Please enable JavaScript to view this site.

To start the analysis, either click the Analyze button in the Analysis section of the toolbar, or open the Analyze menu and select Regression and curves > Cox proportional hazards regression (Cox proportional hazards regression can also be found in the "Survival analyses" section of the Analyze menu and Analyze Data dialog). The analysis parameters dialog that appears contains eight tabs:

Model. Specify the time-to-event (response) variable for the analysis and the outcome (event/censor) variable. These two variables are required for every Cox proportional hazards regression analysis. Specify which value (or level) represents observations with "Events", which represents observations that were “Censored”, and how to handle any other values or levels for the selected variable. Also on this tab, indicate how Prism should handle ties in elapsed times. Finally, select which predictor variables, interactions, and transforms you would like to include in the model using the “Define model” section

Reference level. Set a reference level for any categorical predictor variable in the specified model. The reference level indicates a “baseline” or “usual” level of the categorical variable, and is important for interpretation of analysis results

Predictions. Use the model estimated by Prism to predict survival probabilities using values for each of the predictor variables along with a specified elapsed time

Compare. Choose a second model and specify how the fit of the two models should be compared

Options. Specify which results Prism should report (note, the Goodness-of-fit, Residuals, and Graphs tabs also contain important options for customizing the results output from this analysis)

Goodness-of-fit. Specify which analysis metrics Prism should report. Each of these provide some insight into how well the model fits the given data

Residuals. Select which analysis residuals Prism should report and graph. Note that the "residuals" from Cox proportional hazards regression are mathematically different from residuals for linear regression. Instead, these are values that are used to answer similar questions and test similar assumptions about the regression model that standard residuals are typically used for

Graphs. Use the model estimated by Prism to generate predicted survival curves using values for selected predictor variables in the model spanning all observed time points in the data

© 1995-2019 GraphPad Software, LLC. All rights reserved.