Comparative Effectiveness Research: Opportunities in Surgical Oncology

Comparative Effectiveness Research: Opportunities in Surgical Oncology

Comparative Effectiveness Research: Opportunities in Surgical Oncology Heather B. Neuman, MD, MS, and Caprice C. Greenberg, MD, MPH Comparative effect...

167KB Sizes 0 Downloads 79 Views

Comparative Effectiveness Research: Opportunities in Surgical Oncology Heather B. Neuman, MD, MS, and Caprice C. Greenberg, MD, MPH Comparative effectiveness research (CER) is especially applicable to surgical oncology because of the numerous challenges associated with conducting surgical randomized controlled trials, and the opportunity to apply various CER methodologies to answer surgical questions. In this article, several past randomized trials or attempted trials are described to demonstrate challenges related to feasibility, patient selection and generalizability, and timeliness trial results to inform clinical practice. Thus, there is a gap between these “efficacy” studies (ie, randomized trials) and “effectiveness” research, which is performed in a less controlled setting (not randomized) but is able to examine patient outcomes in the “real world.” Retrospective analyses and pragmatic trials are other important methods for answering CER questions in surgical oncology, with examples of these studies being conducted in prostate, breast, and rectal cancers. Multiple current initiatives by the American College of Surgeons and the Alliance for Clinical Trials in Oncology continue to expand the infrastructure for CER in surgical oncology. Semin Radiat Oncol 24:43-48 C 2014 Elsevier Inc. All rights reserved.

omparative effectiveness research (CER) is the “conduct and synthesis of research comparing the benefits and harms of different interventions and strategies to prevent, diagnose, treat and monitor health conditions in ‘real world’ settings.”1 Effectiveness research differs from traditional efficacy trials in that it attempts to account for the patient (ie, comorbidities), physician (ie, surgical experience), and systems factors (ie, volume of a procedure performed at a specific hospital) that affect outcomes observed when surgical interventions move from the highly controlled clinical trial setting into widespread clinical practice. Importantly, CER assesses outcomes that are directly relevant to patients, extending beyond the typical mortality data to include other patientcentered outcomes such as bowel function after proctectomy or lymphedema after lymph node dissections. Finally, CER describes results at the population as well as at the subgroup level to directly inform what may be considered “best practice” for a specific individual.2 CER is especially applicable to surgery both because of the numerous challenges associated with conducting surgical

C

Department of Surgery, UW Madison School of Medicine and Public Health, Madison, WI. The authors declare no conflict of interest. Address reprint requests to Heather B. Neuman, MD, Department of Surgery, UW Madison School of Medicine and Public Health, Health Sciences Learning Center, H4/726 CSC, 600 Highland Ave, Madison, WI 537927375. E-mail: [email protected]

1053-4296/13/$-see front matter & 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.semradonc.2013.09.003

randomized controlled trials (RCTs) as well as the appropriateness of applying various CER methodologies to surgical questions. In this review, we briefly discuss reasons for the gap between efficacy and effectiveness in surgery, describe in more detail why surgery is so well suited for CER, examine challenges in applying CER to surgical oncology, and describe current national CER initiatives in surgical oncology.

The Gap Between Efficacy and Effectiveness in Surgery Conducting traditional RCTs in surgery is challenging. Although RCTs are considered the gold standard in testing the efficacy of new treatment paradigms, RCTs have not been universally employed to test new surgical interventions. The field of surgery has less regulation and oversight than other aspects of medicine (such as drug interventions) and many new surgical treatments are introduced directly into clinical practice without prospective trials. Consequently, surgical treatments are less likely than medical therapies to have RCT evidence supporting practice.3 An example of how surgical techniques can become readily disseminated into clinical practice without supporting RCT data is that of the sentinel lymph node (SLN) biopsy. SLN for breast cancer was initially reported in two small single-institution studies.4,5 A RCT was initiated in 1999 to validate the efficacy of SLN biopsy by randomizing patients to SLN biopsy followed by immediate 43

H.B. Neuman and C.C. Greenberg

44 axillary lymph node dissection vs SLN biopsy and axillary lymph node dissection only if the SLN had metastatic disease; survival outcomes for this trial were reported in 2010 and confirmed the accuracy and safety of SLN biopsy for breast cancer.6 However, in the intervening time, the SLN concept was quickly and widely disseminated into clinical practice. Even before the results of National Surgical Adjuvant Breast and Bowel Project-B32 being reported in 2010, SLN biopsy was considered the standard of care in the management of the clinically node negative axilla,7 with SLN being performed in 59% of patients with early stage breast cancer across the country in 2004.8 Numerous other such examples exist and represent a challenge when considering RCTs in surgery. Randomization itself is difficult with surgical interventions and can impair the ability to complete surgical trials. Both patients and surgeons bring personal preferences about the optimal management of cancer to treatment discussions. In general, patients dislike the idea of random allocation to a treatment arm, especially if 1 of the randomization arms includes a less invasive surgical or nonoperative arm.9,10 This is highlighted in the experience of the surgical prostatectomy vs interstitial radiation intervention trial.11 In response to slow accrual, the investigators implemented a multidisciplinary educational intervention for all eligible patients to provide a balanced, comprehensive view of the treatment options being examined in the trial. An increase in accrual was achieved, with patients advocating an increased desire to be randomized after more fully understanding the treatment options and the true equipoise between them.11 However, despite this very intensive effort of the investigators to overcome challenges with randomizing patients between an operative and nonoperative intervention, the trial was closed owing to low accrual. Like patients, surgeons have a difficult time accepting true equipoise between treatments options without the unconscious belief that 1 option is preferable for their patients.12 This has significant implications for trial accrual, as surgeons are unlikely to recommend a clinical trial to a patient if they already have a preconceived notion of what is the “right” treatment option for them. Given this difficulty, inclusion of a placebo arm in the form of a sham operation is all but impossible with a few notable exceptions.13 In addition to affecting accrual, the perspectives of both patients and surgeons have the potential to introduce significant bias into a clinical trial, which may then affect the generalizability of trial results. Surgeons select those patients for clinical trial participation who they think may be best suited for trial; this is not always aligned with trial eligibility. As an example, the Cancer and Leukemia Group B C9343 randomized older women (470 years of age) with early stage breast cancer to lumpectomy with tamoxifen and radiation therapy vs lumpectomy with tamoxifen alone.14 When the trial was initiated in 1994, the eligibility criteria included women with tumors up to 4 cm in size and did not specify estrogen receptor status. These eligibility criteria were revised in 1996 in an attempt to “broaden participation by physicians concerned about the upper size limit” to include only women with tumors o2 cm in size and excluding women known to be estrogen receptor negative. The trial was ultimately completed

demonstrating that lumpectomy with tamoxifen only may be a reasonable option for older patients with breast cancer. It is noteworthy that only 14 of the 636 women enrolled in the trial had a tumor 42 cm. This speaks to the importance of physician perceptions of equipoise for successful trial accrual but also highlights how physician perceptions can introduce potential bias and limit the generalizability of results. These challenges to conducting surgical RCTs contribute directly to the observed gap between outcomes within a clinical trial setting and outcomes in “the real world.” In addition, the rigorous controls for confounding and the strict eligibility requirements employed in RCTs play a role. Often, older patients or those with significant comorbidities are excluded from clinical trials. These factors heavily influence surgical decision making in “real-world” situations and can have a significant effect on the clinical outcomes observed. As surgeons attempt to apply clinical trial data to a broader patient population (including subgroups excluded from clinical trials), the outcomes will inevitably be affected. This is particularly germane in surgery where comorbidity and operative risk (often exclusion criteria in efficacy trials) play a critical role in decision making. Additionally, clinical trials have comprehensive quality control built into them to ensure that the only factor that changes between patients is the intervention being tested. For surgical trials, this usually includes quality control for the surgery itself, with many studies requiring surgeons to be credentialed before trial participation. The approach to credentialing can be very simple, such as requiring participating surgeons to have performed a certain number of the procedures. However, for more technically challenging procedures, the requirements for surgeons may be more extensive. As an example, the clinical outcomes of surgical therapy (COST) study randomized patients with colon cancer to laparoscopic vs open resection.15 Because laparoscopic colectomy is a procedure with a known learning curve, the trial required participating surgeons to video record and submit performance of at least 20 laparoscopic colon cases before trial participation to demonstrate that adequate oncologic technique was performed. The investigators hoped that this credentialing process would minimize between-surgeon variation in the quality of surgery performed and this hypothesis was confirmed in a subsequent subset analysis comparing individual surgeon outcomes by volume.16 However, a “credentialing process” does not exist for surgical procedures in “real-world” clinical practice and it is uncertain whether the outcomes observed in clinical trials would translate into general practice. Similarly, differences between hospital structure and processes of care between a clinical trial setting and “real world” may influence the outcomes observed.

Surgery and CER Surgery is ideally suited for CER, including both retrospective analyses and prospective pragmatic clinical trials. Unlike interventions in other disciplines (such as receipt of a pharmaceutical agent), surgical events tend to be discrete and easily identifiable. For every surgical intervention, a

Opportunities in surgical oncology distinct date of service (and subsequent surgical bill) exists to identify the date of the operation, thus allowing the intervention of interest to be readily identified in administrative databases. Further, surgical interventions have meaningful short-term outcomes that can facilitate both prospective and retrospective studies. Both in-hospital end points, such as reoperation and postoperative infections, as well as 30-day outcomes are relevant and facilitate the direct comparison of 2 different surgical procedures in an extremely efficient fashion without the need for extended follow-up. Importantly, these short-term outcomes can also be identified in many administrative data sets and can be readily linked back to the intervention of interest. Additionally, there are other rich data sources available to assess surgical outcomes, such as the National Surgical Quality Improvement Project (NSQIP), a quality improvement system which provides risk-adjusted outcomes data, focusing on 30-day outcomes including complications, mortality, and reoperation, to participating hospitals.17,18 Some surgical questions do not lend themselves to being evaluated in an efficacy RCT but are very appropriate for a retrospective CER approach. To date, only 40% of questions related to surgical interventions could have been evaluated by a RCT.19 Surgical techniques in the era of robotic and laparoscopic surgery are quickly evolving. Conducting RCTs to test each new technique or application of each new technique separately before adoption into practice is not feasible or practical in the light of the pervasiveness and effectiveness of direct-to-consumer advertising and subsequent patient demand for less invasive approaches. In such circumstances, CER allows for the efficient assessment and comparison of new technologies and procedures to the current standard of care to provide the “best evidence” for or against new techniques. Other surgical questions are not amenable to randomization for ethical reasons, such as randomizing women with a BRCA mutation to risk-reducing mastectomy vs enhanced screening with breast MRI studies. CER is the only reasonable means of assessing outcomes between management strategies for highly preference-sensitive surgical decisions like these and provides important data to support patients and surgeons in decision making. Retrospective CER may also be beneficial in generating preliminary data to establish equipoise between treatment options, enhancing accrual to a subsequent surgical RCT. Given the importance of surgeon and patient buy-in for the successful completion of a clinical trial, robust population level supporting data may facilitate trial accrual. A relevant example is completion lymph node dissections for melanoma patients with SLN metastases. Although completion nodal dissection is considered the standard of care for these patients,20 a multicenter RCT (the Multicenter Selective Lymphadenectomy Trial II) is currently accruing patients to test the question of whether a completion lymph node dissection is required for all patients.21 Before trial initiation, numerous retrospective studies, including a multicenter study, examined this question and found similar disease-free and melanoma-specific survival between patients with SLN metastases who did and did not undergo a completion lymph node dissection. Although these

45 retrospective studies are clearly open to selection bias, observational data like these support the underlying rationale of the Multicenter Selective Lymphadenectomy Trial-II study and ongoing accrual. Although CER would likely not be enough to establish a nonoperative approach to the management of axillary metastases for melanoma, it has helped to garner enough interest for fairly brisk trial accrual. Finally, CER can be used to examine how findings from a surgical RCT translate into everyday clinical practice. We previously discussed the limitations of surgical RCT and how these limitations make the outcomes uncertain when surgical interventions are implemented outside a clinical trial. Observational studies comparing outcomes of 2 different surgical interventions in real-world settings are an effective way to examine how surgical techniques and procedures are implemented and disseminated. For example, the dissemination of laparoscopic colectomy after publication of the COST trial15 was studied by examining the Commission on Cancer's National Cancer Data Base.22 Overall outcomes after laparoscopic colectomy in this population sample were good, with equivalent local recurrence rates and improved survival compared with an open approach. However, when comparing survival of patients identified as having undergone laparoscopic colectomy in the National Cancer Data Base to those in the COST trial, significantly lower 5-year overall survival rates were observed (64.1% vs 76.4%). This again highlights the gap between the artificial clinical trial setting and general clinical practice and stresses the importance of using CER to examine surgical procedures in real-world settings.

Examples of CER in Surgical Oncology The field of Surgical Oncology is especially appropriate for the application of CER. Cancer registries such as individual state tumor registries, the national Surveillance Epidemiology and End Results program and the Commission on Cancer's National Cancer Database, all provide incredibly rich data for examining outcomes after cancer surgery. These databases can be linked to administrative claims data that supplement the registry data by providing patient-, provider-, and system-level information in a longitudinal fashion. These linked registry data provide not only the short-term outcomes relevant for cancer surgery but also long-term outcomes; these outcomes are often not feasible to capture through RCTs owing to relatively low event rates and extended disease-free intervals mandating long periods for follow-up. A number of examples of CER in surgical oncology exist in the literature that are illustrative of the range of questions CER can address. A few examples representing the diverse methodological approaches relevant to CER are described ahead in more detail to demonstrate how CER can be applied in surgical oncology. (1) Management of rectal cancer for patients with a clinical complete response to neoadjuvant chemoradiation therapy. Increasing evidence suggests that up to 40%

H.B. Neuman and C.C. Greenberg

46 of rectal cancer patients treated with neoadjuvant chemoradiation therapy have a pathologic complete response.23 However, it is clinically challenging to identify patients with no residual disease who can safely be managed nonoperatively with observation. This question was addressed in a decision-analysis study to model the options of surgery vs observation for rectal cancer patients with a clinical complete response after neoadjuvant chemoradiation, accounting for the decreased quality of life associated with a low anterior resection or abdominal-perineal resection with its requisite permanent stoma and the increased risk of recurrence if residual disease remains. This study demonstrated that with the current limitations in clinicians' ability to accurately assess patients' response to neoadjuvant chemoradiation, proctectomy is the preferred option for most patients.23 A number of active National Institutes of Health-sponsored trials are extending this work and currently investigating optimal methods of assessing response to neoadjuvant therapy. (2) Surgical management of ductal carcinoma in situ. In a comparative effectiveness analysis of the Surveillance Epidemiology and End Results-Medicare database, outcomes after lumpectomy vs lumpectomy plus radiation for women with ductal carcinoma in situ were examined. Outcomes assessed included risk of a subsequent ipsilateral breast event or subsequent mastectomy or both. They determined that radiation therapy reduced all outcomes. Further, it was estimated that for healthy women aged 66-79 years, 11 patients with high-risk disease and 15-16 patients with low-risk disease would need to be treated with radiation therapy to prevent a breast cancer–related event at 5 years.24 (3) Laparoscopic vs robotic prostatectomy for patients with prostate cancer. Although open radical prostatectomy is the most common treatment for patients with prostate cancer, there has been a rapid shift to less invasive methods, first with laparoscopy and more recently with robotic surgery.25 It is unlikely that a clinical trial will ever be designed to assess head-to-head laparoscopic vs robotic prostatectomy. However, this question was examined with a systematic review, including economic modeling, to study the relative clinical benefit and costeffectiveness of each procedure.26 This analysis suggested that outcomes (including the need for blood transfusion, organ injury rates, and negative margins) were generally better for robotic surgery. Cost associated with the robotic procedure was high largely because of capital equipment costs, but cost per patient was reduced as the surgical volume increased. Although very few studies included in this systematic review were considered to be at low risk of bias, this may be the most definitive evaluation of the laparoscopic vs robotic prostatectomy question. Although numerous examples of how CER can be successfully applied to critical surgical oncology questions exist, challenges to expanding CER in surgical oncology remain.

The challenges with patient and physician bias limiting willingness to participate in randomized trials exist for pragmatic or cluster randomized trials just as for efficacy trials. For this reason, prospective randomized CER remains limited in surgery and needs to be addressed as this field moves forward. In the meantime, prospective observational studies and retrospective analyses can answer many important questions. Many challenges for these study designs relate to assessment of the specific outcomes most relevant after cancer surgery. First, owing to the complicated nature of many cancer surgeries, the traditional in-hospital or 30-day outcomes described in noncancer surgical studies may not capture all relevant events. For example, for pancreatic surgery, postoperative complications and readmissions occur as long as 90 days after surgery and would be missed if only events within a 30-day window were collected.27 Further, many of the 30-day outcomes captured within databases such as the NSQIP focus on general end points, such as urinary tract infection, surgical site infection, or pneumonia. Although these outcomes are critical for improving the quality of overall surgical care, the lack of procedure-specific complications (pancreatic duct leak, colon anastomotic leak, and seroma after lymph node dissection) make it challenging to compare 2 different surgical approaches to an oncologic problem in a meaningful way using traditional data sources. Finally, some surgical morbidity (such as lymphedema) after oncologic procedures may not present itself in the “perioperative” period at all and requires longer-term follow-up to comprehensively capture all affected patients. In addition, other patient-centered outcomes that are not routinely captured by quality improvement databases or cancer registries are becoming increasingly important as the population of cancer survivors grows. Functional outcomes after surgery, such as sexual or urinary function after prostatectomy or bowel function after proctectomy, can significantly affect patient's quality of life. As mortality and recurrence after surgical procedures improves, comprehensively examining these patient-reported outcomes is critical to improving the quality of surgical oncologic care. Beginning to develop infrastructure to routinely collect this data is a challenge. Finally, although questions surrounding morbidity are critical, if we are truly going to address the effectiveness of specific surgical interventions for cancer, outcomes such as local recurrence (arguably the most critical when assessing cancer surgical techniques) need to be evaluated. Although cancer registries attempt to capture local recurrence in some circumstances, the quality of these data is variable and significantly limits the ability to use currently available data to examine local recurrence.

Initiatives to Advance CER in Surgical Oncology Although significant challenges to moving CER forward in the field of surgical oncology exist, the surgical oncology community has several new initiatives to develop infrastructure and advance the field. The importance of surgical oncologists in

Opportunities in surgical oncology developing the landscape for CER in cancer has been recognized by the American College of Surgeons and has led to the development of several new programs. The first major initiative will promote the collection of surgical oncology–specific outcomes using the American College of Surgeons NSQIP.17 Until recently, NSQIP sampled a proportion of all procedures performed at an institution to generate hospital-specific performance reports and limited the number of any 1 type of procedure sampled at a hospital. However, the program has recently evolved to allow an option for “Procedure Targeted” reporting.28 If hospitals choose this option, they may choose to sample and collect data on 100% of patients who undergo any of 30 specific high-risk procedures across 9 specialties including oncology. Targeted procedures include colectomy, pancreatectomy, and esophagectomy. As a part of this program, procedure-specific as well as more general variables and outcomes will be collected. For example, patients undergoing pancreatectomy will have data on preoperative variables such as jaundice, intraoperative variables such as vascular resection, and outcomes such as pancreatic leak collected. As this procedure-targeted reporting continues to evolve, it will greatly enhance our ability to use this data resource for surgical oncology CER.28,29 Additionally, the American College of Surgeons has recently developed the Oncology NSQIP National Cancer Institute Center Consortium. The consortium consists of 51 (of 60) National Cancer Institute-accredited hospitals that currently participate in the NSQIP. The goals of this consortium are to compare oncology-specific data among similar centers and improve and expand risk adjustment, process measures, and short-term outcomes currently utilized by the NSQIP. It is anticipated that this consortium will be a leader in future surgical oncology quality improvement initiatives and will provide the infrastructure and resources required for both prospective and retrospective CER.29 The second major initiative involves the Alliance for Clinical Trials in Oncology, formed in 2011 after the merger of the American College of Surgeons Oncology Group, the Cancer and Leukemia Group B, and the North Central Cancer Treatment Group.30 Although the infrastructure of the Clinical Trials Cooperative Group Program was built to support efficacy trials, it can be leveraged to facilitate retrospective and prospective CER. The cooperative groups have collected detailed, quality-checked data on patient characteristics, pathology results, treatments received, and long-term outcomes. These data include reliable longitudinal tracking of outcomes such as recurrence and toxicity that are not available in any other setting but are crucial to CER studies of cancer care. In addition, the cooperative groups have built infrastructure to facilitate collaboration of patients, providers, policymakers, researchers, and federal partners as stakeholders throughout the scientific process, another critical component of this area of research. This potential to develop CER was recognized by the leadership as the Alliance was formed and a joint initiative between the Cancer Control Program and the American College of Surgeons Clinical Research Program (ACS-CRP) was developed. Given our focus on surgical oncology, we

47 concentrate on the ACS-CRP.31 Through the ACS-CRP, the multidisciplinary collaboration and research infrastructure of the Alliance and its partners will be expanded to allow increased research activities in health services, comparative effectiveness, and patient-centered outcomes research while also providing a vehicle for dissemination and implementation of new knowledge. A critical partner in this initiative is the Commission on Cancer (CoC) of the ACS.31 In addition to the National Cancer Data Base (discussed previously), the CoC brings a network of 1500 hospitals dedicated to improving the quality of cancer care and the infrastructure and motivation to participate in research. Through these resources, the CoC offers important opportunities to develop and disseminate new approaches to cancer care.32 As a result, the partnership between the Alliance and the CoC provides a unique and valuable opportunity to perform activities across the spectrum of CER and Patient-Centered Outcomes Research, from analysis of existing data and the conduct of prospective trials in a real-world setting to dissemination and implementation of research and immediate practice of change.

References 1. Services USDoHaH. Federal Coordinating Council for Comparative Effectivness Research: Report to the President and the Congress; 2009 2. Medicine IIo. Initial National Priorities for Comparative Effectiveness Research. Washington, DC; 2009 3. Farrokhyar F, Karanicolas PJ, Thoma A, et al: Randomized controlled trials of surgical interventions. Ann Surg 251:409-416, 2010 4. Krag DN, Weaver DL, Alex JC, et al: Surgical resection and radiolocalization of the sentinel lymph node in breast cancer using a gamma probe. Surg Oncol 2:335-340, 1993 5. Giuliano AE, Kirgan DM, Guenther JM, et al: Lymphatic mapping and sentinel lymphadenectomy for breast cancer. Ann Surg 220:391-401, 1994 6. Krag DN, Anderson SJ, Julian TB, et al: Sentinel-lymph-node resection compared with conventional axillary-lymph-node dissection in clinically node-negative patients with breast cancer: Overall survival findings from the NSABP B-32 randomised phase 3 trial. Lancet Oncol 11:927-933, 2010 7. Lyman GH, Giuliano AE, Somerfield MR, et al: American Society of Clinical Oncology guideline recommendations for sentinel lymph node biopsy in early-stage breast cancer. J Clin Oncol 23:7703-7720, 2005 8. Rescigno J, Zampell JC, Axelrod D: Patterns of axillary surgical care for breast cancer in the era of sentinel lymph node biopsy. Ann Surg Oncol 16:687-696, 2009 9. Solomon MJ, Pager CK, Young JM, et al: Patient entry into randomized controlled trials of colorectal cancer treatment: Factors influencing participation. Surgery 133:608-613, 2003 10. Gattellari M, Ward JE, Solomon MJ: Randomized controlled trials in surgery: Percieved barriers and attitudes of Australian Colorectal Surgeons. Dis Colon Rectum 44:1413-1420, 2001 11. Wallace K, Fleshner N, Jewett M, et al: Impact of a multi-disciplinary patient education session on accrual to a difficult clinical trial: The Toronto experience with the surgical prostatectomy versus interstitial radiation intervention trial. J Clin Oncol 24:4158-4162, 2006 12. Kaafarani HM: Surgeon preference and variation of surgical care. Am J Surg 201:709-711, 2011 13. Moseley JB, O’Malley K, Petersen NJ, et al: A controlled trial of arthroscopic surgery for osteoarthritis of the knee. N Engl J Med 347:81-88, 2002 14. Hughes KS, Schnaper LA, Berry D, et al: Lumpectomy plus tamoxifen with or without irradiation in women 70 years of age or older with early breast cancer. N Engl J Med 351:971-977, 2004

48 15. Fleshman J, Sargent DJ, Green E, et al: Laparoscopic colectomy for cancer is not inferior to open surgery based on 5-year data from the COST Study Group trial. Ann Surg 246:655-662, 2007; ([discussion 62-4]) 16. Larson DW, Marcello PW, Larach SW, et al: Surgeon volume does not predict outcomes in the setting of technical credentialing: results from a randomized trial in colon cancer. Ann Surg 248:746-750, 2008 17. American College of Surgeons National Surgical Quality Improvement Program, Available at: 〈http://site.acsnsqip.org/)〉. Accessed April 30, 2013 18. Ingraham AM, Richards KE, Hall BL, et al: Quality improvement in surgery: The American College of Surgeons National Surgical Quality Improvement Program approach. Adv Surg 44:251-267, 2010 19. Solomon MJ, McLeod RS: Should we be performing more randomized controlled trials evaluating surgical operations? Surgery 118:459-467, 1995 20. National Comprehensive Cancer Network Clinical Practice Guidelines in Oncology—Melanoma, 2010. Available at: 〈http://www.nccn.org/profes sionals/physician_gls/pdf/melanoma.pdf〉. Accessed April 24, 2013 21. Morton DL: Overview and update of the phase III Multicenter Selective Lymphadenectomy Trials (MSLT-I and MSLT-II) in melanoma. Clin Exp Metastasis 29:699-706, 2012 22. Bilimoria KY, Bentrem DJ, Nelson H, et al: Use and outcomes of laparoscopic-assisted colectomy for cancer in the united states. Arch Surg 143:832-840, 2008 23. Neuman HB, Elkin EB, Guillem JG, et al: Treatment for patients with rectal cancer and a clinical complete response to neoadjuvant therapy: A decision analysis. Dis Colon Rectum 52:863-871, 2009

H.B. Neuman and C.C. Greenberg 24. Smith BD, Haffty BG, Buchholz TA, et al: Effectiveness of radiation therapy in older women with ductal carcinoma in situ. J Natl Cancer Inst 98:1302-1310, 2006 25. Hu JC, Hevelone ND, Ferreira MD, et al: Patterns of care for radical prostatectomy in the United States from 2003 to 2005. J Urol 180: 1969-1974, 2008 26. Ramsay C, Pickard R, Robertson C, et al: Systematic review and economic modelling of the relative clinical benefit and cost-effectiveness of laparoscopic surgery and robotic surgery for removal of the prostate in men with localised prostate cancer. Health Technol Assess:16, 2012 27. Ahmad SA, Edwards MJ, Sutton JM, et al: Factors influencing readmission after pancreaticoduodenectomy: A multi-institutional study of 1302 patients. Ann Surg 256:529-537, 2012 28. A.C.S. NSQIP Procedure Targeted. Available at: 〈http://site.acsnsqip.org/ program-specifics/program-options/procedure-targeted-program/〉; Accessed April 23, 2013 29. Merkow RP, Bilimoria KY: Currently available quality improvement initiatives in surgical oncology. Surg Oncol Clin North Am 21:367-375, 2012; (vii) 30. Bertagnolli MM: New Alliance formed to address clinical trials in oncology. Bull Am Coll Surg 97:45-47, 2012 31. Nelson H: Introducing the ACS Clinical Research Program of the Alliance. Bull Am Coll Surg 97:40-41, 2012 32. Commission on Cancer. Cancer Program Standards 2012: Ensuring Patient-Centered Care. Version 1.1, in American College of Surgeons (ed), 2012