SPARC: Methodology

Calculating age-adjusted rates (AA Rates) by a risk factor, such as race/ethnicity or birthplace, requires that both numerator data, i.e., counts of cancer cases for incidence or deaths for mortality, and population denominator data, i.e., populations at risk of developing or dying from a cancer, are stratified by the same risk factor. When population data by the risk factor of interest is not collected in Census, it becomes necessary to estimate the populations from sample surveys; thus, the populations are subject to sampling errors.

Sampling errors will cause bias in estimating AA Rates and comparing AA Rates across subpopulations to assess disparity. The NCI, in collaboration with researchers at the University of California at Davis, developed a new inference method with bias adjustment of sampling error.

Using this new method, SPARC produces bias-corrected point estimates of age-adjusted rates (Jiang et al., 2021), along with standard errors and 95% confidence intervals. The absolute and relative difference between two age-adjusted rates, i.e., Rate Difference and Rate Ratio, can also be produced (Jiang et al., 2022). For details about the inference methods for bias-corrected age-adjusted rates, Rate Difference, and Rate Ratio, please refer to:

  • Jiang J, Feuer EJ, Li Y, Nguyen T, Yu M*. Inference about age-standardized rates with sampling errors in the denominators. Stat Methods Med Research. 2021 Feb;30(2):535-548 [AbstractExternal Website Policy]
  • Jiang J, Li Y, Nguyen T, Yu M. Inference about ratios of age-standardized rates with sampling errors in the population denominators for estimating both rates. Stat Med. 2022 May 20;41(11):2052-2068 [Abstract]