Advertisement

Can postoperative process of care utilization or complication rates explain the volume-cost relationship for cancer surgery?

Open AccessPublished:April 21, 2017DOI:https://doi.org/10.1016/j.surg.2017.03.004

      Background

      Past studies identify an association between provider volume and outcomes, but less is known about the volume-cost relationship for cancer surgery. We analyze the volume-cost relationship for 6 cancer operations and explore whether it is influenced by the occurrence of complications and/or utilization of processes of care.

      Methods

      Medicare hospital and inpatient claims for the years 2005 through 2009 were analyzed for 6 cancer resections: colectomy, rectal resection, pulmonary lobectomy, pneumonectomy, esophagectomy, and pancreatic resection. Regressions were first estimated to quantify the association of provider volume with costs, excluding measures of complications and processes of care as explanatory variables. Next, these variables were added to the regressions to test whether they weakened any previously observed volume-cost relationship.

      Results

      Higher hospital volume is associated with lower patient costs for esophagectomy but not for other operations. Higher surgeon volume reduces costs for most procedures, but this result weakens when processes of care are added to the regressions. Processes of care that are frequently implemented in response to adverse events are associated with 14% to 34% higher costs. Utilization of these processes is more prevalent among low-volume versus high-volume surgeons.

      Conclusion

      Processes of care implemented when complications occur explain much of the surgeon volume–cost relationship. Given that surgeon volume is readily observed, better outcomes and lower costs may be achieved by referring patients to high-volume surgeons. Increasing patient access to surgeons with lower rates of complications may be the most effective strategy for avoiding costly processes of care, controlling expenditure growth.
      Cancer is the second most costly medical condition in the United States with total direct expenses reaching over 88.6 billion in 2011.

      Agency for Healthcare Research and Quality. Total expenses and percent distribution for selected conditions by type of service: United States, 2011. Med Expend Panel Surv Househ Compon Data. Available from: https://meps.ahrq.gov/data_stats/tables_compendia_hh_interactive.jsp?_SERVICE=MEPSSocket0&_PROGRAM=MEPSPGM.TC.SAS&File=HCFY2011&Table=HCFY2011_CNDXP_C&_Debug.

      In a survey conducted in early 2010 by the American Cancer Society Cancer Action Network, 12% of the respondents put off getting a recommended cancer treatment or test due to costs, and 20% of families with someone with cancer have trouble paying for basic necessitates or other bills.
      • American Cancer Society Cancer Action Network
      A national poll facing cancer in the health care system.
      Since the costs of cancer care are expected to rise more quickly than overall medical expenditures,
      • National Cancer Institute
      Cancer trends progress report 2011-2012.
      costs will continue to be a burden for cancer patients and the overall health system that must be addressed.
      Although several studies have examined the association between case volume and patient costs for complex cancer operations, a thorough examination of cost triggers is lacking. Some studies analyzed the association between hospital volume and costs, without adjusting for a potential association between surgeon volume and costs.
      • Sosa J.A.
      • Bowman H.M.
      • Gordon T.A.
      • Bass E.B.
      • Yeo C.J.
      • Lillemoe K.D.
      • et al.
      Importance of hospital volume in the overall management of pancreatic cancer.
      • Swisher S.G.
      • Deford L.
      • Merriman K.W.
      • Walsh G.L.
      • Smythe R.
      • Vaporicyan A.
      • et al.
      Effect of operative volume on morbidity, mortality, and hospital use after esophagectomy for cancer.
      • Kuo E.Y.
      • Chang Y.
      • Wright C.D.
      Impact of hospital volume on clinical and economic outcomes for esophagectomy.
      Other studies examined the relationship between provider volume and hospital charges, even though charges may not reflect the true cost of providing care.
      • Harmon J.W.
      • Tang D.G.
      • Gordon T.A.
      • Bowman H.M.
      • Choti M.A.
      • Kaufman H.S.
      • et al.
      Hospital volume can serve as a surrogate for surgeon volume for achieving excellent outcomes in colorectal resection.
      • Konety B.R.
      • Dhawan V.
      • Allareddy V.
      • O’Donnell M.A.
      Association between volume and charges for most frequently performed ambulatory and nonambulatory surgery for bladder cancer. Is more cheaper?.
      • Al-Qurayshi Z.
      • Robins R.
      • Hauch A.
      • Randolph G.W.
      • Kandil E.
      Association of surgeon volume with outcomes and cost savings following thyroidectomy.
      • Damle R.N.
      • Flahive J.M.
      • Davids J.S.
      • Maykel J.A.
      • Sweeney W.B.
      • Sturrock P.R.
      • et al.
      Surgeon volume correlates with reduced mortality and improved quality in the surgical management of diverticulitis.
      • Damle R.N.
      • Macomber C.W.
      • Flahive J.M.
      • Davids J.S.
      • Sweeney W.B.
      • Sturrock P.R.
      • et al.
      Surgeon volume and elective resection for colon cancer: an analysis of outcomes and use of laparoscopy.
      • Shi H.Y.
      • Chang H.T.
      • Culbertson R.
      • Chen Y.J.
      • Liao Y.C.
      • Hou M.F.
      Breast cancer surgery volume-cost associations: hierarchical linear regression and propensity score matching analysis in a nationwide Taiwan population.
      Finally, some studies examined only one procedure, one US state or a socialized health care system (the United Kingdom).
      • Harmon J.W.
      • Tang D.G.
      • Gordon T.A.
      • Bowman H.M.
      • Choti M.A.
      • Kaufman H.S.
      • et al.
      Hospital volume can serve as a surrogate for surgeon volume for achieving excellent outcomes in colorectal resection.
      • Konety B.R.
      • Dhawan V.
      • Allareddy V.
      • O’Donnell M.A.
      Association between volume and charges for most frequently performed ambulatory and nonambulatory surgery for bladder cancer. Is more cheaper?.
      • Al-Qurayshi Z.
      • Robins R.
      • Hauch A.
      • Randolph G.W.
      • Kandil E.
      Association of surgeon volume with outcomes and cost savings following thyroidectomy.
      • Damle R.N.
      • Flahive J.M.
      • Davids J.S.
      • Maykel J.A.
      • Sweeney W.B.
      • Sturrock P.R.
      • et al.
      Surgeon volume correlates with reduced mortality and improved quality in the surgical management of diverticulitis.
      • Damle R.N.
      • Macomber C.W.
      • Flahive J.M.
      • Davids J.S.
      • Sweeney W.B.
      • Sturrock P.R.
      • et al.
      Surgeon volume and elective resection for colon cancer: an analysis of outcomes and use of laparoscopy.
      • Shi H.Y.
      • Chang H.T.
      • Culbertson R.
      • Chen Y.J.
      • Liao Y.C.
      • Hou M.F.
      Breast cancer surgery volume-cost associations: hierarchical linear regression and propensity score matching analysis in a nationwide Taiwan population.
      • Bachmann M.
      • Peters T.
      • Harvey I.
      Costs and concentration of cancer care: evidence for pancreatic, oesophageal and gastric cancers in National Health Service hospitals.
      Understanding the nature of the volume-cost relationship would aid in determining whether directing cancer patients to high-volume providers would be a means to control rising health care costs.
      Previous research indicated that high surgeon volume, rather than high hospital volume, is associated with lower costs for cancer operations.
      • Ho V.
      • Aloia T.
      Hospital volume, surgeon volume, and patient costs for cancer surgery.
      More recent work corroborated the lack of association between hospital volume and costs but did not examine surgeon volume.
      • Nathan H.
      • Atoria C.L.
      • Bach P.B.
      • Elkin E.B.
      Hospital volume, complications, and cost of cancer surgery in the elderly.
      Other previous work showed that complications and use of processes of care increase costs.
      • Nathan H.
      • Atoria C.L.
      • Bach P.B.
      • Elkin E.B.
      Hospital volume, complications, and cost of cancer surgery in the elderly.
      • Short M.N.
      • Aloia T.A.
      • Ho V.
      The influence of complications on the costs of complex cancer surgery.
      • Short M.N.
      • Ho V.
      • Aloia T.A.
      Impact of processes of care aimed at complication reduction on the cost of complex cancer surgery.
      This study builds on the previous research by seeking to determine which practice differences are driving factors for the volume-cost relationship. This knowledge may be used to generate better-targeted policies that focus on changes that facilitate more cost-effective behavior across the spectrum of surgeon and/or hospital volume.

      Methods

      The analytical methods used were similar to our previously published studies of the association between operative complications and costs
      • Short M.N.
      • Aloia T.A.
      • Ho V.
      The influence of complications on the costs of complex cancer surgery.
      and the association between processes of care and costs.
      • Short M.N.
      • Ho V.
      • Aloia T.A.
      Impact of processes of care aimed at complication reduction on the cost of complex cancer surgery.
      Patient-level Medicare claims data from all 50 states for the years 2005-2009 were searched to identify patients with International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnosis and procedure codes who underwent one of the resections in this study
      • Ho V.
      • Aloia T.
      Hospital volume, surgeon volume, and patient costs for cancer surgery.
      : colectomy, rectal resection, pulmonary lobectomy, pancreatic resection, esophagectomy, and pneumonectomy. The 100% Medicare Provider and Analysis Review files were merged with the Carrier files in order to obtain both hospital- and surgeon-related charges for each inpatient stay where resection occurred. ICD-9-CM diagnosis and procedure codes were used to confirm that the cancer indication matched the resection type as a requirement for inclusion in the study.
      • Ho V.
      • Aloia T.
      Hospital volume, surgeon volume, and patient costs for cancer surgery.
      This study was approved by Rice University's Institutional Review Board.

      Costs and volume

      The Medicare Provider and Analysis Review data provided detailed information on hospital charges by revenue center for each discharge. Hospital charges are drawn from each hospital's chargemaster, which lists a fee that the hospital associates with each single procedure and supply item used for a patient's care.
      • Reinhardt U.E.
      The pricing of U.S. hospital services: chaos behind a veil of secrecy.
      Charges were deflated by the All-Urban Consumer Price Index to reflect 2005 dollars. After previously published studies, costs for each hospital stay were estimated by multiplying the reported patient charge by the hospital's cost-to-charge ratio in the same year.
      • Ho V.
      • Aloia T.
      Hospital volume, surgeon volume, and patient costs for cancer surgery.
      • Meltzer D.
      • Chung J.
      • Basu A.
      Does competition under Medicare Prospective Payment selectively reduce expenditures on high-cost patients?.
      • Cutler D.M.
      • Huckman R.S.
      Technological development and medical productivity: the diffusion of angioplasty in New York state.
      • Ho V.
      Learning and the evolution of medical technologies: the diffusion of coronary angioplasty.
      • Ho V.
      • Petersen L.A.
      Estimating cost savings from regionalizing cardiac procedures using hospital discharge data.
      • Healthcare Cost and Utilization Project
      Cost-to-charge ratio files.
      Medicare and other insurers do not require physicians to file cost reports, so we have no comprehensive source of data on physician costs. Instead, all physician billings to Medicare for the patient's hospital stay were identified in the Carrier claims files and summed to represent physician costs associated with the admission. While physician billings do not represent the resource costs of physician labor, they do reflect the amount that society must incur for the care of cancer patients covered by Medicare.
      Hospital and physician costs were then summed to calculate the explanatory variable of total costs for the entire inpatient stay. Hospital procedure volume was constructed by summing the total number of operations performed by each hospital in each year for each of the 6 cancer resections. Surgeon volume was calculated in the same manner.

      Complications and processes of care

      Complications were identified using the Agency for Healthcare Research and Quality hospital-level Patient Safety Indicators (PSI) methodology.
      • Short M.N.
      • Aloia T.A.
      • Ho V.
      The influence of complications on the costs of complex cancer surgery.
      • Agency for Healthcare Research and Quality
      Patient Safety Indicators overview.
      • Rivard P.E.
      • Luther S.L.
      • Christiansen C.L.
      • Zhao S.
      • Loveland S.
      • Elixhauser A.
      • et al.
      Using patient safety indicators to estimate the impact of potential adverse events on outcomes.
      • Romano P.S.
      • Geppert J.J.
      • Davies S.
      • Miller M.R.
      • Elixhauser A.
      • McDonald K.M.
      A national profile of patient safety in U.S. hospitals.
      These indicators focus on potentially preventable instances of complications and other iatrogenic events resulting from exposure to the health care system. Processes of care represent actions that providers take to achieve high-quality care and/or favorable patient outcomes.
      • Rubin H.R.
      • Pronovost P.
      • Diette G.B.
      The advantages and disadvantages of process-based measures of health care quality.
      After previous research on the volume and process of care in high-risk cancer operations for various cancers,
      • Short M.N.
      • Ho V.
      • Aloia T.A.
      Impact of processes of care aimed at complication reduction on the cost of complex cancer surgery.
      • Hollenbeck B.K.
      • Wei Y.
      • Birkmeyer J.D.
      Volume, process of care, and operative mortality for cystectomy for bladder cancer.
      • Birkmeyer J.D.
      • Sun Y.
      • Goldfaden A.
      • Birkmeyer N.J.O.
      • Stukel T.A.
      Volume and process of care in high-risk cancer surgery.
      the processes used were identified using ICD-9-CM procedure codes and Current Procedure Terminology (CPT) codes in the claims data. These processes are mostly items that are potentially optional and pre-emptively controllable by the operative and anesthetic teams, but they also include services that are used to avoid and/or ameliorate complications. Some of the processes are intraoperative, and some are postoperative.
      The specific codes were obtained from a previous study
      • Birkmeyer J.D.
      • Sun Y.
      • Goldfaden A.
      • Birkmeyer N.J.O.
      • Stukel T.A.
      Volume and process of care in high-risk cancer surgery.
      that defined processes based on a comprehensive literature review (including the literature on hospital safety practices) supplemented by the input of clinician experts. Once a candidate list was developed, ICD-9 and CPT coding manuals were consulted, and pilot tests were conducted to determine the subset of processes that could be measured in inpatient and outpatient claims data.

      Hospital, surgeon, and patient characteristics

      Data from the American Hospital Association Annual Survey of hospitals was used to create indicator variables for complex medical technologies, which are more likely to be at high-volume hospitals (whether the treating hospital has a computed tomography scanner, magnetic resonance imaging, and positron emission tomography), and the nurse-to-patient ratio. In addition, hospital-specific indicator variables were used to control for all observed and unobserved characteristics such as a hospital's reputation for high quality in the community.
      Patient-level characteristics used for risk adjustment include sex, age, race, and income. Secondary diagnosis codes were used to construct indicator variables for the 29 conditions comprising the Elixhauser comorbidity index.
      • Elixhauser A.
      • Steiner C.
      • Harris D.R.
      • Coffey R.M.
      Comorbidity measures for use with administrative data.
      Indicator variables were included for transfer patients and patients who had urgent/emergency admissions. Cancer stage was measured using secondary diagnosis codes for nodal involvement and organ metastasis.
      • Harmon J.W.
      • Tang D.G.
      • Gordon T.A.
      • Bowman H.M.
      • Choti M.A.
      • Kaufman H.S.
      • et al.
      Hospital volume can serve as a surrogate for surgeon volume for achieving excellent outcomes in colorectal resection.
      To adjust for disease-specific differences in procedure complexity and patient casemix, indicator variables were defined for particular operations and operative approaches specific to each procedure. Operating room time was measured as total hours claimed for anesthesia.
      • Short M.N.
      • Ho V.
      • Aloia T.A.
      Impact of processes of care aimed at complication reduction on the cost of complex cancer surgery.
      • Birkmeyer J.D.
      • Sun Y.
      • Goldfaden A.
      • Birkmeyer N.J.O.
      • Stukel T.A.
      Volume and process of care in high-risk cancer surgery.
      Procedure and tumor sites were defined on the basis of previous studies of these operations that used ICD-9-CM procedure and diagnosis codes.
      • Harmon J.W.
      • Tang D.G.
      • Gordon T.A.
      • Bowman H.M.
      • Choti M.A.
      • Kaufman H.S.
      • et al.
      Hospital volume can serve as a surrogate for surgeon volume for achieving excellent outcomes in colorectal resection.
      • Birkmeyer J.D.
      • Warshaw A.L.
      • Finlayson S.R.G.
      • Grove M.R.
      • Tosteson A.N.
      Relationship between hospital volume and late survival after pancreaticoduodenectomy.
      • Birkmeyer J.D.
      • Finlayson S.R.G.
      • Tosteson A.N.
      • Sharp S.M.
      • Warshaw A.L.
      • Fisher E.S.
      Effect of hospital volume on in-hospital mortality with pancreaticoduodenectomy.
      • Patti M.G.
      • Corvera C.U.
      • Glasgow R.E.
      • Way L.W.
      A hospital’s annual rate of esophagectomy influences the operative mortality rate.
      • Romano P.S.
      • Mark D.H.
      Patient and hospital characteristics related to in-hospital mortality after lung cancer resection.
      • Kang C.Y.
      • Halabi W.J.
      • Luo R.
      • Pigazzi A.
      • Nguyen N.T.
      • Stamos M.J.
      Laparoscopic colorectal surgery: a better look into the latest trends.
      The Medicare Physician Identification and Eligibility Registry file was used to identify the subspecialty of the surgeon operating on each patient.

      Statistical analysis

      Regressions were estimated separately for each of the 6 cancer resections using the natural log of total costs as the dependent variable. The explanatory variables of interest are the number of resections of the same type performed by the patient's surgeon and hospital in the calendar year, PSIs, and processes of care. We adjusted for the hospital, surgeon, and patient characteristics listed above. All regressions were performed using panel data methods, which involves the inclusion of year and hospital fixed effects. The 0/1 indicator variables for each hospital imply that volume coefficients quantify the relationship between within-hospital changes in volume and costs. Regressions were estimated using the xtreg command in Stata 12.1 (StataCorp, College Station, TX) with specifications to provide robust standard errors that account for clustering of patient data within hospitals.
      Regressions were first performed excluding PSIs and processes of care. We used Stata's mfp command to fit a multivariable fractional polynomial model for each procedure. Those polynomial models are flexible regression specifications that yield the best possible fit for the association between volume and costs. We looked for instances where the coefficient(s) on surgeon or hospital volume were statistically significant (P ≤ .05). Next, we re-estimated the regressions adding each group of factors (ie, PSIs and processes of care) individually and then jointly. We then checked whether the coefficient(s) on provider volume changed from being statistically significant to insignificant. We used the statistically significant coefficients on provider volume to predict the change in costs that would result from receiving an operation from a provider at the 5th, 50th, and 95th percentile of the volume distribution.

      Results

      Table I illustrates the differences in the procedures examined and the magnitude of the costs associated with them. The number of procedures performed in our sample varies from 2,981 pneumonectomies up to 150,733 colectomies. The median costs of an admission with resection range from $14,959 for colectomy up to $34,287 for esophagectomy.
      Table ISummary of total inpatient costs by procedure
      NMinMaxMedianMean
      Colectomy150,733$2,274$429,176$14,959$20,163
      Pulmonary lobectomy52,202$4,155$522,620$17,830$22,907
      Rectal resection25,892$2,349$201,246$15,473$20,063
      Pancreatic resection12,135$6,818$622,543$28,799$38,661
      Esophagectomy3,857$8,922$488,187$34,287$49,997
      Pneumonectomy2,981$6,225$295,048$19,846$27,553
      Table II compares the PSI and process of care rates between the highest and lowest tertiles of surgeon volume for each procedure. Differences in the PSI rates between tertiles are small and few are statistically significant. Not many are greater than 1 percentage point, and none are greater than 5 percentage points. Where there are statistically significant differences between the surgeon groups for processes of care, higher-volume surgeons tend to use fewer processes than low-volume surgeons, although higher-volume surgeons tend to use more arterial lines and epidural anesthesia paired with daily epidural management. Higher-volume surgeons tend to have lower rates of transfusions, consultations, and total parenteral nutrition (TPN) than low-volume surgeons. The differences are more pronounced for less common operations.
      Table IIAverage PSI and process of care rates by surgeon tertile
      ColectomyPulmonary lobectomyRectal resection
      Low-volHigh-volDifferenceLow-volHigh-volDifferenceLow-volHigh-volDifference
      1–510–1021–26–931–25–46
      PSIs
       Accidental puncture/laceration2.47%2.05%0.42%
      P < .001.
      1.11%1.25%−0.15%3.12%2.99%0.13%
       Anesthesia complications0.02%0.01%0.01%0.00%0.05%−0.05%
      P < .001.
      0.05%0.03%0.03%
       Death among operative inpatients with serious treatable complications1.58%1.30%0.28%
      P < .01.
      1.52%1.50%0.02%1.20%0.95%0.25%
       Decubitus ulcer0.88%0.65%0.23%
      P < .01.
      0.48%0.32%0.16%0.68%0.79%−0.11%
       Foreign body left in during procedure0.03%0.02%0.01%0.00%0.03%−0.03%
      P < .01.
      0.07%0.06%0.01%
       Iatrogenic pneumothorax0.23%0.12%0.10%
      P < .01.
      3.71%2.48%1.23%
      P < .001.
      0.12%0.11%0.01%
       Postoperative hemorrhage/hematoma0.25%0.26%−0.01%0.21%0.13%0.08%0.43%0.43%0.00%
       Postoperative hip fracture0.02%0.01%0.01%0.03%0.01%0.02%0.01%0.00%0.01%
       Postoperative physiologic and metabolic derangement0.07%0.07%−0.01%0.15%0.13%0.02%0.08%0.10%−0.03%
       Postoperative respiratory failure1.78%1.97%−0.18%2.70%3.03%−0.33%2.67%2.64%0.03%
       Postoperative thromboembolism (PE/DVT)1.83%1.78%0.05%1.50%1.22%0.28%1.33%1.60%−0.26%
       Postoperative wound dehiscence0.50%0.36%0.14%
      P < .05.
      0.24%0.26%−0.01%
      Processes of care
       Pulmonary artery catheter1.40%1.21%0.19%
      P < .05.
      2.25%1.89%0.36%0.80%1.34%−0.54%
      P < .01.
       Arterial line17.34%15.88%1.45%
      P < .001.
      74.03%79.38%−5.35%
      P < .001.
      18.48%21.08%−2.59%
      P < .001.
       Central venous catheter
      Central venous catheter alone.
      20.97%19.66%1.31%28.07%24.48%3.59%19.98%20.04%−0.06%
       Epidural anesthesia14.50%15.39%−0.89%
      P < .01.
      52.67%54.26%−1.59%25.22%25.03%0.19%
       Daily epidural management13.11%13.47%−0.36%46.24%49.64%−3.39%
      P < .001.
      21.93%23.04%−1.11%
       Transfusion PRBC or autotransfusion24.06%21.04%3.02%
      P < .001.
      12.68%8.10%4.57%
      P < .001.
      20.95%16.52%4.43%
      P < .001.
       Total parenteral nutrition7.38%6.10%1.28%
      P < .001.
      0.78%0.46%0.32%
      P < .05.
      6.32%5.43%0.89%
      P < .05.
       Critical care consultations17.05%12.95%4.10%
      P < .001.
      19.09%14.49%4.60%
      P < .001.
      12.34%9.75%2.59%
      P < .001.
       Inpatient consultations30.91%29.44%1.47%
      P < .001.
      30.65%24.75%5.90%
      P < .001.
      24.51%22.22%2.28%
      P < .01.
      Pancreatic resectionEsophagectomyPneumonectomy
      Low-volHigh-volDifferenceLow-volHigh-volDifferenceLow-volHigh-volDifference
      14–7214–3712–17
      PSIs
       Accidental puncture/laceration3.68%2.89%0.79%5.11%3.94%1.18%2.08%1.87%0.21%
       Anesthesia complications
      P < .05.
       Death among operative inpatients with serious treatable complications4.40%1.83%2.56%
      P < .001.
      5.86%2.76%3.10%
      P < .01.
      4.52%4.81%−0.28%
       Decubitus ulcer0.97%0.74%0.23%1.04%1.70%−0.66%0.47%0.38%0.09%
       Foreign body left in during procedure0.13%0.10%
       Iatrogenic pneumothorax0.55%0.30%0.26%
       Postoperative hemorrhage/hematoma1.57%1.12%0.44%0.20%0.31%−0.11%0.40%0.44%−0.04%
       Postoperative hip fracture
       Postoperative physiologic and metabolic derangement0.23%0.06%0.17%0.40%0.05%0.35%0.27%0.11%0.16%
       Postoperative respiratory failure6.98%5.22%1.76%
      P < .05.
      16.93%12.39%4.54%
      P < .05.
      4.49%4.87%−0.38%
       Postoperative thromboembolism (PE/DVT)2.09%2.07%0.02%3.92%2.92%1.00%2.04%1.70%0.34%
       Postoperative wound dehiscence0.18%0.35%−0.16%0.20%0.16%0.04%
      Processes of care
       Pulmonary artery catheter5.00%4.73%0.27%8.54%8.32%0.22%4.09%3.48%0.61%
       Arterial line61.76%77.69%−15.93%
      P < .001.
      82.17%89.26%−7.09%
      P < .001.
      81.90%82.93%−1.03%
       Central venous catheter
      Central venous catheter alone.
      57.34%56.81%0.53%58.89%49.79%9.10%36.29%35.57%0.72%
       Epidural anesthesia37.82%43.95%−6.12%
      P < .001.
      51.89%59.26%−7.38%
      P < .01.
      57.41%60.30%−2.90%
       Daily epidural management33.29%44.38%−11.09%
      P < .001.
      48.81%61.89%−13.08%
      P < .001.
      51.64%56.01%−4.36%
      P < .05.
       Transfusion PRBC or autotransfusion25.53%20.60%4.93%
      P < .001.
      12.86%10.33%2.53%19.50%13.54%5.97%
      P < .001.
       Total parenteral nutrition12.80%9.56%3.23%
      P < .01.
      6.21%2.33%3.87%
      P < .001.
      1.01%0.89%0.12%
       Critical care consultations29.49%21.85%7.64%
      P < .001.
      38.88%34.07%4.81%
      P < .05.
      25.40%23.65%1.75%
       Inpatient consultations41.23%23.88%17.35%
      P < .001.
      37.49%21.35%16.14%
      P < .001.
      33.48%27.67%5.81%
      P < .01.
      PRBC, Packed red blood cells; PE/DVT, pulmonary embolism/deep vein thrombosis.
      P < .001.
      P < .01.
      P < .05.
      § Central venous catheter alone.
      For brevity, we report estimates of the relation between provider volume and costs from the cost regressions in the Supplementary Table. We found that esophagectomy is the only procedure of the 6 we studied that exhibits a statistically significant relationship between higher volume and lower costs. The relationship was strictly linear. Using regression estimates that excluded PSIs and processes of care, each additional esophagectomy performed per year at a hospital lowered patient costs 0.6% (P = .016). When PSIs and processes of care were included in the cost regressions, the estimated reduction in costs for each additional esophagectomy was 0.4% (P = .037).
      In cases where the estimated association between surgeon volume and patient costs were statistically significant, the relation was nonlinear. Therefore, the estimated cost savings for each additional operation varied according to the number of operations performed. Table III reports the predicted change in patient costs associated with surgeon volume at the 5th, 50th, and 95th percentile of the distribution for each of the 6 procedures. When the predictions were derived from adjusted regression estimates that exclude measures of PSIs and processes of care, the estimated cost savings are substantial for 4 of the 6 operations. If a surgeon performs the median number of operations (eg, ranging from 2 per year for pancreatic resection to 7 per year for colectomy), the estimated cost savings relative to performing 1 operation per year range from 2.4% to 2.8% for colectomy, pulmonary lobectomy, rectal resection, and pancreatic resection.
      Table IIIEstimated percentage change in costs from increasing surgeon volume, with and without adjustment for Patient Safety Indicators and processes of care, by percentile of procedure-specific volume distribution
      PercentileColectomyPulmonary lobectomyRectal resection
      Annual surgeon volumew/o PSI/POCw/PSI/POCAnnual surgeon volumew/o PSI/POCw/PSI/POCAnnual surgeon volumew/o PSI/POCw/PSI/POC
      5th
      Because the 5th percentile was 1 per year for each procedure, the estimated change in costs is calculated for an increase in surgeon volume from 1 to 2 procedures per year.
      1−0.601−2.001−1.4−0.3
      50th7−2.403−2.503−2.6−0.7
      95th22−5.4017−5.9010−7.6−3.0
      PercentilePancreatic resectionEsophagectomyPneumonectomy
      Annual surgeon volumew/o PSI/POCw/PSI/POCAnnual surgeon volumew/o PSI/POCw/PSI/POCAnnual surgeon volumew/o PSI/POCw/PSI/POC
      5th
      Because the 5th percentile was 1 per year for each procedure, the estimated change in costs is calculated for an increase in surgeon volume from 1 to 2 procedures per year.
      1−2.8−1.4100100
      50th2−2.8−1.4200100
      95th14−8.5−4.1800300
      POC, Processes of care.
      Because the 5th percentile was 1 per year for each procedure, the estimated change in costs is calculated for an increase in surgeon volume from 1 to 2 procedures per year.
      If a surgeon is at the 95th percentile for the number of operations performed per year, the estimated cost savings relative to performing one operation per year are even greater, ranging from 5.4% for colectomy to 8.5% for pancreatic resection. Applying the mean costs for each operation from Table I, these estimates suggest a savings of $1,089 per patient for colectomy and $3,286 for pancreatic resection. The regression estimates revealed no significant relationship between surgeon volume and costs for the remaining 2 procedures, esophagectomy and pneumonectomy.
      When we predicted cost savings associated with increased surgeon volume using regressions that added PSI measures and processes of care, patient costs were predicted to decline with higher volume for only 2 of the 6 operations: rectal resection and pancreatic resection. Moreover, the estimated cost decreases were smaller than those obtained from regressions that excluded PSI measures and processes of care. A surgeon performing 2 pancreatic resections per year (50th percentile) was estimated to reduce costs by 1.4 percent, and a surgeon performing 14 operations per year (95th percentile) was estimated to reduce costs by 4.1 percent.
      Additional analyses revealed that the estimated association between surgeon volume and costs when only PSIs were included in the regressions were similar to the estimates excluding both PSIs and processes of care. In contrast, when only processes of care were added to the regressions, we observed the same weakening of the relation between surgeon volume and costs that we found when both PSIs and processes of care were accounted for.
      Table IV lists the coefficients for the PSI and processes of care variables from the full regression. In general, more PSIs and higher use of processes of care are associated with increased costs of cancer care. Many PSIs are associated with cost increases of 20% or more, dwarfing the effect of volume on costs in absolute value.
      Table IVEstimates for remaining variables in regression with PSI and processes of care
      ColectomyPulmonary LobectomyRectal ResectionPancreatic ResectionEsophagectomyPneumonectomy
      PSIs
       Accidental puncture/laceration0.114
      P < .001.
      (14.08)
      0.0961
      P < .001.
      (5.95)
      0.0880
      P < .001.
      (5.84)
      0.0968
      P < .001.
      (4.06)
      0.0671 (1.64)0.0341 (0.49)
       Anesthesia complications−0.0949
      P < .05.
      (−2.02)
      0.0573 (0.93)−0.0664 (−0.52)
       Death among surgical inpatients with serious treatable complications0.0441
      P < .01.
      (2.88)
      0.330
      P < .001.
      (13.10)
      0.0357 (0.93)0.135
      P < .01.
      (3.09)
      0.142
      P < .01.
      (2.59)
      0.224
      P < .001.
      (3.33)
       Decubitus ulcer0.254
      P < .001.
      (17.25)
      0.385
      P < .001.
      (9.34)
      0.226
      P < .001.
      (6.71)
      0.296
      P < .001.
      (6.89)
      0.256
      P < .001.
      (4.05)
      0.495
      P < .001.
      (3.36)
       Foreign body left in during procedure0.141 (1.35)0.161 (1.43)0.256
      P < .01.
      (2.72)
       Iatrogenic pneumothorax0.156
      P < .001.
      (4.37)
      0.0184 (1.81)0.0897 (1.72)0.0828 (0.91)
       Postoperative hemorrhage/hematoma0.293
      P < .001.
      (11.77)
      0.226
      P < .001.
      (3.59)
      0.285
      P < .001.
      (6.95)
      0.347
      P < .001.
      (8.63)
      0.105 (0.71)0.148 (1.32)
       Postoperative hip fracture0.589
      P < .001.
      (8.59)
      0.362
      P < .001.
      (3.31)
       Postoperative physiologic and metabolic derangement0.348
      P < .001.
      (6.87)
      0.191
      P < .01.
      (3.03)
      0.280
      P < .01.
      (2.84)
      0.208 (1.20)0.217 (1.02)0.611 (1.92)
       Postoperative respiratory failure0.356
      P < .001.
      (29.04)
      0.564
      P < .001.
      (29.24)
      0.338
      P < .001.
      (13.96)
      0.358
      P < .001.
      (11.10)
      0.398
      P < .001.
      (13.44)
      0.441
      P < .001.
      (6.11)
       Postoperative thromboembolism (PE/DVT)0.283
      P < .001.
      (31.19)
      0.265
      P < .001.
      (13.39)
      0.259
      P < .001.
      (11.17)
      0.258
      P < .001.
      (9.71)
      0.192
      P < .001.
      (4.46)
      0.242
      P < .01.
      (3.18)
       Postoperative wound dehiscence0.521
      P < .001.
      (23.44)
      0.551
      P < .001.
      (9.18)
      0.397
      P < .001.
      (4.85)
      0.422
      P < .001.
      (3.67)
      Processes of care
       Pulmonary artery catheter0.220
      P < .001.
      (15.51)
      0.210
      P < .001.
      (12.76)
      0.189
      P < .001.
      (5.97)
      0.228
      P < .001.
      (9.27)
      0.286
      P < .001.
      (6.73)
      0.220
      P < .01.
      (3.02)
       Arterial line0.111
      P < .001.
      (25.80)
      0.00442 (0.94)0.0810
      P < .001.
      (10.47)
      0.0409
      P < .001.
      (4.09)
      −0.0407 (−1.54)−0.0237 (−0.91)
       Central venous catheter
      Central venous catheter alone.
      0.203
      P < .001.
      (47.61)
      0.129
      P < .001.
      (19.41)
      0.177
      P < .001.
      (20.76)
      0.0988
      P < .001.
      (10.01)
      0.170
      P < .001.
      (6.87)
      0.147
      P < .001.
      (6.37)
       Epidural anesthesia−0.00428 (−0.90)−0.0110
      P < .05.
      (−2.23)
      0.00769 (0.82)−0.0114 (−0.84)−0.0168 (−0.69)−0.0297 (−1.09)
       Daily epidural management0.0229
      P < .001.
      (4.57)
      0.0108
      P < .05.
      (2.00)
      0.0111 (1.20)−0.00781 (−0.56)0.00395 (0.16)0.00709 (0.26)
       Transfusion PRBC or autotransfusion0.0756
      P < .001.
      (24.88)
      0.0894
      P < .001.
      (14.87)
      0.0699
      P < .001.
      (10.61)
      −0.0263
      P < .05.
      (−2.51)
      −0.0677
      P < .05.
      (−2.44)
      0.0350 (1.43)
       Total parenteral nutrition0.211
      P < .001.
      (37.46)
      0.323
      P < .001.
      (11.49)
      0.291
      P < .001.
      (23.73)
      0.141
      P < .001.
      (6.44)
      0.0742 (1.45)0.0308 (0.28)
       Critical care consultations0.336
      P < .001.
      (68.91)
      0.346
      P < .001.
      (36.40)
      0.322
      P < .001.
      (27.45)
      0.294
      P < .001.
      (18.61)
      0.321
      P < .001.
      (13.81)
      0.340
      P < .001.
      (10.85)
       Inpatient consultations0.238
      P < .001.
      (73.48)
      0.201
      P < .001.
      (37.28)
      0.230
      P < .001.
      (31.84)
      0.214
      P < .001.
      (17.03)
      0.340
      P < .001.
      (13.08)
      0.221
      P < .001.
      (8.73)
      PRBC, Packed red blood cells; PE/DVT, pulmonary embolism/deep vein thrombosis.
      P < .001.
      P < .05.
      P < .01.
      § Central venous catheter alone.
      The processes, which are often performed to ameliorate the impact of complications (TPN, critical care, and inpatient consultations), have the largest coefficients among significant process of care effects, increasing costs between 14% and 34%. Few of these coefficients imply cost increases less than 20%. For the few cases observed in Table II where high-volume surgeons have significantly higher process of care utilization, the estimated relation between the process and costs tends to be insignificant or small.

      Discussion

      For policy purposes, identification of strategies for controlling the cost of the entire patient stay will be more effective in restraining growth in health care costs than examining operative costs alone. Patient treatment costs are hypothesized to depend on the characteristics of the patient as well as the characteristics of the surgeon and hospital from which the patient received treatment. Therefore, this study examined costs per entire inpatient stay, rather than total costs for postoperative hospital care or just the cost of performing an operation. The goal of the study was to determine the mechanism(s) that explains the relationship between provider volume and cost control in major cancer operations.
      Our findings are consistent with a recent study that found no meaningful association between hospital volume and cost.
      • Nathan H.
      • Atoria C.L.
      • Bach P.B.
      • Elkin E.B.
      Hospital volume, complications, and cost of cancer surgery in the elderly.
      Instead, we found that many PSIs are associated with cost increases of 20% or more. Moreover, processes of care that are performed to ameliorate the impact of complications increase costs between 14% and 34%. Regression analyses that adjust for processes of care reduce, and in some cases eliminate, the association between surgeon volume and patient costs that appeared in regressions that only adjusted for patient and hospital characteristics.
      Therefore, the differences in the frequency at which processes of care occur between low- versus high-volume surgeons explain the previously observed relation between higher procedure volume and lower costs.
      Among the 6 cancer operations we examined, patients treated by low-volume surgeons were less likely to receive 2 processes of care (epidural anesthesia and daily epidural management) that have been associated with better patient outcomes.
      • Sanford D.E.
      • Hawkins W.G.
      • Fields R.C.
      Improved peri-operative outcomes with epidural analgesia in patients undergoing a pancreatectomy: a nationwide analysis.
      • Weinbroum A.A.
      Superiority of postoperative epidural over intravenous patient-controlled analgesia in orthopedic oncologic patients.
      However patients treated by low-volume surgeons almost always were significantly more likely to experience transfusions, consultations, and complication-related processes of care (TPN, critical care, and inpatient consultations).
      For example, the need for TPN after any of these procedures indicates that the patient had an unexpected complication or delayed recovery.
      • Braga M.
      • Ljungqvist O.
      • Soeters P.
      • Fearon K.
      • Weimann A.
      • Bozzetti F.
      ESPEN guidelines on parenteral nutrition: surgery.
      These same processes are those that increase the costs of care the most. While these processes are appropriate responses to complications, higher rates of these processes may reflect higher occurrences of adverse events that are not fully captured by the PSI measures we tracked. This lower-quality care may require greater use of costly processes of care, explaining the observed surgeon volume–cost relationship.
      The results suggest 3 potential strategies for cost control that may be undertaken simultaneously. First, steps should be taken to reduce PSI measures of complications, which significantly increase costs. Second, utilization rates of various processes of care may be surrogates for the incidence of non-PSI complications and higher costs. These data are available through administrative and institutional sources. The data could be used to compare outcomes across institutions, provided there was adequate risk adjustment and cautionary notes that there is no evidence that these processes influence the probability of hospital mortality or long-term cancer survival. Third, it may be beneficial to refer patients to high-volume surgeons, because of their enhanced value (higher quality with lower costs). The analysis shows that the use of costly processes associated with adverse events is more prevalent among low-volume surgeons, directly contributing to the high costs of cancer operations.
      There are several caveats to our study. Processes of care and complications may also affect postdischarge costs, but unfortunately we do not have access to postdischarge financial data. Similarly, we do not explore the relation between patient costs and quality, such as cancer recurrence and survival. The use of Medicare claims data limits our analysis to patients aged 65 or older. But given that this demographic comprises over half of new cancer cases each year, this analysis is relevant to a majority of cancer patients.
      • American Cancer Society
      Cancer facts & figures 2014.
      We track processes of care using claims data, which are not as detailed as information that could be obtained through chart abstraction. For example, we were reluctant to adjust for returns to the operating room or percutaneous drain placement. We lacked clinical detail in the claims data to distinguish cases where these processes reflected actions to provide high-quality care versus action to remediate complications.
      • Morton J.M.
      • Brown A.
      • Galanko J.A.
      • Norton J.A.
      • Grimm I.S.
      • Behrns K.E.
      A national comparison of surgical versus percutaneous drainage of pancreatic pseudocysts: 1997-2001.
      • Birkmeyer J.D.
      • Hamby L.S.
      • Birkmeyer C.M.
      • Decker M.V.
      • Karon N.M.
      • Dow R.W.
      Is unplanned return to the operating room a useful quality indicator in general surgery?.
      Similar to past research, we did not consider processes of care specific to individual procedures.
      • Birkmeyer J.D.
      • Sun Y.
      • Goldfaden A.
      • Birkmeyer N.J.O.
      • Stukel T.A.
      Volume and process of care in high-risk cancer surgery.
      Any differences in the performance of procedure-specific processes may therefore be reflected in our estimates of the volume-cost relationship. However, such effects are unlikely to change our conclusion that much of the surgeon volume–cost relationship is explained by higher complication rates among low-volume surgeons that are tracked through remediating processes of care.
      However, given that higher volume cancer providers have been associated with lower mortality,
      • Birkmeyer J.D.
      • Finlayson S.R.G.
      • Tosteson A.N.
      • Sharp S.M.
      • Warshaw A.L.
      • Fisher E.S.
      Effect of hospital volume on in-hospital mortality with pancreaticoduodenectomy.
      • Finlayson E.V.
      • Birkmeyer J.D.
      Effects of hospital volume on life expectancy after selected cancer operations in older adults: a decision analysis.
      • 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.
      • Goodney P.P.
      • Stukel T.A.
      • Lucas F.L.
      • Finlayson E.V.
      • Birkmeyer J.D.
      Hospital volume, length of stay, and readmission rates in high-risk surgery.
      • Schrag D.
      • Cramer L.D.
      • Bach P.B.
      • Cohen A.M.
      • Warren J.L.
      • Begg C.B.
      Influence of hospital procedure volume on outcomes following surgery for colon cancer.
      • Hannan E.L.
      • Racz M.J.
      • Jollis J.G.
      • Peterson E.D.
      Using Medicare claims data to assess provider quality for CABG surgery: does it work well enough?.
      • Hannan E.L.
      • Radzyner M.
      • Rubin D.
      • Dougherty J.
      • Brennan M.F.
      The influence of hospital and surgeon volume on in-hospital mortality for colectomy, gastrectomy, and lung lobectomy in patients with cancer.
      there are multiple overlaps between our study and a recent comparison of cancer mortality in low- versus high-mortality hospitals based on chart review.
      • Revels S.L.
      • Wong S.L.
      • Banerjee M.
      • Yin H.
      • Birkmeyer J.D.
      Differences in perioperative care at low- and high-mortality hospitals with cancer surgery.
      For example, we found negligible differences in the use of pulmonary artery catheters and central venous catheters between low- and high-volume surgeons, and recent analysis found no measurable difference in the use of these processes between low- and high-mortality hospitals.
      • Revels S.L.
      • Wong S.L.
      • Banerjee M.
      • Yin H.
      • Birkmeyer J.D.
      Differences in perioperative care at low- and high-mortality hospitals with cancer surgery.
      The recent analysis found that high-mortality hospitals had significantly lower rates of epidural catheter usage for postoperative pain management; similarly, we found lower rates of daily epidural management for low-volume surgeons for 4 of the 6 cancer operations studied.
      We did not adjust for the use of minimally invasive/robotic surgery in our analyses. These operations were a small (although growing) share of our study sample, which ended in 2009. Studies differ on whether robotic surgery reduces complications or lowers costs relative to open surgery.
      • Yeo H.L.
      • Issacs A.J.
      • Ableson J.S.
      • Milsom J.W.
      • Sedrakyan A.
      Comparison of open, laparoscopic, and robotic colectomies using a large national database: outcomes and trends related to surgery center volume.
      • Ejaz A.
      • Sachs T.
      • He J.
      • Spolverato G.
      • Hirose K.
      • Ahuja N.
      • et al.
      A comparison of open and minimally invasive surgery for hepatic and pancreatic resections using the nationwide inpatient sample.
      Some studies find that hospitals or surgeons performing a higher volume of robotic or minimally invasive operations have fewer complications and lower costs, which may reflect a learning-curve effect.
      • Keller D.S.
      • Hashemi L.
      • Lu M.
      • Delaney C.P.
      Short-term outcomes for robotic colorectal surgery by provider volume.
      • Tran Cao H.S.
      • Lopez N.
      • Chang D.C.
      • Lowy A.M.
      • Bouvet M.
      • Baumgartner J.M.
      • et al.
      Improved perioperative outcomes with minimally invasive distal pancreatectomy.
      • Adam M.A.
      • Thomas S.
      • Youngwirth L.
      • Pappas T.
      • Roman S.A.
      • Sosa J.A.
      Defining a hospital volume threshold for minimally invasive pancreaticoduodenectomy in the United States.
      These findings were noted for both pancreatic operations as well as colorectal operations, which is considered to be less complex.
      These results suggest that as rates of minimally invasive, and potentially robotic, operations increase, the findings in this study are likely to persist: More complications increase patient costs, and the volume-cost relation is likely explained by actions taken by providers to ameliorate the effects of complications. Because we still lack sufficient data regarding the implications of robotic surgery for patient outcomes and costs, particularly for pancreatic resection, this area is worthy of further investigation.
      In conclusion, not all processes of care are closely related to surgeon volume. However, we did find that the processes of care utilized in the postoperative care of patients can explain much of the surgeon volume–cost relationship, primarily through the large cost savings achieved by preventing the need for processes associated with non-PSI adverse events. Given these findings, both process of care utilization and surgeon-volume information may be used to direct patients to safer and lower-cost providers. These data confirm the belief that referral of patients to high-volume surgeons with low complication rates is likely to constrain cost growth and improve patient outcomes.

      Supplementary data

      References

      1. Agency for Healthcare Research and Quality. Total expenses and percent distribution for selected conditions by type of service: United States, 2011. Med Expend Panel Surv Househ Compon Data. Available from: https://meps.ahrq.gov/data_stats/tables_compendia_hh_interactive.jsp?_SERVICE=MEPSSocket0&_PROGRAM=MEPSPGM.TC.SAS&File=HCFY2011&Table=HCFY2011_CNDXP_C&_Debug.

        • American Cancer Society Cancer Action Network
        A national poll facing cancer in the health care system.
        Cancer Action Network, Washington (DC)2010 (Available from:)
        • National Cancer Institute
        Cancer trends progress report 2011-2012.
        National Cancer Institute, Bethesda (MD)2012
        • Sosa J.A.
        • Bowman H.M.
        • Gordon T.A.
        • Bass E.B.
        • Yeo C.J.
        • Lillemoe K.D.
        • et al.
        Importance of hospital volume in the overall management of pancreatic cancer.
        Ann Surg. 1998; 228: 429-438
        • Swisher S.G.
        • Deford L.
        • Merriman K.W.
        • Walsh G.L.
        • Smythe R.
        • Vaporicyan A.
        • et al.
        Effect of operative volume on morbidity, mortality, and hospital use after esophagectomy for cancer.
        J Thorac Cardiovasc Surg. 2000; 119: 1126-1132
        • Kuo E.Y.
        • Chang Y.
        • Wright C.D.
        Impact of hospital volume on clinical and economic outcomes for esophagectomy.
        Ann Thorac Surgery. 2001; 72: 1118-1124
        • Harmon J.W.
        • Tang D.G.
        • Gordon T.A.
        • Bowman H.M.
        • Choti M.A.
        • Kaufman H.S.
        • et al.
        Hospital volume can serve as a surrogate for surgeon volume for achieving excellent outcomes in colorectal resection.
        Ann Surg. 1999; 230: 404-413
        • Konety B.R.
        • Dhawan V.
        • Allareddy V.
        • O’Donnell M.A.
        Association between volume and charges for most frequently performed ambulatory and nonambulatory surgery for bladder cancer. Is more cheaper?.
        J Urol. 2004; 54: 1027-1028
        • Al-Qurayshi Z.
        • Robins R.
        • Hauch A.
        • Randolph G.W.
        • Kandil E.
        Association of surgeon volume with outcomes and cost savings following thyroidectomy.
        JAMA Otolaryngol Neck Surg. 2016; 142: 32-39
        • Damle R.N.
        • Flahive J.M.
        • Davids J.S.
        • Maykel J.A.
        • Sweeney W.B.
        • Sturrock P.R.
        • et al.
        Surgeon volume correlates with reduced mortality and improved quality in the surgical management of diverticulitis.
        J Gastrointest Surg. 2016; 20: 335-342
        • Damle R.N.
        • Macomber C.W.
        • Flahive J.M.
        • Davids J.S.
        • Sweeney W.B.
        • Sturrock P.R.
        • et al.
        Surgeon volume and elective resection for colon cancer: an analysis of outcomes and use of laparoscopy.
        J Am Coll Surg. 2014; 218: 1223-1230
        • Shi H.Y.
        • Chang H.T.
        • Culbertson R.
        • Chen Y.J.
        • Liao Y.C.
        • Hou M.F.
        Breast cancer surgery volume-cost associations: hierarchical linear regression and propensity score matching analysis in a nationwide Taiwan population.
        Surg Oncol. 2013; 22: 178-183
        • Bachmann M.
        • Peters T.
        • Harvey I.
        Costs and concentration of cancer care: evidence for pancreatic, oesophageal and gastric cancers in National Health Service hospitals.
        J Health Serv Res Policy. 2003; 8: 75-82
        • Ho V.
        • Aloia T.
        Hospital volume, surgeon volume, and patient costs for cancer surgery.
        Med Care. 2008; 46: 718-725
        • Nathan H.
        • Atoria C.L.
        • Bach P.B.
        • Elkin E.B.
        Hospital volume, complications, and cost of cancer surgery in the elderly.
        J Clin Oncol. 2015; 33: 107-114
        • Short M.N.
        • Aloia T.A.
        • Ho V.
        The influence of complications on the costs of complex cancer surgery.
        Cancer. 2014; 120: 1035-1041
        • Short M.N.
        • Ho V.
        • Aloia T.A.
        Impact of processes of care aimed at complication reduction on the cost of complex cancer surgery.
        J Surg Oncol. 2015; 112: 610-615
        • Reinhardt U.E.
        The pricing of U.S. hospital services: chaos behind a veil of secrecy.
        Health Aff. 2006; 25: 57-69
        • Meltzer D.
        • Chung J.
        • Basu A.
        Does competition under Medicare Prospective Payment selectively reduce expenditures on high-cost patients?.
        RAND J Econ. 2002; 33: 447-468
        • Cutler D.M.
        • Huckman R.S.
        Technological development and medical productivity: the diffusion of angioplasty in New York state.
        J Health Econ. 2003; 22: 187-217
        • Ho V.
        Learning and the evolution of medical technologies: the diffusion of coronary angioplasty.
        J Health Econ. 2002; 21: 873-885
        • Ho V.
        • Petersen L.A.
        Estimating cost savings from regionalizing cardiac procedures using hospital discharge data.
        Cost Eff Resour Alloc. 2007; 5: 7
        • Healthcare Cost and Utilization Project
        Cost-to-charge ratio files.
        Agency for Healthcare Research and Quality, Rockville (MD)2011 (Available from:)
        • Agency for Healthcare Research and Quality
        Patient Safety Indicators overview.
        Agency for Healthcare Research and Quality, Rockville (MD)2014 (Available from:)
        • Rivard P.E.
        • Luther S.L.
        • Christiansen C.L.
        • Zhao S.
        • Loveland S.
        • Elixhauser A.
        • et al.
        Using patient safety indicators to estimate the impact of potential adverse events on outcomes.
        Med Care Res Rev. 2008; 65: 67-87
        • Romano P.S.
        • Geppert J.J.
        • Davies S.
        • Miller M.R.
        • Elixhauser A.
        • McDonald K.M.
        A national profile of patient safety in U.S. hospitals.
        Health Aff. 2003; 22: 154-166
        • Rubin H.R.
        • Pronovost P.
        • Diette G.B.
        The advantages and disadvantages of process-based measures of health care quality.
        Int J Qual Heal Care. 2001; 13: 469-474
        • Hollenbeck B.K.
        • Wei Y.
        • Birkmeyer J.D.
        Volume, process of care, and operative mortality for cystectomy for bladder cancer.
        Urology. 2007; 69: 871-875
        • Birkmeyer J.D.
        • Sun Y.
        • Goldfaden A.
        • Birkmeyer N.J.O.
        • Stukel T.A.
        Volume and process of care in high-risk cancer surgery.
        Cancer. 2006; 106: 2476-2481
        • Elixhauser A.
        • Steiner C.
        • Harris D.R.
        • Coffey R.M.
        Comorbidity measures for use with administrative data.
        Med Care. 1998; 36: 8-27
        • Birkmeyer J.D.
        • Warshaw A.L.
        • Finlayson S.R.G.
        • Grove M.R.
        • Tosteson A.N.
        Relationship between hospital volume and late survival after pancreaticoduodenectomy.
        Surgery. 1999; 126: 178-183
        • Birkmeyer J.D.
        • Finlayson S.R.G.
        • Tosteson A.N.
        • Sharp S.M.
        • Warshaw A.L.
        • Fisher E.S.
        Effect of hospital volume on in-hospital mortality with pancreaticoduodenectomy.
        Surgery. 1999; 125: 250-256
        • Patti M.G.
        • Corvera C.U.
        • Glasgow R.E.
        • Way L.W.
        A hospital’s annual rate of esophagectomy influences the operative mortality rate.
        J Gastrointest Surg. 1998; 2: 186-192
        • Romano P.S.
        • Mark D.H.
        Patient and hospital characteristics related to in-hospital mortality after lung cancer resection.
        Chest. 1992; 101: 1332-1337
        • Kang C.Y.
        • Halabi W.J.
        • Luo R.
        • Pigazzi A.
        • Nguyen N.T.
        • Stamos M.J.
        Laparoscopic colorectal surgery: a better look into the latest trends.
        Arch Surg. 2012; 147: 724-731
        • Sanford D.E.
        • Hawkins W.G.
        • Fields R.C.
        Improved peri-operative outcomes with epidural analgesia in patients undergoing a pancreatectomy: a nationwide analysis.
        HPB (Oxford). 2015; 17: 551-558
        • Weinbroum A.A.
        Superiority of postoperative epidural over intravenous patient-controlled analgesia in orthopedic oncologic patients.
        Surgery. 2005; 138: 869-876
        • Braga M.
        • Ljungqvist O.
        • Soeters P.
        • Fearon K.
        • Weimann A.
        • Bozzetti F.
        ESPEN guidelines on parenteral nutrition: surgery.
        Clin Nutr. 2009; 28: 378-386
        • American Cancer Society
        Cancer facts & figures 2014.
        American Cancer Society, Atlanta (GA)2014 (Available from:)
        • Morton J.M.
        • Brown A.
        • Galanko J.A.
        • Norton J.A.
        • Grimm I.S.
        • Behrns K.E.
        A national comparison of surgical versus percutaneous drainage of pancreatic pseudocysts: 1997-2001.
        J Gastrointest Surg. 2005; 9: 15-21
        • Birkmeyer J.D.
        • Hamby L.S.
        • Birkmeyer C.M.
        • Decker M.V.
        • Karon N.M.
        • Dow R.W.
        Is unplanned return to the operating room a useful quality indicator in general surgery?.
        Arch Surg. 2001; 136: 405-410
        • Finlayson E.V.
        • Birkmeyer J.D.
        Effects of hospital volume on life expectancy after selected cancer operations in older adults: a decision analysis.
        J Am Coll Surg. 2003; 196: 410-417
        • 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
        • Goodney P.P.
        • Stukel T.A.
        • Lucas F.L.
        • Finlayson E.V.
        • Birkmeyer J.D.
        Hospital volume, length of stay, and readmission rates in high-risk surgery.
        Ann Surg. 2003; 238: 161-167
        • Schrag D.
        • Cramer L.D.
        • Bach P.B.
        • Cohen A.M.
        • Warren J.L.
        • Begg C.B.
        Influence of hospital procedure volume on outcomes following surgery for colon cancer.
        J Am Med Assoc. 2000; 284: 3028-3035
        • Hannan E.L.
        • Racz M.J.
        • Jollis J.G.
        • Peterson E.D.
        Using Medicare claims data to assess provider quality for CABG surgery: does it work well enough?.
        Health Serv Res. 1997; 31: 659-678
        • Hannan E.L.
        • Radzyner M.
        • Rubin D.
        • Dougherty J.
        • Brennan M.F.
        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
        • Revels S.L.
        • Wong S.L.
        • Banerjee M.
        • Yin H.
        • Birkmeyer J.D.
        Differences in perioperative care at low- and high-mortality hospitals with cancer surgery.
        Ann Surg Oncol. 2014; 21: 2129-2135
        • Yeo H.L.
        • Issacs A.J.
        • Ableson J.S.
        • Milsom J.W.
        • Sedrakyan A.
        Comparison of open, laparoscopic, and robotic colectomies using a large national database: outcomes and trends related to surgery center volume.
        Dis Colon Rectum. 2016; 59: 535-542
        • Ejaz A.
        • Sachs T.
        • He J.
        • Spolverato G.
        • Hirose K.
        • Ahuja N.
        • et al.
        A comparison of open and minimally invasive surgery for hepatic and pancreatic resections using the nationwide inpatient sample.
        Surgery. 2014; 156: 538-547
        • Keller D.S.
        • Hashemi L.
        • Lu M.
        • Delaney C.P.
        Short-term outcomes for robotic colorectal surgery by provider volume.
        J Am Coll Surg. 2013; 217: 1063-1069
        • Tran Cao H.S.
        • Lopez N.
        • Chang D.C.
        • Lowy A.M.
        • Bouvet M.
        • Baumgartner J.M.
        • et al.
        Improved perioperative outcomes with minimally invasive distal pancreatectomy.
        JAMA Surg. 2014; 149: 237
        • Adam M.A.
        • Thomas S.
        • Youngwirth L.
        • Pappas T.
        • Roman S.A.
        • Sosa J.A.
        Defining a hospital volume threshold for minimally invasive pancreaticoduodenectomy in the United States.
        JAMA Surg. 2016; ([Epub ahead of print])