Not Just Another Article on Cancer Surgery Volume and Patient Outcomes

I imagine that you are thinking at this time that the article by Bilimoria et al in this issue of Journal of Clinical Oncology is just another article confirming that, for many of the complex cancer operations performed in the United States, high-volume centers have better perioperative mortality and long-term survival rates compared with low-volume hospitals. However, Bilimoria et al look at this issue from a different perspective. Realizing that it is unclear whether quality initiatives in surgical oncology should focus on factors affecting perioperative mortality or long-term survival, the authors’ objective was to determine whether differences in hospital surgical volume have a larger affect on perioperative mortality or on long-term survival. Several points need to be emphasized before we dissect the present article. No one will deny that surgeon and hospital volumes in conjunction with surgical specialty training may indeed influence patient outcome. However, the processes of care and the surgeon as a prognostic factor are rarely brought into the discussion. Interest in a specific field by a surgeon results in further specialization and additional improvement in knowledge and skills. Those surgeons with a specific training in a given specialty generally make greater use of the full range of treatment and diagnostic options available within a multidisciplinary team setting. Specialist training must eventually reduce or indeed prevent extreme variation in surgical outcomes. When it comes to the discussion of case volume, it is important to remember that there are a variety of hospital factors, such as the types of services and technology, multidisciplinary teams, internal quality programs, and American Cancer Society or National Cancer Institute designation that have all been suggested to be associated with superior cancer care. The problem is that the literature rarely considers these factors. Certainly, one of the reasons can be that they are too detailed to abstract. Now let us turn our attention to the article by Bilimoria et al entitled, “Directing Surgical Quality Improvement Initiatives: Comparison of Perioperative Mortality and Long Term Survival for Cancer Surgery.” Using the National Cancer Data Base, outcomes were evaluated for 243,103 patients undergoing one of seven complex cancer resections with well-documented volume outcome relationships. The patients were identified who underwent surgery for nonmetastatic colon, esophageal, gastric, liver, lung, pancreatic, or rectal cancer. (I do not consider colon cancer surgery complex, but that is a minor issue.) The objective of the study was to determine whether differences in hospital surgical volume have a larger effect on perioperative mortality or on long-term survival. Differences between high-volume and lowvolume hospitals were compared for perioperative mortality and long-term survival using two methods. First, multivariable Cox proportional hazards modeling was used to directly compare differences in outcomes by hospital volume at 60 days and 5 years. Second, the authors estimated the number of potentially avoidable deaths at 60 days and 5 years if outcomes at low-volume hospitals were improved to the level of high-volume hospitals. The authors make a good point. That is, if we have a better understanding of whether hospital volume, as a surrogate for processes of care and structural features, has a larger impact on perioperative mortality or long-term survival, this will help direct quality improvement efforts for cancer surgery. The authors, indeed, are the first to directly compare perioperative and long-term outcomes from multiple complex cancer surgeries. Results of this study suggest that quality improvement initiatives in surgical oncology should not only address factors affecting perioperative mortality, but there should also be considerable focus on identifying factors impacting long-term outcomes. It is important to note that only patients surviving beyond 60 days after the index surgery were included in the long-term conditional survival Cox analysis. To allow direct comparison of perioperative mortality and long-term survival analysis, the same covariates were used in the Cox models for both perioperative mortality and 5-year conditional survival analysis. The regression models accounted for clustering of outcomes within hospitals using robust variance estimates. However, this did not qualitatively affect the results. This is important because clustering refers to the fact that outcomes of one provider’s patients will tend to be more like one another than like the outcomes of a different provider’s patients. When this is the case, studies must include many more patients to detect real differences in the care of highand low-volume providers. Hence, although there are studies that suggest surgical patients fare best with hospitals and surgeons that perform a high volume of procedures, most have not accounted for the tendency of patients of one provider to have similar outcomes, or clustering. I also congratulate the authors for recognizing the limitations of their research. They admit that the National Cancer Data Base does not collect surgeon information, an important issue in view of the fact that a surgeon can be considered a prognostic factor. The authors counteract this limitation by stating that quality initiatives to date, particularly from the American College of Surgeons, the American Society of Clinical Oncology, the National Comprehensive Cancer Network, and the National Quality Forum, are primarily focused on assessing hospital-level performance, rather than individual surgeon performance. However, that does not make me feel any better. One limitation admitted by the authors is that long-term outcomes were JOURNAL OF CLINICAL ONCOLOGY E D I T O R I A L VOLUME 26 NUMBER 28 OCTOBER 1 2008