Assessing the effect of a guideline change on drug use prevalence by including the birth cohort dimension: the case of benzodiazepines

• Published in: Pharmacoepidemiology and drug safety 2013-09, Vol.22 (9), p.933-941
• Main Authors: Bijlsma, Maarten J., Hak, Eelko, Bos, Jens, De Jong-van den Berg, Lolkje T. W., Janssen, Fanny
• Format: Article
• Language: English
• Subjects: Adult; Age Distribution; age-period-cohort; Aged; Aged, 80 and over; benzodiazepine; Benzodiazepines - administration & dosage; Benzodiazepines - therapeutic use; Biological and medical sciences; Clinical trial. Drug monitoring; Cohort Studies; Databases, Pharmaceutical; Drug Prescriptions - statistics & numerical data; Drug therapy; drug utilization; Drug Utilization - statistics & numerical data; Drug Utilization - trends; Female; General pharmacology; Guidelines; Humans; Male; Medical sciences; Middle Aged; Netherlands - epidemiology; pharmacoepidemiology; Pharmacoepidemiology - methods; Pharmacoepidemiology - statistics & numerical data; Pharmacology; Pharmacology. Drug treatments; Practice Guidelines as Topic; Prevalence; projection; Sex Distribution; Time series; time series analysis; Young Adult
• ISSN: 1053-8569; 1099-1557
• DOI: 10.1002/pds.3466

ABSTRACT Purpose The aim of this study was to investigate whether including the birth cohort dimension in time series analysis leads to a more accurate estimation of the (long‐term) effect of a guideline change on the trend of benzodiazepine use. Methods We calculated age‐specific (20–84 years) and sex‐specific prevalence of benzodiazepine use per 1000 population per quarter year (1998 to 2008) using a prescription database set in the Netherlands. We studied the prevalence over time by age group and within birth cohorts through interrupted time series analyses to estimate the effect of the guideline change in 2001. Results From 1998 to 2008, the overall age‐standardized prevalence of benzodiazepine use per 1000 population declined from ~54 for men and ~107 for women to ~45 for men and ~85 for women. The relative change increased significantly after 2001 for both sexes and for the majority of age groups. Within birth cohorts, the prevalence increased with age until the year 2001 and leveled thereafter. The age–period approach overall had worse model fit indicators than the within‐cohort approach and predicted larger long‐term effects than the within‐cohort approach. The age–period projection estimated 36% decline in benzodiazepine use relative to 2008, whereas the birth‐cohort projection estimated 8% decline. Conclusion Explicitly following birth cohort trajectories led to models with better fit; the conventional approach estimated a stronger long‐term guideline effect. This has important implications for professional practice. Copyright © 2013 John Wiley & Sons, Ltd.