A Medication Surveillance System Model to Improve Patient Safety Academic Article uri icon


  • BACKGROUND: The recent withdrawal of Vioxx (rofecoxib) from the market and emerging evidence that cyclooxygenase (COX) 2 inhibitors as a class of drugs may raise the risk of adverse cardiovascular events have highlighted the limitations of current postlicensing surveillance systems in the United States. We explored the use of linking national Veterans Health Administration (VHA) administrative datasets into a medication surveillance system to examine hospitalizations for stroke or myocardial infarction (MI) among COX-2 users compared with nonusers. METHODS: We conducted a descriptive study of 719,058 inpatient hospitalizations within the VHA system over a 2-year period, 2002-2003. We analyzed the frequencies of strokes or MI among COX-2 users and nonusers, as well as mutually exclusive celecoxib or rofecoxib groups. We identified a total of 110,583 unique COX-2 users who had 708,617 COX-2 prescriptions. We then linked the COX-2 patients to acute hospital care for stroke or MI and analyzed frequencies by age strata. RESULTS: Approximately 2% of COX-2 users (2246 individuals) had a stroke or MI, which represents 9.2% of all 24,452 hospitalized COX-2 patients. This is slightly higher than the 8.8% of all stroke or MI patients (63,548) accounted for in the total of 719,058 unique hospitalizations. However, adjustments for age produced mixed results in all analyses. CONCLUSIONS: Currently in the United States, medication safety programs are largely focused on monitoring and voluntary reporting of known risks and side effects of drugs as identified by the drug manufacturers, and medication administration process measure reporting. These approaches do not facilitate identification and analysis of a medication's safety where the drug is an independent risk factor for an adverse health outcome that may not have been previously identified. This study demonstrates that a medication surveillance system based on large-scale national administrative datasets can be used to rapidly evaluate a reported medication risk or harm. More importantly, in the absence of a reported medication risk or harm, especially for outpatient medications, a medication surveillance system based on the emerging VHA model could be used to facilitate efficient, timely, and proactive postlicensing medication surveillance. Copyright 2006 by Lippincott Williams & Wilkins.

published proceedings

  • Journal of Patient Safety

author list (cited authors)

  • French, D. D., Spehar, A., Campbell, R., Chirikos, T. N., Bulat, T., Vedenov, D., & Luther, S. L.

complete list of authors

  • French, Dustin D||Spehar, Andrea||Campbell, Robert||Chirikos, Thomas N||Bulat, Tatjana||Vedenov, Dmitry||Luther, Stephen L

publication date

  • January 1, 2006 11:11 AM