A Millennium Bug Still Bites Public Health - An Illustration Using Cancer Mortality

摘要

Accurate estimation of cancer mortality rates and the comparison acrosscancer sites, populations or time periods is crucial to public health, asidentification of vulnerable groups who suffer the most from these diseases maylead to efficient cancer care and control with timely treatment. Because cancermortality rate varies with age, comparisons require age-standardization using areference population. The current method of using the Year 2000 PopulationStandard is standard practice, but serious concerns have been raised about itslack of justification. We have found that using the US Year 2000 PopulationStandard as reference overestimates prostate cancer mortality rates by 12-91%during the period 1970-2009 across all six sampled U.S. states, and alsounderestimates case fatality rates by 9-78% across six cancer sites, includingfemale breast, cervix, prostate, lung, leukemia and colon-rectum. We develop amean reference population method to minimize the bias using mathematicaloptimization theory and statistical modeling. The method corrects the bias tothe largest extent in terms of squared loss and can be applied broadly tostudies of many diseases.