Survival analysis for a treatment in oncology

Project background

A client focusing on several oncology treatments needed relatively accurate outcomes to determine the Economically Justifiable Price of their drug in the new indication. For this, it was necessary to input survival estimates into our model. However, at that point, we had not yet received survival estimates to produce realistic survival distributions. 

Challenges faced

The main challenge was to swiftly set up and perform a survival analysis on earlier data (i.e., Kaplan-Meier curves) from the client. For this project, we had agreed with the client that survival analysis would be performed by their statistical team and not us. However, their statistical team was not able to produce these results at that time, and we had to perform the survival analysis ourselves instead.

The model specialist had no prior knowledge of performing survival analysis and therefore had to quickly perform digitization of Kaplan-Meier curves, adjust the R-code to properly accept the data, and implement the outcomes into the model.

Our solution

With the help of consultants who had worked on the R-code before, it was possible to adjust the script and run the code to produce proper output within a few days. For the other parts of the survival analysis (i.e., digitisation, implementation) we already had an SOP available that adequately helped in performing the analysis.

Our impact

The survival analysis was performed well because after inputting survival estimates into the model, the resulting median survival outcomes were similar/comparable to the reported median survival outcomes from the trial readout. Furthermore, later in the project, after the client’s statistical team performed the survival analysis themselves, the outcomes were very much comparable to the outcomes from our survival analysis.