Objectives: This study sought to understand the stability of and change in benzodiazepine use among incident long-term benzodiazepine users over a five-year period and to investigate predictors of variation in use patterns from adolescence into adulthood. Methods: Long-term use was defined as receipt of benzodiazepine prescriptions for 31 or more cumulative days in a calendar year. Data for 1999-2005 were obtained from the National Health Insurance Research Database in Taiwan. Two age groups of incident long-term users in 2000 were identified-1,758 aged 12-15 and 5,265 aged 16-19-and their benzodiazepine prescription records from 2001 to 2005 were retrieved. Group-based trajectory analyses and polytomous logistic regression were performed to evaluate differential risk of benzodiazepine use over time. Results: From 3% to 5% of the incident benzodiazepine users were long-term users. Four distinct groups of users emerged from the five years of study data: occasional, decelerating, accelerating, and chronic users. Overall, one-quarter were accelerating or chronic users. A history of psychosis or epilepsy, prescription by providers from multiple specialties, and receipt of benzodiazepines with a long half-life or mixed indications significantly increased one's risk of becoming a chronic or accelerating user (range of adjusted odds ratios from 2 to 6). Conclusions: Patient characteristics and attributes of service providers and pharmacological agents played significant roles in benzodiazepine use patterns. Prescribers can reduce the risk of long-term use by assessing whether pediatric patients have received benzodiazepines from multiple doctors for various medical conditions. (Psychiatric Services 62: 900-907, 2011)
