Who declines to respond to the reactions to race module?: findings from the South Carolina Behavioral Risk Factor Surveillance System, 2016–2017

BACKGROUND: The inclusion of self-reported differential treatment by race/ethnicity in population-based public health surveillance and monitoring systems may provide an opportunity to address long-standing health inequalities. While there is a growing trend towards decreasing response rates and selective non-response in health surveys, research examining the magnitude of non-response related to self-reported discrimination warrants greater attention. This study examined the distribution of sociodemographic variables among respondents and non-respondents to the South Carolina Behavioral Risk Factor Surveillance System (SC-BRFSS) Reactions to Race module (6-question optional module capturing reports of race-based treatment). METHODS: Using data from SC-BRFSS (2016, 2017), we examined patterns of non-response to the Reactions to Race module and individual items in the module. Logistic regression models were employed to examine sociodemographic factors associated with non-response and weighted to account for complex sampling design. RESULTS: Among 21,847 respondents, 15.3% were non-responders. Significant differences in RTRM non-response were observed by key sociodemographic variables (e.g., age, race/ethnicity, labor market participation, and health insurance status). Individuals who were younger, Hispanic, homemakers/students, unreported income, and uninsured were over-represented among non-respondents. In adjusted analyses, Hispanics and individuals with unreported income were more likely to be non-responders in RTRM and across item, while retirees were less likely to be non-responders. Heterogeneity in levels of non-responses were observed across RTRM questions, with the highest level of non-response for questions assessing differential treatment in work (54.8%) and healthcare settings (26.9%). CONCLUSIONS: Non-responders differed from responders according to some key sociodemographic variables, which could contribute to the underestimation of self-reported discrimination and race-related differential treatment and health outcomes. While we advocate for the use of population-based measures of self-reported racial discrimination to monitor and track state-level progress towards health equity, future efforts to estimate, assess, and address non-response variations by sociodemographic factors are warranted to improve understanding of lived experiences impacted by race-based differential treatment. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1186/s12889-021-11748-y.