Survival statistics gone awry: pancreatic cancer, a case in point.

Resections for pancreatic cancer have been performed for 65 years, with approximately 20,000 reported. A number of authors claim a 5-year survival rate of 30% to 58%. Review of the literature reveals only about 1,200 5-year survivors; however, 10 times as many individual resected survivors have been reported (in various countries), and nonresected survivors are overlooked. This high survival percentage is obtained by reducing the subset on which calculations are based and by using methods such as the Kaplan-Meier method, which produces higher figures as increasing numbers of patients are lost to follow-up. After adjustments, hardly more than 350 resected survivors could be found. Revision of statistical methods is urgently needed.