gms | German Medical Science

Age adjustment is a crucial component in valid comparison of cancer survival between populations or over time. Traditionally, age adjustment was mostly done by calculating a weighted average of age specific survival rates, with weights reflecting the age distribution of some standard cancer population [Ref. 1], [Ref. 2]. Recently, an alternative method of age adjustment was proposed to overcome practical and conceptual difficulties commonly encountered in this context: rather than calculating a weighted average of age specific survival rates, specific weights are individually assigned to all patients in different age groups in the first place, and one then carries out conventional survival analysis using the weighted individual data [Ref. 3].

To provide an empirical assessment and comparison of age adjusted absolute and relative survival rates obtained with the traditional and the alternative method, we applied age adjustment to an international standard cancer population of 5- and 10-year survival rates of patients diagnosed with 20 common forms of cancers in Finland.

Point estimates and their standard errors obtained with both methods were almost identical for age adjusted 5- and 10-year absolute survival rates. However, the traditional age adjustment often altered the relative survival rates in a counterintuitive way. This pattern was particularly pronounced for the 10-year relative survival rates. For example, 10-year relative survival rates adjusted by the traditional method were often considerably lower than the crude estimates even if the age distribution of the standard cancer population was similar to or more favorable than the age distribution of the study population. The alternative method generally provided more meaningful point estimates with an often smaller standard error for the age adjusted relative survival rates.

Application of the alternative method may enhance both validity and precision of comparative analyses of cancer patient survival. For both the alternative and the traditional methods, special analytical techniques have to be used, however, to derive valid estimates of standard errors of relative survival rates [Ref. 4].