New statistical technique shows more informative picture of survival

Jan 08, 2008

Researchers have developed a new method for presenting clinical trial survival data that includes data from all trial participants unlike the standard method, according to a commentary published online January 8 in the Journal of the National Cancer Institute.

In clinical studies, “time-to-event” data represents the time from the start of a study to an event, such as disease recurrence or death. But often many participants in a study do not experience an event before the study is over, so their survival time is not known. To overcome this data gap, the standard statistical method for presenting time-to-event results, known as the Kaplan-Meier survival curve, involves plotting the proportion of individuals surviving without an event over the period of the study. Using this method, researchers get an estimate of the median survival times. However, these plots also tend to make differences in survival between groups visually appear larger than they actually are.

To address this problem, Patrick Royston, D.Sc., of the Medical Research Council Clinical Trials Unit in London and colleagues developed a new method for plotting survival as a bar graph and tested it on data from a kidney cancer trial. In cases where a participant had not experienced an event, the researchers estimated that person's survival by using their prognosis and length of time in the trial.

Their plots show considerable overlap in survival times between treatment and control groups in the kidney cancer trial, whereas the Kaplan–Meier plots of the same data showed a distinct separation between the two groups. The authors argue that the new method gives a more realistic representation of what are usually small differences between groups.

“The method is surprisingly informative and, we hope, will help physicians and patients to understand more fully the results of clinical trials and the implications of prognostic assessments,” the authors write.

In an accompanying editorial, Janet Wittes, Ph.D., of Statistics Collaborative in Washington, D.C., discusses the challenges in interpreting time-to-events graphs, how this new method addresses these problems, and under what circumstances this method should be used.

“Those of us who work with time-to-event data should now attempt to extend their method to…settings other than survival,” Wittes writes.

Source: Journal of the National Cancer Institute

Explore further: DNA blood test detects lung cancer mutations

Related Stories

Canada revises upward CO2 emission data since 1990

5 minutes ago

Canada revised its greenhouse gas emission data from 1990 to 2013 in a report Friday, showing it had higher carbon dioxide discharges each year, and a doubling of emissions from its oil sands.

Recommended for you

DNA blood test detects lung cancer mutations

Apr 17, 2015

Cancer DNA circulating in the bloodstream of lung cancer patients can provide doctors with vital mutation information that can help optimise treatment when tumour tissue is not available, an international group of researchers ...

Tumors prefer the easy way out

Apr 17, 2015

Tumor cells become lethal when they spread. Blocking this process can be a powerful way to stop cancer. Historically, scientists thought that tumor cells migrated by brute force, actively pushing through whatever ...

User comments : 0

Please sign in to add a comment. Registration is free, and takes less than a minute. Read more

Click here to reset your password.
Sign in to get notified via email when new comments are made.