Division of Bacterial Diseases (DBD) News Bulletin
November 3, 2011: Content on this page kept for historical reasons.
The U.S. Department of Health and Human Services (HHS) recently launched two strategic plans aimed at reducing health disparities. The plans call for HHS to set data standards and upgrade collection and analysis of data on race, ethnicity, primary language and other demographic categories in line with new provisions of the Affordable Care Act. A recent project aimed to improve the quality and usefulness of race reporting as part of the surveillance data generated by the Active Bacterial Core surveillance (ABCs) implemented by Melissa Lewis and Tracy Pondo will help the country track and monitor race-specific and age- by race-specific incidence rates and improve the understanding of racial disparities for disease incidence. The two developed a multiple imputation method as a routine approach for dealing with missing race data, beginning with the 1996 dataset for all ABCs pathogens, and now completed routinely on each new dataset. The two young statisticians are part of the DBD Biostatistics Office which contributes to the planning and execution of analyses across the division and provides innovative statistical approaches to deal with challenging analyses on a daily basis.
Their work was featured at the August 2011 CDC NCHHSTP Health Equity Symposium that specifically highlighted the role of data in informing and shaping public health policy, practice, and research and addressing factors of underlying health inequities. The novel approach developed and being used by Lewis and Pondo, one new to public health surveillance, will keep ABCs and health monitoring activities on par with modern statistical methodology. ABCs is an important active-laboratory-based surveillance system for Haemophilus influenzae, Neisseria meningitidis, group A Streptococcus (GAS), group B Streptococcus (GBS) and Streptococcus pneumoniae in multiple large diverse U.S. populations. However, race is missing for approximately 15 to 20 percent of reported annual cases. Working with missing data and accurately reflecting uncertainty in the reported incidence rates and risk factor analyses is a challenge. Case reports with missing race information results in a large proportion (15-20%) of ABCs data not being used in standard analysis, and it poses a challenge in correct interpretation of the results.
Lewis and Pondo’s multiple imputation method addresses this challenge and will help lead ABCs to better routine race-specific incidence rate reporting, better risk-factor analyses, and analytic consistency among all users of the data in using reported race data. Multiply imputed data provides a dataset with complete data and allows estimation procedures to account for the uncertainty associated with imputing unknown values. This project was only possible with major statistical and methodological enhancements to better improve ABCs.