I want you to visualize a staircase with people on it. No I am serious here, try to imagine there are 100 people who fit on a very large staircase. The people are walking down the staircase as a group. As time passes, some of those people will not be continuing down the staircase. So your “group” gets smaller and smaller over time. Each time a person falls out of the group, it is considered an “event”. But your concern is the time it takes until an event occurs and the rate at which people leave the group. There will be people who never leave the group and we consider those cases as “censored”cases. The people who have the event are your primary focus. Why do we care about these events? You care because in the real world, survival of an outcome is measured by time. Think about it…
Since the “events” are not occurring all at once you want to follow the behavior patterns in a special graph. The graph is called a Kaplan Meier curve, which looks like a series of steps. Horizontal lines on the graph represent the proportion of patients that survive to that time point. Over time your “steps” move in a down-ward slope towards the x-axis line because there is a change in the proportion of survivors. The following are main themes in survivor analysis: the event, the time period under study, censored cases, probability of survivorship.
Now you can try to piece together a demographic profile for survivorship and associate probabilities with it. And the information is useful for studying treatment interventions and risk factors. So the next time you see those “stairs” in a graph, you can identify it as a survival analysis graph or Kaplan Meier curve.
If you enjoyed this blog there are more to follow-Amy
