension) than 10 years ago, crude outcomes following resection have improved significantly, possibly due to the concentration of CLP resections at large, high volume teaching centers with more resources, as shown in these results. Yet despite the increasing annual volume of CLP resections over time, the proportion of malignant diagnoses has remained stable. Current literature discussing the management of certain CLP (such as IPMN) mostly reflects the experiences of a limited number of high volume institutions, and approaches to management often vary. For example, Gaujoux and colleagues, who advocate a selective approach to resection, found that rates of initial resection decreased during a recent 15-year period, from 43% between 1995 and 2005, to 33% between 2005 and 2010. Only 6.5% of 1141 patients undergoing surveillance at their

HPB 2016, 18, 375–382

Figure 1 National trends over time in number of CLP resections, and

proportion of all resections performed for CLP, in Medicare patients between the years 2001 and 2012 (CLP: cystic lesion of the pancreas).

© 2016 International Hepato-Pancreato-Biliary Association Inc. Published by Elsevier Ltd. All rights reserved.

HPB

380

Figure 2 Changes over time in the proportion of pancreatic resections

for CLP performed in Medicare patients with and without a concurrent diagnosis of malignancy (a), and Comparison of the proportion of CLP patients with and without a malignant diagnosis before and after release of international consensus guidelines in 2006 (b) (CLP: cystic lesion of the pancreas).

institution subsequently required resection, of which only 12 (1.7%) were found to have malignancy on final pathology.4 Conversely, Fritz and colleagues, who support an aggressive approach to resection, recently reported their experience with 287 consecutively resected IPMN, and found that 24.6% of lesions considered appropriate for surveillance (“Sendai negative”) had malignant features on final pathology.6 Given these conflicting single-institution reports, it is unclear if these experiences accurately represent the management of CLP more broadly. As such, this study is the first of its kind to examine the surgical management of CLP over time in a national population. According to international consensus guidelines for management of IPMN and mucinous cystic neoplasms (MCN) released in 2006 and updated in 2012, resection is recommended for all main duct IPMN and all MCN. However, a conservative approach to resection is suggested for branch duct IPMN without “high-risk stigmata,” as the authors suggest even lesions >3 cm can be observed without immediate resection.11,12 Our results show that in the years following guideline release, the proportion of CLP patients with malignancy did not change

HPB 2016, 18, 375–382

despite increasing rates of CLP resection. Given that controversy in management of asymptomatic patients with CLP centers on the assessment of malignant potential weighed against the potential risks of morbidity and mortality, these findings may be interpreted in two ways. On one hand, the increasing volume of non-malignant resections could imply that patient selection for resection of CLP remains suboptimal, as one might expect the proportion of malignancy to increase as selective guidelines are applied. Alternatively, given the purpose of guidelines is to identify lesions prior to malignant conversion, these findings may be interpreted as a successful application of CLP guidelines: appropriate resection is preventing a rise in malignancy rates. Although this study is the first of its kind to offer an evaluation of national practice, the lack of tumor-specific data unfortunately limits the assessment of guideline adherence. This study has several limitations. First, without national data regarding patients with a CLP diagnosis who are managed conservatively, we are unable account for the increasing identification of CLP lesions over time in this evaluation. Second, the use of Medicare data limits our analysis to patients aged 65 years and older, which may limit generalizability to younger patients. However, given the threshold for resection may be lower in a younger cohort, we would expect the proportion of CLP patients with malignancy to be lower than reported in this study. Third, although MEDPAR analytic files are the only currently available data source capable of addressing questions of national trends in resection, the use of administrative data can be limited by miscoding and imprecision.30–32 Though random miscoding errors should not introduce bias, the miscoding of CLP diagnoses may underestimate the true rate of resection nationally, while the restriction of data to acute care hospitalizations may similarly underestimate malignant diagnoses not captured until postoperative follow-up. However, these underestimations should not bias the evaluation of trends over time, and suggest that this analysis yields a conservative estimate of national rates of CLP resection. Moreover, previous work corroborates these trends. Klibansky and colleagues,33 using multiple regional datasets to estimate national incidence rates, reported a rising incidence of IPMN, but a relatively stable incidence of IPMNrelated carcinoma, during the years 2000–2005. And as noted above, given that administrative data lacks detailed information (tumor characteristics) necessary to identify “high-risk stigmata” of CLP, we are unable to determine if guideline release influenced resection of this subset of lesions. Conclusions Advancements in radiographic and endoscopic techniques have led to the increasing identification of CLP in the general population. As such, this study reports increasing rates of CLP resection during a recent 12-year period without an associated increase in diagnosis of malignancy, despite the publication of guidelines recommending a more selective approach to resection.

© 2016 International Hepato-Pancreato-Biliary Association Inc. Published by Elsevier Ltd. All rights reserved.

HPB

381

Additional research and prospective data are needed to better understand malignant progression in CLP, to better define criteria to identify high-risk lesions, and to accurately evaluate national practice. Future guidelines regarding CLP management should incorporate stronger evidence, and employ proactive dissemination strategies, to successfully influence practice.34

12. Tanaka M, Fernandez-del Castillo C, Adsay V, Chari S, Falconi M,

Source of funding

15. Fernandez-del Castillo CF, Thayer SP, Ferrone CR, Wargo J. (2014)

Jang JY et al. (2012) International consensus guidelines 2012 for the management of ipmn and mcn of the pancreas. Pancreatology 12: 183–197. 13. Schmidt CM. (2014) Is surgical intervention for cystic neoplasms of the pancreas being underutilized? J Gastrointest Surg 18:184–186. 14. Allen PJ. (2014) Operative resection is currently overutilized for cystic lesions of the pancreas. J Gastrointest Surg 18:182–183.

CPS is supported by a grant from the National Cancer Institute (5T32CA009672-23). JBD is supported by grant from the National Institute on Aging (5R01AG039434-03). These funding sources had no involvement in the manuscript herein.

Surgery for small and asymptomatic branch-duct ipmns. Ann Surg, 259: e47. 16. Correa-Gallego C, Brennan MF, Fong Y, Kingham TP, DeMatteo RP, D’Angelica MI et al. (2014) Liberal resection for (presumed) sendai negative branch-duct intraductal papillary mucinous neoplasms–also

Conflict of interest

not harmless. Ann Surg, 259:e45.

BNR, CPS, TLF and HN have no conflicts of interest or disclosures related to

17. Fritz S, Hackert T, Buchler MW. (2015) Pancreatic intraductal papillary

the content of this manuscript. JBD has an equity interest in ArborMetrix, Inc,

mucinous neoplasm – where is the challenge? Dig Dis 33:99–105.

which provides software and analytics for measuring hospital quality and

18. Goh BKP, Tan Y-M, Cheow P-C, Chung Y-FA, Chow PKH, Wong W-K et al. (2006) Cystic lesions of the pancreas: an appraisal of an aggres-

efficiency. The company had no role in this study.

sive resectional policy adopted at a single institution during 15 years. Am J Surg 1:148–154.

References 1. Spinelli KS, Fromwiller TE, Daniel RA, Kiely JM, Nakeeb A, Komorowski RA et al. (2004) Cystic pancreatic neoplasms: observe or

19. Morris-Stiff G, Falk GA, Chalikonda S, Walsh RM. (2013) Natural history of asymptomatic pancreatic cystic neoplasms. HPB 15:175–181. 20. Birkmeyer JD, Reames BN, McCulloch P, Carr AJ, Campbell WB,

operate. Ann Surg 239:651–657. discussion 657–659. 2. Girometti R, Intini S, Brondani G, Como G, Londero F, Bresadola F et al. (2011) Incidental pancreatic cysts on 3d turbo spin echo magnetic resonance cholangiopancreatography: prevalence and relation with clinical and imaging features. Abdom Imaging 36:196–205. 3. Kimura W, Nagai H, Kuroda A, Muto T, Esaki Y. (1995) Analysis of small cystic lesions of the pancreas. Int J Pancreatol 18:197–206. 4. Gaujoux S, Brennan MF, Gonen M, D’Angelica MI, DeMatteo R, Fong Y

Wennberg JE. (2013) Understanding of regional variation in the use of surgery. Lancet 382:1121–1129. 21. Reames BN, Sheetz KH, Waits SA, Dimick JB, Regenbogen SE. (2014) Geographic variation in use of laparoscopic colectomy for colon cancer. J Clin Oncol 32:3667–3672. 22. Elixhauser A, Steiner C, Harris DR, Coffey RM. (1998) Comorbidity measures for use with administrative data. Med Care 36:8–27.

et al. (2011) Cystic lesions of the pancreas: changes in the presentation

23. Silber JH, Romano PS, Rosen AK, Wang Y, Even-Shoshan O,

and management of 1,424 patients at a single institution over a 15-year

Volpp KG. (2007) Failure-to-rescue: comparing definitions to measure quality of care. Med Care 45:918–925.

time period. J Am Coll Surg 212:590–600. 5. Sawai Y, Yamao K, Bhatia V, Chiba T, Mizuno N, Sawaki A et al. (2010)

24. Ghaferi AA, Osborne NH, Birkmeyer JD, Dimick JB. (2010) Hospital

Development of pancreatic cancers during long-term follow-up of side-

characteristics associated with failure to rescue from complications

branch intraductal papillary mucinous neoplasms. Endoscopy 42: 6. Fritz S, Klauss M, Bergmann F, Hackert T, Hartwig W, Strobel O et al. (2012) Small (sendai negative) branch-duct ipmns: not harmless. Ann 7. Siech M, Tripp K, Schmidt-Rohlfing B, Mattfeldt T, Widmaier U, Gansauge F et al. (1998) Cystic tumours of the pancreas: diagnostic pathologic

observations

and

surgical

consequences.

N Engl J Med 346:1128–1137. volume and operative mortality in the modern era. Ann Surg 260: 244–251. 27. Iezzoni LI, Daley J, Heeren T, Foley SM, Fisher ES, Duncan C et al. (1994) Identifying complications of care using administrative data. Med

Langenbecks Arch Surg 383:56–61. 8. Horvath KD, Chabot JA. (1999) An aggressive resectional approach to cystic neoplasms of the pancreas. Am J Surg 178:269–274. 9. Walsh RM, Vogt DP, Henderson JM, Zuccaro G, Vargo J, Dumot J et al. (2005) Natural history of indeterminate pancreatic cysts. Surgery 138:

Care 32:700–715. 28. Weingart SN, Iezzoni LI, Davis RB, Palmer RH, Cahalane M, Hamel MB et al. (2000) Use of administrative data to find substandard care: validation of the complications screening program. Med Care 38:796–806. 29. Reames BN, Birkmeyer NJ, Dimick JB, Ghaferi AA. (2014) Socioeco-

665–670. discussion 670–661. 10. Allen PJ, D’Angelica M, Gonen M, Jaques DP, Coit DG, Jarnagin WR et al. (2006) A selective approach to the resection of cystic lesions of the pancreas. Trans Meet Am Surg Assoc 124:237–247. 11. Tanaka M, Chari S, Adsay V, Fernandez-del Castillo C, Falconi M, Shimizu M et al. (2006) International consensus guidelines for management of intraductal papillary mucinous neoplasms and mucinous cystic neoplasms of the pancreas. Pancreatology 6:17–32.

HPB 2016, 18, 375–382

et al. (2002) Hospital volume and surgical mortality in the united states. 26. Reames BN, Ghaferi AA, Birkmeyer JD, Dimick JB. (2014) Hospital

Surg 256:313–320.

accuracy,

after pancreatectomy. J Am Coll Surg 211:325–330. 25. Birkmeyer JD, Siewers AE, Finlayson EV, Stukel TA, Lucas FL, Batista I

1077–1084.

nomic disparities in mortality after cancer surgery: failure to rescue. JAMA Surg 149:475–481. 30. Jencks SF, Williams DK, Kay TL. (1988) Assessing hospital-associated deaths from discharge data. The role of length of stay and comorbidities. JAMA 260:2240–2246. 31. Iezzoni LI, Foley SM, Daley J, Hughes J, Fisher ES, Heeren T. (1992) Comorbidities, complications, and coding bias. Does the number of

© 2016 International Hepato-Pancreato-Biliary Association Inc. Published by Elsevier Ltd. All rights reserved.

HPB

382

diagnosis codes matter in predicting in-hospital mortality? JAMA 267: 2197–2203.

33. Klibansky DA, Reid–Lombardo KM, Gordon SR, Gardner TB. (2012) The clinical relevance of the increasing incidence of intraductal papillary

32. Fisher ES, Whaley FS, Krushat WM, Malenka DJ, Fleming C, Baron JA

mucinous neoplasm. Clin Gastroenterol Hepatol 10:555–558.

et al. (1992) The accuracy of medicare’s hospital claims data: progress

34. Reames BN, Shubeck SP, Birkmeyer JD. (2014) Strategies for reducing

has been made, but problems remain. Am J Public Health 82:

regional variation in the use of surgery: a systematic review. Ann Surg

243–248.

259:616–627.

HPB 2016, 18, 375–382

© 2016 International Hepato-Pancreato-Biliary Association Inc. Published by Elsevier Ltd. All rights reserved.