Numerical estimation of the variance of the completeness index applied to cancer data

Abstract

Cancer prevalence is the proportion of people in a population diagnosed with cancer in the past and still alive. One way to estimate prevalence is via population-based registries, where data on diagnosis and life status of all incident cases occurring in the covered population are collected. In this report a numerical method for the estimation of the variance of the completeness index (NUMCOMP) is developed, and comparisons are made with a previous analytical method (VARCOMP) proposed by Gigli et al.(2006). The paper is organized as follows: section 1 introduces the problem; section 2 illustrates the new method; section 3 describes the algorithm in details; finally in section 4 the new and old methods are applied to the following cancer sites: all sites, anus, brain, colorectal for males and females, and to breast cancer for females (all races/ethnicities) and results are compared and commented.