[Skip to Content]
[Skip to Content Landing]
January 2010

Centers of Excellence—Reply

Arch Surg. 2010;145(1):107-108. doi:10.1001/archsurg.2009.257

In reply

Mr Pratt is critical of my use of statistical analysis and states that SRC statisticians as well as those at the University of North Carolina reviewed my analyses and found them to be flawed. Mr Pratt fails to name these statisticians and none of them coauthored his letter. His criticism centers on the fact that I found a statistically significant relationship between procedure volume and outcomes and concluded that volume was not a clinically important predictor of outcomes. I readily acknowledge that there was a statistically significant P value relating procedure volume to complication rates. Nonetheless, P values alone do not tell the entire story. Accepting a P value as the exclusive evidence for acceptance of scientific data is akin to deciding a patient is dead because his or her eyes are closed. Closed eyes are associated with death but other factors must be considered before drawing the conclusion. The reason that a statistically significant volume-outcome relationship can be of little practical importance was summarized in detail in prior publications on the topic.1,2

First Page Preview View Large
First page PDF preview
First page PDF preview