Computational Realization of Period Analysis for Monitoring Cancer Patient Survival

To the Editor: The essential outcome measures for groups of patients with cancer are given by their long-term survival rates. Reflecting this importance, the monitoring of long-term survival rates is routinely performed by many cancer registries throughout the world. The predominant method of estimating such survival figures, typically reported as 5- and 10-year survival rates, is related to a cohort approach in which only patients followed over the full follow-up period of interest are included in the analysis. 1 However, survival estimates derived in this way do not reflect potential recent progress in prognosis. Another type of survival analysis also includes more recently diagnosed patients who have not completed the full follow-up period and who are therefore censored at the closing date of follow-up. 2,3 Even with this approach, the most up-to-date long-term survival estimate reflects survival experience over a time period that dates back a long time in the past, and it may still be quite outdated if there has been recent improvement of survival. Some years ago, a new method of survival analysis, denoted period analysis, was proposed to obtain more up-to-date estimates of long-term survival rates. 4,5 Period analysis exclusively reflects the survival experience of patients within some most recent calendar period, such as the most recent calendar year, for which mortality follow-up is available. This is achieved by left truncation of observations at the beginning of this period in addition to the right censoring mentioned above. Recently, various cancer registries have begun to use period analysis for reporting more up-to-date estimates of long-term survival rates. 6–9 However, the widespread application has been hindered by the lack of computer software. To remove this obstacle in the present case, we have developed simple and easy-to-use computer software that enables one to carry out period analysis of survival with the type of data commonly available in population-based cancer registries. The program uses a life table approach in which conditional survival probabilities for 1-year intervals after diagnosis are combined to derive cumulative survival rates. Both absolute and relative survival rates 10 can be estimated. Two SAS macros are provided which can be run with the SAS statistical software package (version 8 or older). 11 In one macro, called “period,” the relative survival rates are estimated according to the method commonly known as the Ederer II method, 12 whereas in the other macro, called “periodh,” they are estimated according to Hakulinen's method. 13 The macros can be downloaded free of charge from the statistical archive network maintained by the Department of Medical Informatics, Biometry, and Epidemiology at the University of Erlangen-Nuremberg (http://www.imbe.med.uni-erlangen.de/issan/SAS/period/period.htm) or obtained on diskette by written request to the first author. Extensive validation efforts have been made to guarantee the correctness of the macros. Point estimates and standard errors of absolute and relative survival rates, calculated according to Greenwood's method, 14 are provided. More details concerning the technical aspects of the computational realization, as well as the structure of input data for running the software, have been compiled in the documentation of the software, which can obtained as specified above. The source code of the SAS macro is open code under the conditions of the GNU-GPL license. 15 Consequently, the source code can be easily modified by future users to provide such features as additional output or graphical presentations of survival curves. In conclusion, we hope that the easy access to computer software offered here will promote the application of period analysis which has been shown to yield more up-to-date estimates of long-term survival rates than traditional methods of survival analysis. 8,16 Hermann Brenner Timo Hakulinen Olaf Gefeller