[Skip to Navigation]
Sign In
Invited Critique
Dec 2011

The Bayes Theorem Wins: Comment on “Impact of Localization Studies and Clinical Scenario in Patients With Hyperparathyroidism Being Evaluated for Reoperative Neck Surgery”

Author Affiliations

Author Affiliation: Department of Surgery, University of California, San Francisco, VA Medical Center, San Francisco, California.

Arch Surg. 2011;146(12):1403. doi:10.1001/archsurg.2011.1001

“Torture numbers and they will confess to anything,” (Gregg Easterbrook, 1999, New Republic)

“There are three kinds of lies: lies, damned lies, and statistics.” (misattributed to Disraeli by Mark Twain, 1906)

Endocrine surgeons are known to be friendly, humble, and thoughtful intellectuals rather than chestthumpers. The main socially acceptable mode of competition among endocrine surgeons is their excellence in surgical outcome. One measure of endocrine surgeons' prowess is their parathyroidectomy success rate. For primary operations, the success rate is expected to be 96% to 99%. Even for reoperations, the success rate is expected to be 90% to 95%. Success rate is a simple fraction with a numerator and a denominator. As surgeons, we try to improve the percentage by having more accurate diagnosis and localization studies (LSs) and better knowledge of anatomy and embryology. Usually, we concentrate on improving the outcome of individual patients (the numerator). Shin et al1 remind us that the denominator is just as important. They analyzed a large series of reoperative parathyroidectomy from Cleveland Clinic, a center of excellence for complex parathyroid surgery. They show that the expected success rate depends on how and when you measure it. For a patient already scheduled for a reoperative parathyroidectomy, the expected success rate is 96%, but for a patient who has an indication for reoperative parathyroidectomy who has not yet undergone LSs, the expected success rate is only 83% because those with negative LSs usually could not undergo parathyroidectomy. Similarly, the PPVs of their LSs depended on whether nonoperated patients were included in the calculation and whether clinical scenarios (CSs) were considered.

Add or change institution