![[ Visit Client Website ]](images/banner.png)
BACKGROUND: The question used to derive the prevalence estimate of having a dental visit in the past year among adults aged ≥18 years (DV) has been included in the Behavioral Risk Factor Surveillance System (BRFSS) rotating core questionnaire in even-numbered years since 2002. Beginning in 2011, the BRFSS incorporated two methodological changes: 1) the addition of cell phone interviews to the landline interviews; and 2) the use of “raking” weighting methodology. We used 2012 BRFSS data to assess changes in DVs with addition of cell phone interviews and variation in changes by socio-demographic factors and state.
METHODS: We calculated DVs for landline and cell phone combined data using weight _LLCPWT (DVLLCP) and for landline data using weight _LANDWT (DVLL), separately. We assessed changes in DVs with the cell phone interview addition by absolute difference between DVLLCP and DVLL. Analyses were performed overall nationwide and stratified by state, sex, age, race/ethnicity, education level, household income, and marital status. T-tests were used to indicate statistical significance of differences (p≤0.05).
RESULTS: In 2012, although the overall DVLLCP (65.4%, 95% confidence interval (CI): 65.1%, 65.7%) was slightly lower than DVLL (65.7%, 95% CI: 65.3%, 66.2%) by 0.3%, the difference was not significant. Lower DVLLCPs than DVLLs were observed in most subpopulations, with differences ranging from -0.1% among 35-64 year olds, Non-Hispanic blacks and married to -1.7% among 18-34 year olds and Hispanics. Four subpopulations showed higher DVLLCP than DVLL, with differences ranging from 0.1% among high-school graduate to 1.5% among 65+ year olds. Significant differences were observed only among adults with household income ≥$50,000 (-0.8%), males (-1.0%), and 18-34 year olds (-1.7%) and 65+ year olds (1.5%). When stratified by state, DVLLCPs were lower than DVLLs in 32 states and DC, with a median difference of -0.9%, ranging from -0.1% in Colorado and Maine to -3.4% in DC. DVLLCPs were higher than DVLLs in 17 states, with a median difference of 0.4%, ranging from 0.1% in Oregon and Washington to 2.7% in Idaho. Only differences in Michigan (-2.2%) and Utah (-2.9%) were significant.
CONCLUSIONS: Adding cell phone interviews in the 2012 BRFSS slightly shifted the DV estimates downward, overall and for most socio-demographic subpopulations. Most of the differences, however, were not statistically significant.