Using American Community Survey Disability Data to Improve the Behavioral Risk Factor Surveillance System Accuracy
Starting in 2013, the Behavioral Risk Factor Surveillance System (BRFSS) included five of the six disability questions that are now common across national surveys including the American Community Survey (ACS). The common disability questions will potentially facilitate new disability research; however, the BRFSS disability prevalence estimates based on the new disability questions are substantially higher than the ACS-based estimates. We assessed possible explanations for the differences. We find that BRFSS sampling and nonresponse bias is the most likely explanation for the differences between the ACS and BRFSS estimates.
To reduce the BRFSS sampling and nonresponse bias associated with the disability questions and to increase the accuracy of disability prevalence estimates, we re-weight the Massachusetts BRFSS data. We find that the re-weighting methodology is a viable alternative to the current BRFSS weighting methodology.
Importantly, the re-weighing changes BRFSS population and subpopulation estimates compared to estimates based on the original weights, for example population estimates for the prevalence of diabetes, asthma, and smoking are lower with the re-weighted data. This is expected because the re-weighting decreased the disability prevalence estimates and persons with disabilities, on average, have a higher prevalence of diabetes, asthma and smoking compared to persons without disabilities.