Learning Objectives: Identify reasons why prevalence estimates varied between the Adult Tobacco Survey (ATS) and Behavioral Risk Factor Surveillance System (BRFSS).
Abstract: Problem:Smoking prevalence is a principal outcome for evaluating tobacco control efforts, but prevalence estimates differ between ATS and BRFSS.
Methods:We determined New Hampshire smoking prevalence from the 2002 ATS and 2001 BRFSS and examined possible reasons for the observed difference.
Results:Smoking prevalence was 17.9% (95% confidence interval [CI] 16.3–19.5) in ATS and 24.1% (95% CI 22.6–25.6) in BRFSS. Both ATS and BRFSS are population-based, random-digit-dialed telephone surveys of persons aged >=18 years. ATS (response rate 52.6%) included 103 questions related to knowledge, attitudes, and behaviors regarding tobacco. BRFSS (response rate 41.4%) included 138 questions related to health behavior risk factors including tobacco. Questions to determine smoking prevalence were identical in both surveys. The introduction to ATS informed potential participants that it was a survey on health and tobacco; BRFSS was introduced as a survey on health and health practices. The majority of ATS questions addressed tobacco use, whereas tobacco questions on BRFSS followed questions related to nine other health topics. ATS was administered over 3 months, whereas BRFSS was administered throughout the year.
Conclusions:The discrepancy in estimated smoking prevalence between ATS and BRFSS might be caused by 1) differences in survey introduction, which might lead to self-selection of nonsmokers; 2) differences in question placement, which might bias answers; and 3) differences in the timing of survey administration, which might be affected by seasonal variation. Methodological differences between the two surveys prevent a direct comparison of results, thereby complicating the evaluation of tobacco control efforts.
Back to Poster Session
Back to Evaluation and Surveillance
Back to The 2003 National Conference on Tobacco or Health (December 10-12, 2003)