Survival Curves

The information on effectiveness of treatment of children in trials is best illustrated by survival curves. Survival most commonly refers to the time from diagnosis of disease (entry to trial) until death. It may also be from diagnosis to relapse of the disease or death (event free survival), or from remission to relapse (disease free survival).

At any moment, when analysis of the data is carried out a large proportion of patients will hopefully still be alive and in remission. They can be included in the analysis, and their data is as important as that of relapsing patients. This data is termed 'censored', meaning their time to relapse or death is unknown but must be longer than the last reported time they were alive.

The commonly used method to calculate the best estimate of survival at any time from diagnosis or remission is the "actuarial" (also known as "Kaplan-Meier" or "Product limit") method.

The essential calculation is:

Estimated % surviving beyond time T =

no. of patients surviving beyond T x estimated % surviving up to time T
(no. surviving beyond T + no. dying at T)

This calculation is made after each event. The data are then presented as a series of steps, with a drop at each event time. This provides curves which will not differ systematically from the final curve, calculated when all patients have full follow up.

The shapes of the curves can vary greatly. Clearly the curves cannot climb upwards, but some drop earlier than others and flatten out sooner. This reflects the type and stage of disease, as well as the course of disease.

Comparisons between curves can be made for groups of patients either given different treatments (as in randomised trials) or with different characteristics when initially diagnosed (prognostic factors). A statistical test may be carried out to assess the evidence for any difference. A common test is the log rank test, which is more sensitive to later events than other tests.

>>>> UKCCSG Protocols – development and approval process >>>>

Introduction What is a Clinical Trial? Clinical Trial procedure The role of the Chief Investigator The Protocol How is the size of a clinical trial determined? What is a multi-centre trial? Role of coordinating centre/Trials Unit What is randomisation? How is a patient randomised for a particular trial? What is a control in clinical trials? Forms and data management Statistics in clinical trials

The role of the trial coordinator Follow up How is the trial data analysed/evaluated? What are pharmacokinetic studies? Ethical issues Conduct of Clinical Trials/Research What is a blind trial? Interpretation and publication of findings Survival Curves CCLG Protocols – development and approval process References Download PDF