Reviewing Adult Smoking Trends in Alaska in Light of Recent Changes in Surveillance Methods and Weighting

Introduction: In Alaska, smoking decreased significantly in the decade since 2000, based on Behavioral Risk Factor Surveillance System (BRFSS) data weighted using the ‘classic' post-stratified method. In 2012, CDC transitions to reporting BRFSS prevalences using a “raked” weighting method that includes adjusting for demographic factors such as education, race and marital status. Because most states have found that smoking prevalence is 4-7 percentage points higher using the ‘raked' versus ‘classic' estimate, framing the discussion about this change is critical for the tobacco prevention and control community. Methods: In this study, we compared the effect of classic versus raked weights on the Alaska adult smoking trend from 2001 to 2010 using standardized population estimates. Regression analysis was used to identify key effects of raking margin variables on tobacco use and exposure. We used CDC's procedures to develop both classic and raked weights, but utilized a standardized set of population controls to run trends for smoking and smoke exposure. Results: We discuss trend analyses overall and by priority groups, including Alaska Natives, non-Natives of low SES, and young adults. Comparing trends was useful in clarifying program impact in Alaska. Discussion: This presentation describes issues encountered in reproducing raked weighting for BRFSS data from 2001 to 2010, including appropriate population estimates, handling missing data, revised categories for race, and consideration of wireless uptake. We also outline issues of importance to the Alaska Tobacco Prevention and Control Program in preparing communication regarding the changes in estimates of smoking prevalence due to raked weighting and inclusion of cell sample.