Estimating the age-at-onset function using life-table methods

In the analysis of dominantly inherited diseases, the age-at-onset function is often estimated from the observed age-at-onset distribution of cases. This estimate is confounded with the age distribution of the population from which the cases were sampled and is accurate only if there are no competing causes of death. In this paper, we present a straightforward method for calculating a more accurate age-at-onset function under etiologic heterogeneity. We use the life-table approach and survival analysis methods. This method is illustrated using data on first-degree relatives of probands from two sets of families with high cancer incidence: one with breast/ovarian cancer and the other with colon cancer. A comparison of the estimated age-at-onset function obtained by the two methods is presented. In both cases, colon cancer as well as breast/ovarian cancer, the estimates of onset probabilities based on proportion of cases, are consistently higher than those obtained by the life-table method. For breast/ovarian cancer, this difference is not as striking as it is in the case of colon cancer; nevertheless, the method using proportion of cases tends to give a lower estimate of the age-at-onset function (higher probability of being affected at lower age) than the life-table approach.