Robert A. Meguid, Eric S. Weiss, David C. Chang, Malcolm V. Brock, Steven C. Yang; Surgery, Johns Hopkins University, School of Medicine, Baltimore, MD
Objective: Volume-outcome relationships for esophageal cancer resection have been well described with centers of excellence defined by volume. No consensus exists, however, for what constitutes a "high volume" center. We aim to define an objective evidence-based threshold of operative volume associated with improvement in operative outcome for esophageal resections.Methods: Retrospective analysis was performed on patients undergoing esophageal resection (esophagectomy ICD-9-CM 42.4X) for esophageal cancer (ICD-9 150.X) in the 1998-2003 Nationwide Inpatient Sample file (48 million records). Multivariate analysis (MVA) was performed with perioperative death as outcome, and resection volume, age, gender, race, procedure yr, Charlson Comorbidity Index, and academic medical center status as independent variables. Different models of "high" volume (>=1 to >=29 resection/yr) were substituted in the MVA. The goodness-of-fit of each model was compared by pseudo r2, a measurement of the amount of data explained by the model.
Results: The study population consisted of 3,052 patients (653, 21% female, mean age 64yrs). Median annual institution resection volume was 4 (range:1-29, mean 6.4). The overall mortality rate was 9.9%. The mortality rate of "high volume" centers ranged from 9.9% (>=1 resection/yr) to 5.4% (>=10resection/yr). The best model of a "high volume" center was with annual institution resection volume >=10 with a pseudo r2 of 0.04, (4% of variance in the data explained) and the lowest mortality rate of 5.4%. However, there was very little difference in percent variance in the data explained between the best model and other models with different volume cutoffs (range 3.3%-4.0%). In comparison, the model without any volume variable yielded a pseudo r2 of 0.033, or 3.3% of variance in the data explained. This suggests that the volume explained <1% of variance in perioperative death.
Conclusion: The best model for defining a "high volume" center for esophageal resection was an annual institution resection volume >=10. However, little differences exist in the explanatory powers of other models of "high volume" centers. Although volume has an important impact on mortality, volume cutoff alone is insufficient for defining centers of excellence. We suspect that other variables which better define centers of excellence need to be evaluated in future studies.
Back to Annual Meeting
Back to Program Outline
