Introduction: Pharmacotherapy and counseling for tobacco cessation are evidence-based methods that increase successful smoking cessation attempts. Medicaid programs are required to provide coverage for smoking cessation services. Monitoring utilization is desirable for program evaluation and quality improvement. Various methodologies have been used to study utilization. Many factors can influence results, perhaps none more than how smokers are identified. This study evaluated the utilization of smoking cessation services using various methods to estimate the number of smokers within New York State's (NYS's) Medicaid program in 2015.
Methods: Estimates of utilization were generated based on Medicaid claims and encounters and four sources of smoking prevalence: two population surveys, one Medicaid enrollee survey, and diagnosis codes. We compared the percentage of (estimated) smokers utilizing cessation services, and the average number of services used, across fee-for-service and managed care populations, and by cessation service category.
Results: Statewide, smoking prevalence estimates ranged from 10.9% to 31.5%. Diagnosis codes identified less than 45% of smokers estimated by surveys. A similar number of cessation counseling (199106) and pharmacotherapy services (197728) were used, yet more members utilized counseling (126839) than pharmacotherapy (91433). The estimated percentage of smokers who used smoking cessation services ranged from 15.1% to 43.4%, and the estimated average number of cessation services used ranged from 0.31 to 0.90 per smoker.
Conclusion: Smoking prevalence estimates obtained through surveys greatly exceed prevalence observed in diagnosis codes in NYS's Medicaid data. Use of diagnosis codes in the analysis of smoking cessation benefit utilization may result in overestimates.
Implications: Selection of a smoking prevalence data source for similar analyses should ultimately be based on completeness of the data and applicability to the population of interest. Evaluation of smoking cessation benefit utilization and the effectiveness of tobacco control campaigns aimed to increase utilization requires a well-defined methodology which ensures reliable baseline data. Comparing utilization estimates across populations or state lines can be misleading, as differences in how estimations were generated can greatly bias observed results.