iSixSigma

Kaplan-Meier

Definition of Kaplan-Meier:

The Kaplan-Meier method is a nonparametric (actuarial) technique for estimating time-related events (the survivorship function). Ordinarily it is used to analyze death as an outcome. It may be used effectively to analyze time to an endpoint, such as remission.

It is a univariate analysis and is an appropriate starting technique. It estimates the probability of the proportion of individuals in remission at a particular time, starting from the initiation of active date (time zero), is especially applicable when length of follow-up varies from customer to customer, and takes into account those customer lost to follow-up or not yet in remission at end of study (censored customers, assuming the censoring is non-informative). It is therefore the instrument of choice in evaluating remissions following loosing a customer. Since the estimated survival distribution for the cohort study has some degree of uncertainty, 95% confidence intervals may be calculated for each survival probability on the “estimated” curve.

A variety of tests (log-rank, Wilcoxan and Gehen) may be used to compare two or more Kaplan-Meier “curves” under certain well-defined circumstances. Median remission time (the time when 50% of the cohort has reached remission), as well as quantities such as three, five, and ten year probability of remission, can also be generated from the Kaplan-Meier analysis, provided there has been sufficient follow-up of customers.

The Kaplan-Meier technique is usually only useful as a method of preliminary evaluation, since it is purely a descriptive method for the evaluation of one variable.

« Back to Dictionary Index