Abstract
BACKGROUND: Analysis of Medicare data is often used to determine epidemiology, healthcare utilization and effectiveness of disease treatments. We were interested in whether Medicare data could be used to estimate prevalence of tobacco use. Currently, data regarding tobacco use is derived from Behavioral Risk Factor Surveillance System (BRFSS) survey data. We compare administrative claims data for tobacco diagnosis among Medicare beneficiaries to survey (BRFSS) estimates of tobacco use from 2001 to 2014. METHODS: Retrospective cross-sectional study comparing tobacco diagnoses using International Classification of Disease, Ninth Revision (ICD-9) codes for tobacco use in Medicare data to BRFSS data from 2001 to 2014 in adults age ≥ 65 years. Beneficiary data included age, gender, race, socioeconomic status, and comorbidities. Tobacco cessation counselling was also examined using Healthcare Common Procedure Coding System codes. RESULTS: The prevalence of Medicare enrollees aged ≥65 years who had a diagnosis of current tobacco use increased from 2.01% in 2001 to 4.8% in 2014, while the estimates of current tobacco use from BRFSS decreased somewhat (10.03% in 2001 vs. 8.77% in 2014). However, current tobacco use based on Medicare data remained well below the estimates from BRFSS. Use of tobacco cessation counselling increased over the study period with largest increases after 2010. CONCLUSIONS: The use of tobacco-related diagnosis codes increased from 2001 to 2014 in Medicare but still substantially underestimated the prevalence of tobacco use compared to BRFSS data.