Standard errors of non-standardised and age-standardised relative survival of cancer patients.

Abstract

Background:Relative survival estimates cancer survival in the absence of other causes of death. Previous work has shown that standard errors of non-standardised relative survival may be substantially overestimated by the conventionally used method. However, evidence was restricted to non-standardised relative survival estimates using Hakulinen's method. Here, we provide a more comprehensive evaluation of the accuracy of standard errors including age-standardised survival and estimation by the Ederer II method.Methods:Five- and ten-year non-standardised and age-standardised relative survival was estimated for patients diagnosed with 25 common forms of cancer in Finland in 1989–1993, using data from the nationwide Finnish Cancer Registry. Standard errors of mutually comparable non-standardised and age-standardised relative survival were computed by the conventionally used method and compared with bootstrap standard errors.Results:When using Hakulinen's method, standard errors of non-standardised relative survival were overestimated by up to 28%. In contrast, standard errors of age-standardised relative survival were accurately estimated. When using the Ederer II method, deviations of the standard errors of non-standardised and age-standardised relative survival were generally small to negligible.Conclusion:In most cases, overestimations of standard errors are effectively overcome by age standardisation and by using Ederer II rather than Hakulinen's method.