The effect of volume on esophageal cancer resections: What constitutes acceptable resection volumes for centers of excellence?

Objective

Volume–outcome relationships for esophageal cancer resection have been well described with centers of excellence defined by volume. No consensus exists for what constitutes a “high-volume” center. We aim to determine if an objective evidence-based threshold of operative volume associated with improvement in operative outcome for esophageal resections can be defined.

Methods

Retrospective analysis was performed on patients undergoing esophageal resection for cancer in the 1998 to 2005 Nationwide Inpatient Sample. A series of multivariable analyses were performed, changing the resection volume cutoff to account for the range of annual hospital resections. The goodness of fit of each model was compared by pseudo r², the amount of data variance explained by each model.

Results

A total of 4080 patients underwent esophageal resection. The median annual hospital resection volume was 4 (range: 1–34). The mortality rate of “high-volume” centers ranged from 9.94% (≥2 resection/year) to 1.56% (≥30 resections/year). The best model was with an annual hospital resection volume greater than or equal to 15 (3.87% of data variance explained). The difference in goodness of fit between the best model and other models with different volume cutoffs was 0.64%, suggesting that volume explains less than 1% of variance in perioperative death.

Conclusion

Our data do not support the use of volume cutoffs for defining centers of excellence for esophageal cancer resections. Although volume has an incremental impact on mortality, volume alone is insufficient for defining centers of excellence. Volume seems to function as an imperfect surrogate for other variables, which may better define centers of excellence. Additional work is needed to identify these variables.

Abbreviations and Acronyms: AUC, area under the curve, ICD-9, International Classification of Diseases, Ninth Revision, NIS, Nationwide Inpatient Sample

CTSNet classification: 7