How do I enter data for subjects still alive at the end of the study?

When a study subject or participant reaches the end of the study without experiencing the event of interest, the observation for that individual is considered to be censored. You know how much time has elapsed since they entered the study, but you don’t know how much more time will elapse until they experience the event of interest. To enter this data, simply enter the elapsed time of observation for this individual into the X column (or the starting and ending dates of their observation), and enter a zero in the appropriate Y column (to indicate that the observation was censored).

What if two or more subjects experience the event of interest at the same time?

This isn’t a problem. Simply enter the observation data for each individual on separate rows: the elapsed time values for each of these rows will be the same, and the outcome (event/censor) values will be entered into the appropriate row of the outcome variable column. Note that this isn’t restricted to events occurring at the same time, and can be applied to two individuals who had the same elapsed time of observation, with one experiencing the event of interest, and the other being censored. The process for entering the data is the same.

How do I enter data for a subject who died of an unrelated cause?

When performing a survival study where death is the event of interest (common in animal studies), there are different ways that this situation can be handled. Some investigators will treat a death from any cause as the event of interest and record it as such. Others will treat a death from an unrelated cause to be a censored observation. Ideally, this decision should be made in the study design. If the study design is ambiguous, you should decide how to handle these data before unblinding the study. Options on the Model tab of the Cox regression parameters dialog allow you to specify how these "other" outcomes should be handled.

Do the X values have to be entered in order?

No. As long as each row represents a different observation (individual), the data can be entered in any order that you want.

How does Prism distinguish between subjects who are alive at the end of the study and those who dropped out of the study?

There is no distinction in this case. In both situations, the observation is considered censored. You know that the patient was alive and on the study protocol for a certain period of time. After that, you can’t know (because the subject has not yet experienced the event of interest) or can’t use (because the subject stopped following the study protocol) any subsequent information about that subject. Survival analysis calculations treat all censored subjects in the same way. Until the time of censoring, censored subjects are considered “at risk”, and thus continue to contribute towards the calculation of percent survival. After these subjects are censored, there is no way to know if they are “at risk” and so are removed from the survival calculations. These essentially represent missing data.

I already have a life-table showing percent survival at various times. Can I enter this table into Prism?

To perform survival analysis, no. Prism can only analyze survival data if you enter the elapsed time of observation until the event of interest for each subject. If your goal is simply to create a graph of the survival curve for this life-table, then the data can be entered into an XY data table, with time entered into the X column and percent survival entered into the Y column. When switching to the graph for the first time, select the “Connecting line only” graph type from the XY family of graphs and click OK. Open the Format Graph dialog (double click on the graph) and on the Appearance tab of this dialog in the “Show connecting line/curve” section, open the “Style” dropdown menu, and select the “Survival” option. Click OK, and the survival curve for this data will be displayed. Note that this will not include censored observations on the graph.

Can I enter a starting date and ending date rather than total elapsed time for Cox regression?

No. Currently, Cox proportional hazards regression requires values of elapsed time entered in appropriate units (months, weeks, days, etc.). These values do not have to be entered as integers, and can be entered as decimal values.

How do I handle data for subjects that were “enrolled” in a study but never treated?

Most clinical studies follow the “intention to treat” rule. You analyze the data assuming that the subject got the treatment they were assigned to receive, even if the treatment was never given. This decision, of course, should be made as part of the experimental design.