Advertisement

Influence of facility type on survival outcomes after pancreatectomy for pancreatic adenocarcinoma

  • Quyen D. Chu
    Affiliations
    Department of Surgery, Louisiana State University Health Sciences Center-Shreveport, Shreveport, LA, USA

    The Feist-Weiller Cancer Center, Louisiana State University Health Sciences Center-Shreveport, Shreveport, LA, USA
    Search for articles by this author
  • Meijiao Zhou
    Affiliations
    Louisiana Tumor Registry & Epidemiology, Louisiana State University Health Sciences Center-Shreveport, Shreveport, LA, USA

    School of Public Health, Louisiana State University Health Sciences Center, New Orleans, LA, USA
    Search for articles by this author
  • Prakash Peddi
    Correspondence
    Correspondence Prakash Peddi, Division of Hematology and Oncology, Louisiana State University Health Sciences Center-Shreveport, Feist-Weiller Cancer Center, Shreveport, LA 71130, USA.
    Affiliations
    Department of Medicine, Louisiana State University Health Sciences Center-Shreveport, Shreveport, LA, USA

    The Feist-Weiller Cancer Center, Louisiana State University Health Sciences Center-Shreveport, Shreveport, LA, USA
    Search for articles by this author
  • Kaelen L. Medeiros
    Affiliations
    Louisiana Tumor Registry & Epidemiology, Louisiana State University Health Sciences Center-Shreveport, Shreveport, LA, USA

    School of Public Health, Louisiana State University Health Sciences Center, New Orleans, LA, USA
    Search for articles by this author
  • Gazi B. Zibari
    Affiliations
    Department of Surgery, Louisiana State University Health Sciences Center-Shreveport, Shreveport, LA, USA

    John C McDonald Regional Transplant Center, Willis Knighton Health System, Shreveport, LA, USA
    Search for articles by this author
  • Hosein Shokouh-Amiri
    Affiliations
    Department of Surgery, Louisiana State University Health Sciences Center-Shreveport, Shreveport, LA, USA

    John C McDonald Regional Transplant Center, Willis Knighton Health System, Shreveport, LA, USA
    Search for articles by this author
  • Xiao-Cheng Wu
    Affiliations
    Louisiana Tumor Registry & Epidemiology, Louisiana State University Health Sciences Center-Shreveport, Shreveport, LA, USA

    School of Public Health, Louisiana State University Health Sciences Center, New Orleans, LA, USA
    Search for articles by this author
Open ArchivePublished:September 26, 2017DOI:https://doi.org/10.1016/j.hpb.2017.04.017

      Abstract

      Introduction

      Although a volume-outcome relationship has been well established for pancreatectomy, little is known about differences in mortality by facility type. The objective of this study is to evaluate the impact of facility type on short-term and long-term survival outcomes for patients with pancreatic adenocarcinoma who underwent pancreatectomy and identify determinants of overall survival (OS).

      Methods

      A cohort of 33,382 patients with Stage I–III pancreatic adenocarcinoma diagnosed between 1998 and 2011 were evaluated from the National Cancer Data Base. Clinicopathological, sociodemographic and treatment variables were compared among three facility types where patients received resection: (i) community cancer program (CCP), (ii) comprehensive community cancer program (CCCP), and (iii) academic research program (ARP). 5-year OS was calculated using the Kaplan–Meier method.

      Results

      Despite ARP having significantly higher percentage of poorly differentiated tumors, higher T-stage tumors, more positive lymph nodes, and greater circle distance compared to the other facilities, it had the highest 5-yr OS. The 5-yr OS for CCP, CCCP, and ARP was 11.2%, 13.2%, and 16.6%, respectively (P < 0.0001) and the median survival time (months) was 12.4, 15.6 and 19.1, respectively.

      Conclusion

      Patients receiving pancreatic resection at an ARP yielded a higher 5-year OS compared to CCP or CCCP.

      Introduction

      Adenocarcinoma of the pancreas is a formidable disease. It is the fourth leading cause of cancer-related deaths in the U.S.
      • Siegel R.L.
      • Miller K.D.
      • Jemal A.
      Cancer statistics, 2016.
      It is estimated that in 2016, 53,070 new cases of pancreatic cancer will be diagnosed, of which 41,780 deaths will occur in the United States.
      • Siegel R.L.
      • Miller K.D.
      • Jemal A.
      Cancer statistics, 2016.
      Surgery remains the only treatment for potential cure and long-term survival depends on several factors.
      Multiple studies have demonstrated that patients who underwent complex operations such as pancreatectomy at high-volume centers experience lower mortality.
      • Gordon T.A.
      • Bowman H.M.
      • Tielsch J.M.
      • Bass E.B.
      • Burleyson G.P.
      • Cameron J.L.
      Statewide regionalization of pancreaticoduodenectomy and its effect on in-hospital mortality.
      • Birkmeyer J.D.
      • Finlayson E.V.
      • Birkmeyer C.M.
      Volume standards for high-risk surgical procedures: potential benefits of the Leapfrog initiative.
      • Schmidt C.M.
      • Turrini O.
      • Parikh P.
      • House M.G.
      • Zyromski N.J.
      • Nakeeb A.
      • et al.
      Effect of hospital volume, surgeon experience, and surgeon volume on patient outcomes after pancreaticoduodenectomy: a single-institution experience.
      The lower mortality at these high-volume centers may be related to their ability to better rescue patients.
      • Ghaferi A.A.
      • Birkmeyer J.D.
      • Dimick J.B.
      Hospital volume and failure to rescue with high-risk surgery.
      • Ward M.M.
      • Jaana M.
      • Wakefield D.S.
      • Ohsfeldt R.L.
      • Schneider J.E.
      • Miller T.
      • et al.
      What would be the effect of referral to high-volume hospitals in a largely rural state?.
      • Khuri S.F.
      • Henderson W.G.
      The case against volume as a measure of quality of surgical care.
      Despite this knowledge, no studies have evaluated how the volume-outcome relationship is affected by facility type. In other words, are there specific components of care, besides volume, within a specific facility that contribute toward better outcomes than other facilities? In addition, are there differences in referral patterns, socioeconomic factors, or access to care that might give specific centers an advantage over another center? Finally, do particular facilities have better quality of care (i.e. less likely to have positive surgical margins following a pancreatectomy) than others that translates to improved outcomes? Of note, although data on the volume-outcome relationship exist, most if not all are based on short-term perioperative mortality (i.e. 30-days or 90-days mortality) rather than on long-term outcomes such as overall survival (OS). The objective of this study is to evaluate the impact of facility type on short-term and long-term survival outcomes for patients with pancreatic adenocarcinoma who underwent pancreatectomy and evaluate the impact of volume on long-term surgical outcomes.

      Methods

      Data source

      The nationally recognized National Cancer Data Base (NCDB) is a joint project of the Commission on Cancer (CoC) of the American College of Surgeons and the American Cancer Society. More than 1500 CoC-accredited facilities contribute clinical information to the database. Approximately 70% of newly diagnosed cancer cases in the U.S and 30 million historical records are captured in the database. The data sets in the PUF were de-identified and were in compliant with the privacy requirements of the Health Insurance Portability and Accountability Act (HIPAA). The study was exempted from Institutional Review Board (IRB) approval by the Louisiana State University Health Sciences Center-Shreveport.

      Study population

      A cohort of 33,382 cases of patients with stage I–III pancreatic adenocarcinoma (ICD-0-3; C25.0-C25.9) who were diagnosed in 1998–2011 and received surgery (the site-specific codes ranged from 25 to 90 in the SEER 2003 and Site-Specific Surgery of Primary Site Codes), with known vital status in the NCDB was analyzed to determine significant factors associated with overall survival (OS). Patients were staged based on the 7th edition of the AJCC/TNM staging system.
      • Edge S.B.
      • Compton C.C.
      The American Joint Committee on Cancer: the 7th edition of the AJCC cancer staging manual and the future of TNM.
      According to the NCDB's PUF dictionary, facility type was classified as (i) community cancer program (CCP), (ii) comprehensive community cancer program (CCCP), (iii) academic research program (ARP), and (iv) others.

      Categories of accreditation. http://www.facs.org/cancer/coc/categories.html. Accessed November 9, 2014.

      Community cancer programs are those that treat between 100 and 500 newly diagnosed cancer patients each year, and patients may be referred to another facility for part of their diagnosis or treatment.

      Categories of accreditation. http://www.facs.org/cancer/coc/categories.html. Accessed November 9, 2014.

      Comprehensive community cancer programs are those that treat more than 500 newly diagnosed cancer patients each year.

      Categories of accreditation. http://www.facs.org/cancer/coc/categories.html. Accessed November 9, 2014.

      Academic and research institutions are those that treat more than 500 newly diagnosed cancer patients each year while offering postgraduate medical education programs.

      Categories of accreditation. http://www.facs.org/cancer/coc/categories.html. Accessed November 9, 2014.

      The treating facility was the hospital where the patients received their pancreatic cancer surgery.
      Facility location was categorized into regions within the United States based on the U. S. Census Regions and Divisions: (i) New England, (ii) Mid-Atlantic, (iii) South Atlantic, (iv) East North Central, (v) East South Central, (vi) West North Central, (vii) West South Central, (viii) Mountain, and (ix) Pacific.
      Race/ethnicity was categorized as White, Black, American Indian/Aleutian or Eskimo, Asian or Pacific Islander, and Others. Insurance status was classified as (i) uninsured, (ii) private insurance/managed care, (iii) Medicaid, (iv) Medicare, and (v) other governmental insurance. Patients with both private insurance and Medicare were grouped in the private insurance category. Median household income level was classified as (i) <$38,000, (ii) $38,000–$47,999, (iii) $48,000–$62,999, and (iv) ≥$63,000. Education level was classified into percentage of adults (age ≥ 25 years) who did not graduate from high school in the area based on the 2012 American Community Survey data: (i) ≥21%, (ii) 13–20.9%, (iii) 7–12.9%, and (iv) <7%. Charlson/Deyo comorbidity score has been recorded since 2003 and was reported as 0, 1 or 2.
      • Charlson M.E.
      • Pompei P.
      • Ales K.L.
      • MacKenzie C.R.
      A new method of classifying prognostic comorbidity in longitudinal studies: development and validation.
      • Deyo R.A.
      • Cherkin D.C.
      • Ciol M.A.
      Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases.
      Population density was classified as (i) metropolitan with ≥1 million, (ii) metropolitan with 250,000 to 1 million, (iii) metropolitan with <250,000, (iv) urban with ≥20,000 and adjacent to a metropolitan area, (v) urban with ≥20,000 but not adjacent to a metropolitan area, (vi) urban with <20,000 and adjacent to a metropolitan area, (vii) urban with <20,000 but not adjacent to a metropolitan area, (viii) rural with <2500 and adjacent to a metropolitan area, and (ix) rural with <2500 but not adjacent to a metropolitan area. Great circle distance is measured in miles and defined as being between the patient's residence at diagnosis and the hospital that reported the case.
      Annual Hospital volume (AHV) was defined as the average number of pancreatic surgeries per year, which was based on the cases each hospital reported to the NCDB. The mean AHV of pancreatectomies was calculated for currently accredited CoC hospitals during the study period (1998–2011), and grouped as 4 categories: <5 patients, 5–<10 patients, 10–<20 patients and ≥20 patients.

      Statistical analysis

      The nonparsimonious approach using variables such as age, race/ethnicity, etc. was used to construct models. Descriptive statistics for the different variable were reported. Univariable analysis of each variable was performed using chi-square test for categorical data and ANOVA for numerical data. The Kaplan–Meier method was used for survival analysis. Univariable Cox proportional hazard regression was used to identify factors significantly associated with the risk of deaths for all causes. Factors that were statistically significant in univariable analysis were included in the multivariable model. The multivariable Cox proportional hazards regression analysis was used to determine independent significant factors associated with the risk of death for all causes, and the hazard ratios (HR) and confidence intervals (CI) were calculated. A p-value ≤0.05 was considered statistically significant. Multicollinearity was detected using Variance Inflation Factors (VIF). VIF greater than 4 indicates multicollinearity. All statistical analyses were performed using SAS Version 9.4 statistical software, (SAS Institute Inc., Cary, NC, U.S.A., 2013).

      Results

      The median follow-up was 16.3 months. Summary statistics on patient characteristics and treatment outcomes are shown in Table 1. The median age of the entire cohort was 67 years (range, 18–90). The proportion of males and females was similar. Most patients were White (86%) and managed in an academic research program (60%). South Atlantic region had the highest percentage of patients (21.9%) while the New England region had the lowest (4.2%). Approximately 16% of centers performed ≥20 pancreatectomy annually. Most patients had health insurance (97.4%), resided in metropolitan communities (79%), had a median household income of ≥$63,000 (31.3%), resided in a high education area, had to travel less than 50 miles to their nearest hospital that reported the case (73.5%), and had zero comorbidity score (49.9%). The majority of patients underwent a pancreaticoduodenectomy (60.4%), had negative margins (72.9%), moderately differentiated tumors (48.9%), positive lymph nodes (61.6%), less than 15 lymph nodes retrieved (62.9%), T3/4 disease (71.0%), N1 disease (61.5%), and stage II disease (67.1%), The majority received chemotherapy (55.3%), while most did not receive radiation therapy (60.9%).
      Table 1Summary statistics for stage I–III pancreatic cancer
      VariableFrequency (percent)
      Total patients33,382
      SexMale16,750 (50.1)
      Female16,632 (49.8)
      Race/ethnicityWhite28,777 (86.2)
      Black3246 (9.7)
      American Indian, Aleutian, or Eskimo67 (0.2)
      Asian/Pacific Islander711 (2.1)
      Other, unknown or missing581 (1.7)
      Facility typeCommunity cancer program1166 (3.4)
      Comprehensive community cancer center12,214 (36.5)
      Academic research program20,002 (59.9)
      Facility locationNew England1412 (4.2)
      Mid Atlantic5536 (16.5)
      South Atlantic7320 (21.9)
      East North Central5970 (17.8)
      East South Central2303 (6.9)
      West North Central2837 (8.5)
      West South Central3136 (9.3)
      Mountain1440 (4.3)
      Pacific3428 (10.2)
      Annual hospital volume<5 patients14,512 (43.4)
      5–9 patients5126 (15.3)
      10–19 patients8161 (24.4)
      ≥20 patients5420 (16.2)
      Missing, not available, or unknown163 (0.4)
      Primary payorUninsured858 (2.5)
      Private insurance12,771 (38.2)
      Medicaid1321 (3.9)
      Medicare17,077 (51.1)
      Other government259 (0.7)
      Missing, not available, or unknown1096 (3.2)
      Income<$38,0005787 (17.3)
      $38,000–47,9997445 (22.3)
      $48,000–62,9998541 (25.5)
      $63,000+10,454 (31.3)
      Patients had missing information1155 (3.4)
      Education≥21%5230 (15.6)
      13–20.9%8302 (24.8)
      7–12.9%10,456 (31.3)
      <7%8249 (24.7)
      Patients had missing information1145 (3.4)
      Surgical resectionPartial pancreatectomy8335 (24.9)
      Pancreaticoduodenectomy20,163 (60.4)
      Total pancreatectomy3840 (11.5)
      Not specified1044 (3.1)
      Surgical marginsNo residual tumor24,329 (72.8)
      Residual tumor7711 (23.1)
      Patients had missing information1342 (4.0)
      Pathologic stage groupI5216 (15.6)
      II22,381 (67.0)
      III5785 (17.3)
      ChemotherapyNone13,196 (39.5)
      Yes18,473 (55.3)
      Patients had missing information1713 (5.1)
      Radiation therapyNone20,339 (60.9)
      Yes12,795 (38.3)
      Patients had missing information248 (0.7)
      Comorbidities/deyo score016,660 (49.9)
      16940 (20.7)
      21860 (5.5)
      Urban/rural location (2013 in dataset)Patients had missing information7922 (23.7)
      Metro ≥1 million16,921 (50.6)
      Metro 250 k to 1 million6472 (19.3)
      Urban <250k3004 (9.0)
      Urban ≥20k adjacent metro1524 (4.5)
      Urban ≥20k not adjacent metro407 (1.2)
      Urban <20k adjacent metro1804 (5.4)
      Urban <20k not adjacent metro948 (2.8)
      Rural <2500 adjacent metro265 (0.7)
      Rural <2500 not adjacent metro342 (1.0)
      Patients had missing information1695 (5.0)
      Great circle distance<50 miles24,523 (73.4)
      ≥50 miles7740 (23.1)
      Patients had missing information1119 (3.3)
      Table 2 delineates the association of facility type with socio-demographics and clinical factors. Note that ARP represents most the cases (59.9%) while CCP represent the minority of the cases (3.5%). Most patients in all three facility types had Medicare or private insurance while ARP had slightly more percentage of patients belonging to the highest income bracket, highest education level, and higher percentage with no residual tumor. Compared to CCCP and CCP, ARP had higher percentage of patients with poorly differentiated tumors, nodal positivity, and T3/4 disease. Finally, patients in the ARP group were more likely to receive chemotherapy (60% versus 57% for CCP and 50% for CCP), although they had a slightly lower use of radiation therapy (37.6%) compared to CCCP (40.2%) and CCP (38.3%) (P < 0.01). The 30-day and 90-day mortality rates were significantly lower at ARP compared to CCCP and CCP; the 30-day mortality rates for ARP, CCCP, and CCP were 3.5%, 6.0%, and 9.9%, respectively (P < 0.01); the 90-day mortality rates for ARP, CCCP, and CCP were 7.1%, 10.7%, and 16.9%, respectively (P < 0.01).
      Table 2Demographics of stage I–III pancreatic cancer population by facility type
      Variable, N = 33,382Community cancer program N (%)Comprehensive community cancer program N (%)Academic/research program N (%)P-value
      Total patients1166 (3.4)12,214 (36.5)20,002 (59.9)
      SexMale580 (49.7)6011 (49.2)10,159 (50.7)0.02
      Female586 (50.2)6203 (50.7)9843 (49.2)
      Race/ethnicityWhite937 (80.9)10,743 (88.9)17,097 (87.4)<0.01
      Black145 (12.5)1079 (8.9)2022 (10.3)
      American Indian, Aleutian, or Eskimo2 (0.1)27 (0.2)38 (0.1)
      Asian/Pacific Islander73 (6.3)235 (1.9)403 (2.0)
      Facility locationNew England95 (8.1)268 (2.1)1049 (5.2)<0.01
      Mid Atlantic136 (11.6)1129 (9.2)4271 (21.3)
      South Atlantic183 (15.6)3051 (24.9)4086 (20.4)
      East North Central309 (26.5)1939 (15.8)3722 (18.6)
      East South Central61 (5.2)1229 (10.0)1013 (5.0)
      West North Central74 (6.3)854 (6.9)1909 (9.5)
      West South Central107 (9.1)1194 (9.7)1835 (9.1)
      Mountain13 (1.1)750 (6.1)677 (3.3)
      Pacific188 (16.1)1800 (14.7)1440 (7.2)
      Annual hospital volume<5 patients1161 (100)8968 (74.0)4383 (21.9)<0.01
      5–9 patients01734 (14.3)3392 (17.0)
      10–19 patients01411 (11.6)6750 (33.8)
      ≥20 patients005420 (27.1)
      Primary payorUninsured45 (3.9)284 (2.3)529 (2.7)<0.01
      Private insurance397 (34.8)4624 (38.4)7750 (40.5)
      Medicaid54 (4.7)428 (3.5)839 (4.3)
      Medicare640 (56.1)6603 (54.9)9834 (51.4)
      Other government4 (0.3)85 (0.7)170 (0.8)
      Income<$38,000205 (18.2)2150 (18.1)3432 (17.8)<0.01
      $38,000–47,999334 (29.7)2834 (23.9)4277 (22.1)
      $48,000–62,999289 (25.7)3304 (27.9)4948 (25.6)
      $63,000+294 (26.2)3537 (29.9)6623 (34.3)
      Education≥21%220 (19.6)2000 (16.9)3010 (15.6)<0.01
      13–20.9%344 (30.6)3045 (25.7)4913 (25.4)
      7–12.9%393 (35.0)3898 (32.9)6165 (31.9)
      <7%165 (14.7)2886 (24.4)5198 (26.9)
      Surgical resectionPartial pancreatectomy327 (29.3)2956 (24.9)5052 (26.1)<0.01
      Pancreaticoduodenectomy652 (58.4)7467 (62.9)12,044 (62.2)
      Total pancreatectomy136 (12.2)1444 (12.1)2260 (11.6)
      Surgical marginsNo residual tumor805 (72.1)8670 (73.6)14,854 (77.5)<0.01
      Residual tumor311 (27.8)3107 (26.3)4293 (22.4)
      Lymph nodes positiveNo435 (39.22)4411 (37.31)6977 (35.85)<0.01
      Yes674 (60.78)7411 (62.69)12,482 (64.15)
      Lymph nodes retrieved<15 LN831 (75.6)8455 (71.4)11,695 (59.5)<0.01
      ≥15 LN267 (24.3)3383 (28.5)7938 (40.4)
      GradeWell differentiated170 (15.32)1324 (11.41)1714 (9.21)<0.01
      Moderately differentiated583 (52.52)6092 (52.49)9650 (51.87)
      Poorly differentiated350 (31.53)4099 (35.32)7006 (37.66)
      Undifferentiated, anaplastic7 (0.63)90 (0.78)233 (1.25)
      Pathologic stage groupI221 (18.9)2034 (16.6)2961 (14.8)<0.01
      II638 (54.7)7823 (64.0)13,920 (69.5)
      III307 (26.3)2357 (19.3)3121 (15.6)
      ChemotherapyNone558 (50.0)5046 (43.0)7592 (40.3)<0.01
      Yes558 (50.0)6680 (56.9)11,235 (59.6)
      Radiation therapyNone714 (61.6)7244 (59.7)12,381 (62.3)<0.01
      Yes444 (38.3)4878 (40.2)7473 (37.6)
      30-day mortality115 (9.9%)726 (6.0%)702 (3.5%)<0.01
      90-day mortality195 (16.9%)1290 (10.7%)1390 (7.1%)<0.01
      There was a significant difference in the volumes across facility types; 61% of ARP had annual pancreatic surgery for ≥10 patients, 12% of CCCP had ≥10 patients, and none of CCP had ≥10 patients (P < 0.01). The 30-day and 90-day mortality rates were also significantly lower at hospitals that performed high volume of pancreatectomy; the 30-day mortality rate for hospitals that performed ≥20 pancreatectomies annually was 2% compared to 6.7% for hospitals that performed <5 pancreatectomies annually (P < 0.01); similarly, the 90-day mortality rate for hospitals that performed ≥20 pancreatectomies annually was 5.1% compared to 11.6% for hospitals that performed <5 pancreatectomies annually (P < 0.01) (Table 2).
      Fig. 1 shows the Kaplan–Meier OS curve for the three facility types. The 5-year OS rate was 16.6% for ARP, 13.2% for CCCP, and 11.2% for CCP (P < 0.01). The median survival was 19.1 mos for ARP, 15.6 mos for CCCP, and 12.4 mos for CCP. Table 3 shows univariable factors associated with OS and Table 4 shows multivariable factors associated with OS. Independent predictors associated with high HR include: facility type, older age, male gender, facility location, population density, low hospital volume, Medicare recipient, low income, low education level, high comorbidity index, residual margins, histology other than well-differentiated, nodal involvement, <15 lymph nodes retrieved, stage III, and lack of receipt of chemotherapy or radiation therapy. Compared to hospitals that performed ≥20 pancreatectomy per year, those that performed <5 per year had a hazard ratio of 1.23 (p < 0.01; 95% CI = 1.17–1.30). There is no multicollinearity issue based on the VIF (≤2). Of note, the 30 day and 90-day mortality rates were significantly different among the different facility types. For the 30-day mortality rate, the hazard ratio for CCP vs. ARP was 1.38 (p = 0.01; 95% CI = 1.08–1.77), and the hazard ratio for CCCP vs. ARP was 1.12 (p = 0.11; 95% CI = 0.98–1.29). For the 90-day mortality rate, the hazard ratio for CCP vs. ARP was 1.39 (p < 0.01; 95% CI = 1.13–1.69), and the hazard ratio for CCCP vs. ARP was 1.13 (p = 0.03; 95% CI = 1.02–1.26). Fig. 2 shows differences of HR based on the 9 geographic locations; compared to Pacific region, the HR for mortality was higher in the West South Central region (HR = 1.08), East North Central region (HR = 1.15), East South Central region (HR = 1.11), and South Atlantic Region (HR = 1.10). There were no differences in HR among the Pacific, Mountain, West North Central, Mid-Atlantic, and New England regions.
      Figure 1
      Figure 1Kaplan–Meier overall survival curve of the three facility types in patients undergoing pancreatectomy for pancreatic cancer. The 5-yr OS for ARP, CCCP and CCP was 16.6%, 13.2% and 11.2%. The median survival time was 19.1, 15.6 and 12.4 months
      Table 3Univariable analysis of factors associated with overall survival in stage I–III pancreatic tumor
      VariableHazard ratio95 CI (P-value)
      Age1.0141.013–1.015 (<0.01)
      Facility typeAcademic research program1.00
      Community cancer program1.441.35–1.53 (<0.01)
      Comprehensive community cancer program1.201.17–1.23 (<0.01)
      SexFemale1.00
      Male1.061.04–1.09 (<0.01)
      Race/ethnicityWhite1.00
      Black1.000.96–1.04 (0.93)
      American Indian, Aleutian, or Eskimo0.820.62–1.08 (0.17)
      Asian/Pacific Islander0.880.80–0.96 (<0.01)
      Facility locationPacific1.00
      New England0.990.93–1.06 (0.83)
      Mid Atlantic0.960.91–1.01 (0.10)
      South Atlantic1.071.03–1.12 (<0.01)
      East North Central1.071.02–1.12 (0.008)
      East South Central1.131.07–1.20 (<0.01)
      West North Central0.970.92–1.03 (0.35)
      West South Central1.091.03–1.15 (<0.01)
      Mountain1.181.10–1.26 (<0.01)
      Annual hospital volume≥20 patients1.00
      10–19 patients1.040.997–1.08 (0.07)
      5–9 patients1.061.02–1.11 (0.009)
      <5 patients1.291.24–1.34 (<0.01)
      Primary payorPrivate insurance1.00
      Uninsured1.141.05–1.23 (<0.01)
      Medicaid1.171.10–1.25 (<0.01)
      Medicare1.281.25–1.31 (<0.01)
      Other government0.990.86–1.14 (0.89)
      Income$63,000+1.00
      $48,000–62,9991.101.06–1.13 (<0.01)
      $38,000–47,9991.171.13–1.21 (<0.01)
      <$38,0001.231.19–1.28 (<0.01)
      Education<7%1.00
      ≥21%1.191.15–1.24 (<0.01)
      13–20.9%1.171.13–1.21 (<0.01)
      7–12.9%1.081.05–1.12 (<0.01)
      Comorbidities/dey score01.00
      11.101.06–1.13 (<0.01)
      21.301.23–1.37 (<0.01)
      Missing1.211.17–1.24 (<0.01)
      Lymph nodes positiveNo (reference)1.00
      Yes1.531.49–1.57 (<0.01)
      Lymph nodes retrieved≥15 LN1.00
      <15 LN1.101.07–1.13 (<0.01)
      T-StageT1/21.00
      T3/41.341.30–1.37 (<0.01)
      N-StageN01.00
      N11.531.49–1.57 (<0.01)
      Pathologic stage groupI1.00
      II1.551.50–1.61 (<0.01)
      III2.031.95–2.12 (<0.01)
      GradeWell differentiated (reference)1.0
      Moderately differentiated1.381.32–1.44 (<0.01)
      Poorly differentiated1.781.70–1.86 (<0.01)
      Undifferentiated, anaplastic1.571.38–1.79 (<0.01)
      ChemotherapyYes1.00
      None1.561.52–1.60 (<0.01)
      Radiation therapyYes1.00
      None1.341.31–1.37 (<0.01)
      Urban/rural locationMetro ≥1 million (reference)1.00
      Metro 250 k to 1 million1.081.05–1.12 (<0.01)
      Metro <250k1.081.03–1.13 (<0.01)
      Urban ≥20k adjacent metro1.101.03–1.16 (<0.01)
      Urban ≥20k not adjacent metro1.090.98–1.22 (0.11)
      Urban <20k adjacent metro1.101.04–1.16 (<0.01)
      Urban <20k not adjacent metro1.181.10–1.27 (<0.01)
      Rural <2500 adjacent metro1.171.03–1.33 (0.02)
      Rural <2500 not adjacent metro1.141.02–1.29 (0.03)
      Great circle distance≥50 miles (reference)1.00
      <50 miles1.041.01–1.07 (<0.01)
      Surgical resectionPartial pancreatectomy (reference)1.00
      Pancreaticoduodenectomy1.131.10–1.17 (<0.01)
      Total pancreatectomy1.141.09–1.19 (<0.01)
      Table 4Multivariable analysis of factors associated with overall survival in stage I–III pancreatic cancer
      VariableHazard ratio95% CI (P-value)
      Facility typeAcademic research program (reference)1.00
      Community cancer program1.221.12–1.32 (<0.01)
      Comprehensive community cancer program1.051.01–1.09 (<0.01)
      SexFemale (reference)1.00
      Male1.081.05–1.11 (<0.01)
      Race/ethnicityWhite (reference)1.00
      Black0.970.93–1.03 (0.31)
      American Indian, Aleutian, or Eskimo0.860.62–1.19 (0.37)
      Asian/Pacific Islander0.920.83–1.02 (0.13)
      Facility locationPacific (reference)1.00
      New England1.020.94–1.11 (0.62)
      Mid Atlantic1.060.99–1.12 (0.08)
      South Atlantic1.101.04–1.16 (<0.01)
      East North Central1.151.08–1.21 (<0.01)
      East South Central1.111.03–1.19 (0.005)
      West North Central1.050.98–1.13 (0.15)
      West South Central1.081.01–1.15 (0.03)
      Mountain1.070.99–1.16 (0.11)
      Hospital volume≥20 patients (reference)1.00
      <5 patients1.231.17–1.30 (<0.01)
      5–<10 patients1.061.01–1.12 (0.03)
      10–<20 patients1.071.02–1.13 (<0.01)
      Primary payorPrivate insurance (reference)1.00
      Uninsured1.030.94–1.13 (0.53)
      Medicaid1.161.08–1.25 (0.01)
      Medicare1.061.02–1.10 (<0.01)
      Other government0.990.84–1.16 (0.89)
      Income$63,000 + (reference)1.00
      <$38,0001.111.04–1.18 (0.01)
      $38,000–47,9991.040.99–1.09 (0.17)
      $48,000–62,9991.020.98–1.07 (0.30)
      Education<7% (reference)1.00
      ≥21%1.061.00–1.13 (0.06)
      13–20.9%1.091.03–1.14 (0.01)
      7–12.9%1.041.00–1.09 (0.05)
      Surgical resectionPartial pancreatectomy (reference)1.00
      Pancreaticoduodenectomy1.091.05–1.12 (<0.01)
      Total pancreatectomy1.131.08–1.19 (<0.01)
      Surgical marginsNo residual tumor (reference)1.00
      Residual tumor1.541.49–1.59 (<0.01)
      GradeWell differentiated (reference)1.0
      Moderately differentiated1.371.30–1.44 (<0.01)
      Poorly differentiated1.791.70–1.89 (<0.01)
      Undifferentiated, anaplastic1.881.61–2.20 (<0.01)
      Lymph nodes positiveNo (reference)1.00
      Yes1.361.31–1.41 (<0.01)
      Lymph nodes retrieved≥15 LN (reference)1.00
      <15 LN1.091.05–1.12 (<0.01)
      Pathologic stage groupI (reference)1.00
      II1.341.27–1.41 (<0.01)
      III1.491.39–1.60 (<0.01)
      ChemotherapyYes (reference)1.00
      None1.591.53–1.65 (<0.01)
      Radiation therapyYes (reference)1.00
      None1.141.10–1.18 (<0.01)
      Comorbidities/deyo score0 (reference)1.00
      11.061.02–1.10 (0.004)
      21.251.18–1.33 (<0.0001)
      Missing1.071.02–1.12 (0.01)
      Figure 2
      Figure 2Hazard Ratio for survival based on geographical location

      Discussion

      There are multiple factors that determine outcomes following complex operations such as pancreatectomy. One such major determinant is hospital case-volume.
      • Gordon T.A.
      • Bowman H.M.
      • Tielsch J.M.
      • Bass E.B.
      • Burleyson G.P.
      • Cameron J.L.
      Statewide regionalization of pancreaticoduodenectomy and its effect on in-hospital mortality.
      • Birkmeyer J.D.
      • Finlayson E.V.
      • Birkmeyer C.M.
      Volume standards for high-risk surgical procedures: potential benefits of the Leapfrog initiative.
      • Schmidt C.M.
      • Turrini O.
      • Parikh P.
      • House M.G.
      • Zyromski N.J.
      • Nakeeb A.
      • et al.
      Effect of hospital volume, surgeon experience, and surgeon volume on patient outcomes after pancreaticoduodenectomy: a single-institution experience.
      Many investigators have reported that patients who underwent pancreatectomy at high-volume centers experience higher survival rate.
      • Gordon T.A.
      • Bowman H.M.
      • Tielsch J.M.
      • Bass E.B.
      • Burleyson G.P.
      • Cameron J.L.
      Statewide regionalization of pancreaticoduodenectomy and its effect on in-hospital mortality.
      • Birkmeyer J.D.
      • Finlayson E.V.
      • Birkmeyer C.M.
      Volume standards for high-risk surgical procedures: potential benefits of the Leapfrog initiative.
      • Schmidt C.M.
      • Turrini O.
      • Parikh P.
      • House M.G.
      • Zyromski N.J.
      • Nakeeb A.
      • et al.
      Effect of hospital volume, surgeon experience, and surgeon volume on patient outcomes after pancreaticoduodenectomy: a single-institution experience.
      Birkmeyer et al. reported an inverse relationship between hospital volume and operative mortality, which led the Leapfrog group, a large coalition of private and public health care groups to advocate for the need for high-risk surgical procedures to be performed at high-volume centers.
      • Birkmeyer J.D.
      • Finlayson E.V.
      • Birkmeyer C.M.
      Volume standards for high-risk surgical procedures: potential benefits of the Leapfrog initiative.
      Consistent with these studies, our study also confirmed the volume-outcome relationship. The hazard ratio was 1.23 (p < 0.01) for hospitals that performed <5 pancreatectomies annually compared to those that performed ≥20. Because 61% of ARP centers had performed ≥10 cases annually compared to 12% for CCCP and none for CCP, as expected, the better outcome observed with ARP centers can partly be explained by this volume-outcome relationship.
      Another reason for the better outcomes at high-volume centers is thought to be due to the surgeon's experience/volume.
      • Birkmeyer J.D.
      • Finks J.F.
      • O'Reilly A.
      • Oerline M.
      • Carline A.M.
      • Nunn A.R.
      • et al.
      Surgical skill and complication rates after bariatric surgery.
      • Birkmeyer J.D.
      • Stukel T.A.
      • Siewers A.E.
      • Goodney P.P.
      • Wennberg D.E.
      • Lucas F.L.
      Surgeon volume and operative mortality in the United States.
      • Mamidanna R.
      • Ni Z.
      • Anderson O.
      • Spiegelhalter S.D.
      • Bottle A.
      • Aylin P.
      • et al.
      Surgeon volume and cancer esophagectomy, gastrectomy, and pancreatectomy: a population-based study in England.
      Historically, a high volume pancreatic surgeon was classified as one having performed 3–5 pancreatectomies annually.
      • Birkmeyer J.D.
      • Stukel T.A.
      • Siewers A.E.
      • Goodney P.P.
      • Wennberg D.E.
      • Lucas F.L.
      Surgeon volume and operative mortality in the United States.
      • Urbach D.R.
      • Austin P.C.
      Conventional models overestimate the statistical significance of volume-outcome associations, compared with multilevel models.
      • Geraedts M.
      • de Cruppé W.
      • Blum K.
      • Ohmann C.
      Implementation and effects of Germany's minimum volume regulations: results of the accompanying research.
      • Ho V.
      • Heslin M.J.
      • Yun H.
      • Howard L.
      Trends in hospital and surgeon volume and operative mortality for cancer surgery.
      These thresholds, however, were arbitrarily selected without critical analysis. However, in a recent large population-based cohort study of 9116 pancreatectomies from England, Mamidanna et al. found that the threshold for classifying pancreatic surgeon as a high-volume practitioner was 12–31 cases annually.
      • Birkmeyer J.D.
      • Finks J.F.
      • O'Reilly A.
      • Oerline M.
      • Carline A.M.
      • Nunn A.R.
      • et al.
      Surgical skill and complication rates after bariatric surgery.
      Although the relationship among volume, surgeon's experience and outcome following pancreatectomy has been well described, to our knowledge, no prior studies have evaluated the relationship between outcomes and facility type, especially long-term outcomes rather than perioperative outcomes. Our study is the first to demonstrate a facility-outcome relationship for patients undergoing pancreatectomy for cancer. After controlling for demographic and clinical factors, we found that facility type independently predicted the risk of deaths from all causes; the risk was lowest among ARP patients than those of CCP and CCCP.
      There are several possible explanations for this observation. Perhaps ARP centers treat healthier patients than other centers. Using the Charlson comorbidity index as a surrogate of surgical risk, we found that ARP had a slightly higher percentage of patients with a zero Charlson comorbidity index than CCP and CCCP. However, we believe that the absolute difference of 3%–4% is unlikely to be clinically significant and therefore doubt if patient selection plays a role.
      Perhaps the low survivorship in CCP and CCCP is related to failure to rescue. Many studies have shown that mortality from high-risk cancer operations such as pancreatectomy may not necessarily be related to differences in complication rate, but rather to a hospital's ability to quickly recognize and effectively address a major complication.
      • Ghaferi A.A.
      • Birkmeyer J.D.
      • Dimick J.B.
      Hospital volume and failure to rescue with high-risk surgery.
      • Ward M.M.
      • Jaana M.
      • Wakefield D.S.
      • Ohsfeldt R.L.
      • Schneider J.E.
      • Miller T.
      • et al.
      What would be the effect of referral to high-volume hospitals in a largely rural state?.
      • Khuri S.F.
      • Henderson W.G.
      The case against volume as a measure of quality of surgical care.
      It is plausible that ARP centers have better supporting infrastructure to intervene when patients have major complications following pancreatectomy.
      • Ghaferi A.A.
      • Birkmeyer J.D.
      • Dimick J.B.
      Hospital volume and failure to rescue with high-risk surgery.
      • Aiken L.H.
      • Clarke S.P.
      • Sloane D.M.
      • Sochalski J.
      • Silber J.H.
      Hospital nurse staffing and patient mortality, nurse burnout, and job dissatisfaction.
      • Aiken L.H.
      • Clarke S.P.
      • Cheung R.B.
      • Sloane D.M.
      • Silber J.H.
      Educational levels of hospital nurses and surgical patient mortality.
      • Wennberg D.E.
      • Lucas F.L.
      • Birkmeyer J.D.
      • Bredenberg C.E.
      • Fisher E.S.
      Variation in carotid endarterectomy mortality in the Medicare population: trial hospitals, volume, and patient characteristics.
      • Pronovost P.J.
      • Angus D.C.
      • Dorman T.
      • Robinson K.A.
      • Dremsizov T.T.
      • Young T.L.
      Physician staffing patterns and clinical outcomes in critically ill patients: a systematic review.
      In our study, we found that the 30-day and 90-day mortality rates were significantly higher in CCP and CCCP when compared to ARP, thus suggesting that ARP may be better at rescuing patients.
      One of the unique aspects of our study is that we were able to demonstrate that long-term survival was also associated with facility type. Many studies have evaluated the volume-outcome relationship based on short-term endpoint (i.e. perioperative mortality), but to our knowledge, ours is the first to look at such a relationship from a longitudinal perspective. It is plausible that compared to CCCP and CCP, ARP centers possess unique components that result in better OS. ARP centers had lower percentage of patients who were uninsured and higher percentage of those with private insurance. Additionally, it has higher percentage of patients belonging to the higher income bracket (≥$63,000) and higher education area. These socioeconomic differences may partly explain the long-term survival advantage witnessed by ARP centers.
      Another plausible reason for improved OS at ARP centers is the quality of the surgery performed. Positive margins portend a poor outcome in resected pancreatic cancer,
      • Chang D.K.
      • Johns A.L.
      • Merrett N.D.
      • Gill A.J.
      • Colvin E.K.
      • Scarlett C.J.
      • et al.
      Margin clearance and outcome in resected pancreatic cancer.
      • Kimbrough C.W.
      • St Hill C.R.
      • Martin R.C.
      • McMasters K.M.
      • Scoggins C.R.
      Tumor-positive resection margins reflect an aggressive tumor biology in pancreatic cancer.
      • Van den Broeck A.
      • Sergeant G.
      • Ectors N.
      • Van Steenbergen W.
      • Aerts R.
      • Topal B.
      Patterns of recurrence after curative resection of pancreatic ductal adenocarcinoma.
      • Ethun C.G.
      • Delman K.A.
      The importance of surgical margins in melanoma.
      and our study found that positive resection margin significantly increases mortality by 54%. Compared to ARP, CCP and CCCP had a significantly higher percentage of patients with positive resection margins. The likelihood of having clear margins following a pancreatectomy may be related to case-volume. Our study, which supports the findings of others,
      • Bilimoria K.Y.
      • Talamonti M.S.
      • Sener S.F.
      • Bilimoria M.M.
      • Stewart A.K.
      • Winchester D.P.
      • et al.
      Effect of hospital volume on margin status after pancreaticoduodenectomy for cancer.
      found that high-volume centers such as those at ARP centers were more likely to achieve negative surgical margins.
      Finally, access to care might also play a critical role in OS. In our study, patients treated at ARP tend to belong to a higher socioeconomic stratum than those treated at CCCP or CCP. It has been proven in multiple studies that adjuvant chemotherapy or chemoradiotherapy confer a survival benefit compared to post-pancreatectomy observation alone.
      • Neoptolemos J.P.
      • Stocken D.D.
      • Tudur Smith C.
      • Bassi C.
      • Ghaneh P.
      • Owen E.
      • et al.
      Adjuvant 5-fluorouracil and folinic acid vs observation for pancreatic cancer: composite data from the ESPAC-1 and -3(v1) trials.
      • Oettle H.
      • Neuhaus P.
      • Hochhaus A.
      • Hartmann J.T.
      • Gellert K.
      • Ridwelski K.
      • et al.
      Adjuvant chemotherapy with gemcitabine and long-term outcomes among patients with resected pancreatic cancer: the CONKO-001 randomized trial.
      • Liao W.C.
      • Chien K.L.
      • Lin Y.L.
      • Wu M.S.
      • Lin J.T.
      • Wang H.P.
      • et al.
      Adjuvant treatments for resected pancreatic adenocarcinoma: a systematic review and network meta-analysis.
      While survival benefit of adjuvant chemotherapy in the form of six months of gemcitabine or 5-fluorouracil plus leucovorin is very well established, the same cannot be said about chemo-radiotherapy. European Study Group for Pancreatic cancer 1 trial
      • Neoptolemos J.P.
      • Stocken D.D.
      • Friess H.
      • Bassi C.
      • Dunn J.A.
      • Hickey H.
      • et al.
      A randomized trial of chemoradiotherapy and chemotherapy after resection of pancreatic cancer.
      (ESPAC-1) trial reported inferior outcomes patients undergoing chemo-radiotherapy compared to patients who did not receive chemo-radiotherapy. Gastrointestinal Tumor Study Group
      • Gastrointestinal Tumor Study Group
      Further evidence of effective adjuvant combined radiation and chemotherapy following curative resection of pancreatic cancer.
      (GITSG) and European Organization for Research and Treatment of Cancer
      • Klinkenbijl J.H.
      • Jeekel J.
      • Sahmoud T.
      • van Pel R.
      • Couvreur M.L.
      • Veenhof C.H.
      • et al.
      Adjuvant radiotherapy and 5-fluorouracil after curative resection of cancer of the pancreas and periampullary region: phase III trial of the EORTC gastrointestinal tract cancer cooperative group.
      (EORTC) demonstrated improvements in median survival in patients undergoing chemo-radiotherapy compared to who did not receive chemo-radiotherapy. Although controversial, 6 weeks of 5-fluorouracil based chemo-radiotherapy followed by chemotherapy is still considered an acceptable for of adjuvant therapy. More patients in ARP centers received adjuvant chemotherapy compared to CCP and CCCP centers (59.7% vs 50% and 56.9%, p < 0.01). On the other hand, adjuvant chemo-radiation was less adopted in ARP centers compared to CCP and CCCP (37.6% vs 38.3% and 40.2%). This differential distribution of adjuvant chemotherapy and chemo-radiation can partly contribute to differences in outcomes between these centers.
      Our study does have several limitations. One such limitation is that patients may have visited more than one facility and therefore it is difficult to delineate the true impact of facility type. Referral patterns and how patients select their treatment center are not captured in the database. Another limitation is that the study does not account for possible migration of surgeons moving from one facility type to another. Also, since there is no information on accreditation status of facilities in the NCDB PUF data, the calculation of AHV was based on some assumptions. This study does not account for exact adjuvant chemotherapy protocol or adjuvant chemo-radiotherapy protocol followed after the surgery. The study does not tease out which factor exactly contributes to advantageous survival outcomes: is this a direct measure of better surgical or better choice of adjuvant therapy following the surgery or a combination of both. Finally, our analysis lacks cause-specific survival information since the cause of death was not captured in NCDB.

      Conclusion

      In summary, we report a facility-outcome relationship for patients who underwent pancreatectomy for pancreatic adenocarcinoma. This relationship was based on long-term outcome rather than short-term outcome (i.e. perioperative mortality). Academic Research Program had better outcome than Community Cancer Program and Comprehensive Community Cancer Program following pancreatectomy, despite it having higher percentages of high grade tumors, higher stage tumors, and positive lymph nodes. ARP is likely a surrogate for critical components that allow ARP to deliver higher quality of care such as lower mortality rate, lower rate of positive margins, higher number of lymph nodes retrieved, higher percentage of patients receiving adjuvant chemotherapy, and higher surgical volume.

      Funding

      This work was supported by the Charles D. Knight, Sr. Endowed Professor of Surgery.

      Acknowledgment

      The data used in the study are derived from a de-identified NCDB file. The American College of Surgeons and the Commission on Cancer have not verified and are not responsible for the analytic or statistical methodology employed, or the conclusions drawn from these data by the investigator.

      Conflict of interest

      None declared.

      References

        • Siegel R.L.
        • Miller K.D.
        • Jemal A.
        Cancer statistics, 2016.
        CA Cancer J Clin. 2016; 66: 7-30
        • Gordon T.A.
        • Bowman H.M.
        • Tielsch J.M.
        • Bass E.B.
        • Burleyson G.P.
        • Cameron J.L.
        Statewide regionalization of pancreaticoduodenectomy and its effect on in-hospital mortality.
        Ann Surg. 1998; 228: 71-78
        • Birkmeyer J.D.
        • Finlayson E.V.
        • Birkmeyer C.M.
        Volume standards for high-risk surgical procedures: potential benefits of the Leapfrog initiative.
        Surgery. 2001; 130: 415-422
        • Schmidt C.M.
        • Turrini O.
        • Parikh P.
        • House M.G.
        • Zyromski N.J.
        • Nakeeb A.
        • et al.
        Effect of hospital volume, surgeon experience, and surgeon volume on patient outcomes after pancreaticoduodenectomy: a single-institution experience.
        Arch Surg. 2010; 145: 634-640
        • Ghaferi A.A.
        • Birkmeyer J.D.
        • Dimick J.B.
        Hospital volume and failure to rescue with high-risk surgery.
        Med Care. 2011; 49: 1076-1081
        • Ward M.M.
        • Jaana M.
        • Wakefield D.S.
        • Ohsfeldt R.L.
        • Schneider J.E.
        • Miller T.
        • et al.
        What would be the effect of referral to high-volume hospitals in a largely rural state?.
        J Rural Health. 2004; 20: 344-354
        • Khuri S.F.
        • Henderson W.G.
        The case against volume as a measure of quality of surgical care.
        World J Surg. 2005; 29: 1222-1229
        • Edge S.B.
        • Compton C.C.
        The American Joint Committee on Cancer: the 7th edition of the AJCC cancer staging manual and the future of TNM.
        Ann Surg Oncol. 2010; 17: 1471-1474
      1. Categories of accreditation. http://www.facs.org/cancer/coc/categories.html. Accessed November 9, 2014.

        • Charlson M.E.
        • Pompei P.
        • Ales K.L.
        • MacKenzie C.R.
        A new method of classifying prognostic comorbidity in longitudinal studies: development and validation.
        J Chronic Dis. 1987; 40: 373-383
        • Deyo R.A.
        • Cherkin D.C.
        • Ciol M.A.
        Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases.
        J Clin Epidemiol. 1992; 45: 613-619
        • Birkmeyer J.D.
        • Finks J.F.
        • O'Reilly A.
        • Oerline M.
        • Carline A.M.
        • Nunn A.R.
        • et al.
        Surgical skill and complication rates after bariatric surgery.
        N Engl J Med. 2013; 369: 1434-1442
        • Birkmeyer J.D.
        • Stukel T.A.
        • Siewers A.E.
        • Goodney P.P.
        • Wennberg D.E.
        • Lucas F.L.
        Surgeon volume and operative mortality in the United States.
        N Engl J Med. 2003; 349: 2117-2127
        • Mamidanna R.
        • Ni Z.
        • Anderson O.
        • Spiegelhalter S.D.
        • Bottle A.
        • Aylin P.
        • et al.
        Surgeon volume and cancer esophagectomy, gastrectomy, and pancreatectomy: a population-based study in England.
        Ann Surg. 2016; 263: 727-732
        • Urbach D.R.
        • Austin P.C.
        Conventional models overestimate the statistical significance of volume-outcome associations, compared with multilevel models.
        J Clin Epidemiol. 2005; 58: 391-400
        • Geraedts M.
        • de Cruppé W.
        • Blum K.
        • Ohmann C.
        Implementation and effects of Germany's minimum volume regulations: results of the accompanying research.
        Dtsch Arztebl Int. 2008; 105: 890-896
        • Ho V.
        • Heslin M.J.
        • Yun H.
        • Howard L.
        Trends in hospital and surgeon volume and operative mortality for cancer surgery.
        Ann Surg Oncol. 2006; 13: 851-858
        • Aiken L.H.
        • Clarke S.P.
        • Sloane D.M.
        • Sochalski J.
        • Silber J.H.
        Hospital nurse staffing and patient mortality, nurse burnout, and job dissatisfaction.
        JAMA. 2002; 288: 1987-1993
        • Aiken L.H.
        • Clarke S.P.
        • Cheung R.B.
        • Sloane D.M.
        • Silber J.H.
        Educational levels of hospital nurses and surgical patient mortality.
        JAMA. 2003; 290: 1617-1623
        • Wennberg D.E.
        • Lucas F.L.
        • Birkmeyer J.D.
        • Bredenberg C.E.
        • Fisher E.S.
        Variation in carotid endarterectomy mortality in the Medicare population: trial hospitals, volume, and patient characteristics.
        JAMA. 1998; 279: 1278-1281
        • Pronovost P.J.
        • Angus D.C.
        • Dorman T.
        • Robinson K.A.
        • Dremsizov T.T.
        • Young T.L.
        Physician staffing patterns and clinical outcomes in critically ill patients: a systematic review.
        JAMA. 2002; 288: 2151-2162
        • Chang D.K.
        • Johns A.L.
        • Merrett N.D.
        • Gill A.J.
        • Colvin E.K.
        • Scarlett C.J.
        • et al.
        Margin clearance and outcome in resected pancreatic cancer.
        J Clin Oncol. 2009; 27: 2855-2862
        • Kimbrough C.W.
        • St Hill C.R.
        • Martin R.C.
        • McMasters K.M.
        • Scoggins C.R.
        Tumor-positive resection margins reflect an aggressive tumor biology in pancreatic cancer.
        J Surg Oncol. 2013; 107: 602-607
        • Van den Broeck A.
        • Sergeant G.
        • Ectors N.
        • Van Steenbergen W.
        • Aerts R.
        • Topal B.
        Patterns of recurrence after curative resection of pancreatic ductal adenocarcinoma.
        Eur J Surg Oncol. 2009; 35: 600-604
        • Ethun C.G.
        • Delman K.A.
        The importance of surgical margins in melanoma.
        J Surg Oncol. 2016; 113: 339-345
        • Bilimoria K.Y.
        • Talamonti M.S.
        • Sener S.F.
        • Bilimoria M.M.
        • Stewart A.K.
        • Winchester D.P.
        • et al.
        Effect of hospital volume on margin status after pancreaticoduodenectomy for cancer.
        J Am Coll Surg. 2008; 207: 510-519
        • Neoptolemos J.P.
        • Stocken D.D.
        • Tudur Smith C.
        • Bassi C.
        • Ghaneh P.
        • Owen E.
        • et al.
        Adjuvant 5-fluorouracil and folinic acid vs observation for pancreatic cancer: composite data from the ESPAC-1 and -3(v1) trials.
        Br J Cancer. 2009; 100: 246-250
        • Oettle H.
        • Neuhaus P.
        • Hochhaus A.
        • Hartmann J.T.
        • Gellert K.
        • Ridwelski K.
        • et al.
        Adjuvant chemotherapy with gemcitabine and long-term outcomes among patients with resected pancreatic cancer: the CONKO-001 randomized trial.
        JAMA. 2013; 310: 1473-1481
        • Liao W.C.
        • Chien K.L.
        • Lin Y.L.
        • Wu M.S.
        • Lin J.T.
        • Wang H.P.
        • et al.
        Adjuvant treatments for resected pancreatic adenocarcinoma: a systematic review and network meta-analysis.
        Lancet Oncol. 2013; 14: 1095-1103
        • Neoptolemos J.P.
        • Stocken D.D.
        • Friess H.
        • Bassi C.
        • Dunn J.A.
        • Hickey H.
        • et al.
        A randomized trial of chemoradiotherapy and chemotherapy after resection of pancreatic cancer.
        N Engl J. 2004; 350: 1200-1210
        • Gastrointestinal Tumor Study Group
        Further evidence of effective adjuvant combined radiation and chemotherapy following curative resection of pancreatic cancer.
        Cancer. 1987; 59: 2006-2010
        • Klinkenbijl J.H.
        • Jeekel J.
        • Sahmoud T.
        • van Pel R.
        • Couvreur M.L.
        • Veenhof C.H.
        • et al.
        Adjuvant radiotherapy and 5-fluorouracil after curative resection of cancer of the pancreas and periampullary region: phase III trial of the EORTC gastrointestinal tract cancer cooperative group.
        Ann Surg. 1999; 230: 776-782