New methods of handling cases of unknown age in cancer registry data

The essential assumption of random missing age behind the "conventional method" of handling cancer patients of unknown age does not often hold. This article is to introduce four alternative methods based on more acceptable assumptions.

More cases with unknown age are allocated to the older age-groups in all the new methods. In the "weighting method," cases of unknown age are distributed according to distribution of cases of known age, whereas in the "last-group method," all of them are added to the oldest age-group. In the "progressive method," unknown-age cases are added to the age-groups above 60 progressively (weighting=1/63, 2/63, 4/63, 8/63, 16/63, and 32/63), whereas in the "additive method," they are allocated to the age-groups above 60 additively (weighting=1/21, 2/21, 3/21, 4/21, 5/21, and 6/21). Data were from the Cancer in Five Continent database, vol. VIII.

Age-standardized rates for "All sites" in Zaragoza (Spain), Cali (Colombia), Algiers (Algeria), and Gambia showed that results by all the methods differed, the magnitude ranging from 0.1 to 3.1% depending on the method, registry, sex, and the defined last age-group.

Conventional and weighting methods are not based on acceptable assumptions. The last-group method is not stable because it depends on the defined age-group as last (65+, 75+ or 85+). Both progressive and additive methods have more acceptable assumptions. The progressive method is preferable above all others because it can produce an age-specific curve with the expected exponential increase.