Tools for Cancer Survival Analysis
Analysis program (strel) and life tables
The computer program strel is designed for relative survival analysis. It is based on the maximum likelihood approach to survival estimation using individual tumour records (Estève et al., 1990). You will need a licence for Stata®, in which the program is written, but our program strel is free. It has been widely used for cancer survival analyses in the UK and many other countries. You can download the most recent version of strel once you have registered. To do this, go to the ‘Register for Access’ page, complete and submit the form, and a username and password will be sent to you.
Relative survival is the ratio of the observed cumulative probability of survival in the study group and the survival that would have been expected if the group had only been subject to the background mortality in the general population (obtained from life tables).
A suite of life tables is available to registered users (go to ‘Login for Registered Users’ and enter your username and password). You will have access to the life tables we have constructed for England and Wales containing age-sex-mortality rates for five deprivation groups based on various indices of deprivation, by geographic region and calendar year or period. Life tables for Scotland, Northern Ireland and the Republic of Ireland are also available.
Sub-national life tables have been smoothed either by applying Ewbank's 4-parameter model life table system to the observed mortality rates with the English Life Table 1991 as standard (archive), or by a Poisson model (published in 2009). You can perform the Ewbank procedure with your own data using our Stata® command ewblft, which you can also download free, once you have registered.
Both strel and ewblft are copyright to the London School of Hygiene & Tropical Medicine. When you register to obtain access you will be agreeing to accept the terms of the licences, so please read them carefully.
Handling missing data
Missing data frequently create problems in the analysis of population-based datasets, such as those collected by cancer registries. We have published a tutorial article explaining how to deal with such data in survival analysis:
Modelling relative survival in the presence of incomplete data: a tutorial Int J Epidemiol. 2010; 39: 118-28
Ula Nur, Lorraine G Shack, Bernard Rachet, James R Carpenter and Michel P Coleman
We estimated relative survival for 29 563 colorectal cancer patients who were diagnosed between 1997 and 2004 and registered in the North West Cancer Intelligence Service. The method of multiple imputation was applied to account for the common example of incomplete stage at diagnosis, under the “missing at random” (MAR) assumption. Multiple imputation greatly improved the results by exploiting all the information in the incomplete records. This method also helped to ensure efficient inferences about survival were made from the multivariate regression analyses.
Modelling relative survival in the presence of incomplete data, STATA code (PDF 0.44 MB)
