Catherine Barber1, David Gagnon2, Jennifer Fonda2, Kelly Cho2, John Hermos2, Matthew Miller1,3. 1. Harvard Injury Control Research Center, Harvard School of Public Health, Boston, MA, USA. 2. VA Boston Healthcare System, Massachusetts Area Veterans Research and Information Center (MAVERIC), Boston, MA, USA. 3. Department of Health Sciences, Northeastern University, Boston, MA, USA.
Abstract
PURPOSE: The study aimed to test whether directives on opioid prescribing released by the Veterans Health Administration (VHA) or the Food and Drug Administration (FDA) had an impact on prescribing among VHA providers. METHODS: We used the VHA's linked pharmacy and patient medical records database to identify new prescriptions written for propoxyphene, fentanyl, and controlled release (CR) oxycodone between 1/1/2000 and 12/31/2009. We plotted the monthly proportion of these prescriptions that complied with components of four specific safety alerts or directives for these substances issued by the VHA or FDA between 1/1/2001 and 12/31/2008. We modeled compliance using interrupted time series analysis and a generalized additive model with the addition of an indicator variable to flag prescriptions that followed the directive's release date. RESULTS: A total of 32.2 million new prescriptions for fentanyl, oxycodone CR, and propoxyphene were written for VHA patients meeting inclusion criteria. Compliance with guidelines in the directives increased steadily throughout the entire study period, with no clinically meaningful inflection point near the date of each directive's release. Generalized additive modeling and interrupted time series analysis found that the indicator flag slightly improved the fit of the data, but visual inspection of the plots revealed no change at a level of practical significance. CONCLUSIONS: While prescribing compliance increased throughout the period, release of FDA and VHA alerts and guidelines did not appear to contribute to this change. Given the fivefold increase in the rate of drug-related overdose deaths since 1990, identifying effective methods to communicate safety messages and change prescriber behavior remains a priority for future work.
PURPOSE: The study aimed to test whether directives on opioid prescribing released by the Veterans Health Administration (VHA) or the Food and Drug Administration (FDA) had an impact on prescribing among VHA providers. METHODS: We used the VHA's linked pharmacy and patient medical records database to identify new prescriptions written for propoxyphene, fentanyl, and controlled release (CR) oxycodone between 1/1/2000 and 12/31/2009. We plotted the monthly proportion of these prescriptions that complied with components of four specific safety alerts or directives for these substances issued by the VHA or FDA between 1/1/2001 and 12/31/2008. We modeled compliance using interrupted time series analysis and a generalized additive model with the addition of an indicator variable to flag prescriptions that followed the directive's release date. RESULTS: A total of 32.2 million new prescriptions for fentanyl, oxycodone CR, and propoxyphene were written for VHA patients meeting inclusion criteria. Compliance with guidelines in the directives increased steadily throughout the entire study period, with no clinically meaningful inflection point near the date of each directive's release. Generalized additive modeling and interrupted time series analysis found that the indicator flag slightly improved the fit of the data, but visual inspection of the plots revealed no change at a level of practical significance. CONCLUSIONS: While prescribing compliance increased throughout the period, release of FDA and VHA alerts and guidelines did not appear to contribute to this change. Given the fivefold increase in the rate of drug-related overdose deaths since 1990, identifying effective methods to communicate safety messages and change prescriber behavior remains a priority for future work.