The Survival of Survival Curves

Survival analysis, or Kaplan-Meier analysis, is a powerful tool that is critical to randomized clinical trials. It allows one to determine how often patients within a group experience a particular endpoint, despite enrollment of patients at variable times, follow-up lasting for variable durations, and loss of patients before they reach an endpoint. Data from patients who are lost to follow-up, or who exit a study before completion, can still be used after they are censored. Two Kaplan-Meier curves can be compared using a log-rank test to determine if one group of patients fared better or worse. Kaplan-Meier analysis is traditionally used to compare survival rates between treatment groups in medicine, but can also be used to compare the incidence of other health-related endpoints such as a specific complication, a stroke, or a combined clinical endpoint. It is used in other fields as well, such as engineering and economics. It is difficult to open any medical journal today and not run into a Kaplan-Meier survival curve. Despite this, however, very few people know who Kaplan and Meier were.

On November 12, 2012, the University of Chicago Departments of Statistics and Health Studies held its Paul Meier Memorial Lecture, to honor the “Meier” of “Kaplan and Meier.” Dr. Paul Meier died on August 7, 2011, at age 87. The lecture was an important annual reminder of the contributions of Dr. Meier, who along with Edward L. Kaplan, introduced the Kaplan-Meier estimator in 1958.¹ At that time, Dr. Meier was studying cancer at Johns Hopkins University, and Dr. Kaplan was working at Bell Labs on a better way to measure the survival of vacuum tubes. Dr. Kaplan died in 2006. Their publication is said to be one of the most frequently cited of all research papers ever. In 1945, Dr. Meier received his bachelor’s degree in physics and mathematics from Oberlin, and in 1957, he joined the faculty at the University of Chicago, where he became Chairman of the Statistics Department. He was a strong advocate of the use of randomized clinical trials in medicine at a time when medical decision making was largely based on anecdotal experiences and biased, nonrandomized studies. He helped convince the national regulatory health agencies of the importance of randomized trials when making decisions related to drug efficacy and approval.

Despite the frequent use of survival curves in the medical literature, and the fact that the curves are still commonly referred to as Kaplan-Meier curves, it is surprising that so few people know who Dr. Kaplan and Dr. Meier were. Other paired names in medicine, such as Watson and Crick, are much more widely known. The answer may be that Kaplan and Meier were mere statisticians, a profession that is rarely in the limelight. Nonetheless, it is impressive how much impact two statisticians had on so many disciplines when they published a paper in a statistical journal in 1958.

Reference

1. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Amer Statist Assn. 1958:53;457-481.

Read last month's editorial: "Keeping Medical Language Patient-Centric."