SPARC: Survey-based Population-adjusted Rate Calculator
Population denominators for calculating rates of cancer incidence and mortality by basic demographics are routinely derived from decennial censuses. Despite small errors due to surveying operations and processing, census-based populations are considered error-free for the purpose of calculating cancer rates (i.e., no sampling error because it is a census). Populations stratified by risk factors such as immigration status, education ascertainment, and hysterectomy status, can only be derived from national sample surveys, such as American Community Survey (ACS) and Behavioral Risk Factor Surveillance System (BRFSS), and are subject to sampling errors.
Not accounting for sampling error in the denominators of cancer rates not only underestimates the variability of the rates, but also produces biased estimates of the rates themselves (i.e., the estimated rates are spuriously high). The NCI, in collaboration with Dr. Jiming Jiang from University of California at Davis, developed a novel method to correct for such sampling errors in estimating age-adjusted rates.
For details about this method, please refer to:
Jiang J, Feuer EJ, Li Y, Nguyen T, Yu M. Inference about age-standardized rates with sampling errors in the denominators. Statistical Methods in Medical Research. October 2020. doi:10.1177/0962280220962516 [Abstract]
Development of SPARC
The Survey-based Population-adjusted Rate Calculator (SPARC) is a web-based tool currently under development to implement this novel method. The first application that will be released under SPARC will calculate age-adjusted rates of US mortality for US-born and foreign-born Americans with population denominators derived from a time-series of ACS samples.
In this first iteration of SPARC, users will be able to calculate age-adjusted mortality rates for US-born and foreign-born Americans with the following characteristics:
- Single year deaths from 2006 to 2017;
- Cause of death coded by ICD-10 mortality coding classification;
- Ability to stratify by age, sex, and race/ethnicity*;
- Standard output will include:
- death counts,
- population estimates and standard errors of the population estimates,
- bias-corrected age-adjusted rates and standard errors of bias-corrected age-adjusted rates,
- 95% confidence intervals of bias-corrected age-adjusted rates, and
- arithmetic difference between two age-adjusted rates and the standard error;
- Optional output will include age-adjusted rates calculated without correcting for sampling errors and corresponding standard error.
*Race/ethnicity is limited to Non-Hispanic White, Non-Hispanic Black, Non-Hispanic Asian/Pacific Islander, and Hispanic. Denominators for Hispanic by detailed race and Non-Hispanic American Indian & Alaska Natives are not stable for use in calculating age-adjusted rates due to ACS sample size constraint.
For more information about SPARC and more detailed information about its planned release, please contact us.