Midkiff K, Johannes CB, Mohan AK. Utility of multiple databases in a surveillance program for prescription narcotic analgesics. Poster presented at the 24th ICPE International Conference on Pharmacoepidemiology & Therapeutic Risk Management; March 19, 2008. Copenhagen, Denmark. [abstract] Pharmacoepidemiol Drug Saf. 2008 Mar; 17(S1-294):S86.

BACKGROUND: The surveillance of possible abuse and misuse of products is important in pharmacovigilance programs for prescription narcotic analgesics, but information needed may not be readily available from a single data source.

OBJECTIVES: To describe how data from two different sources were used to monitor overmedication and non-medical use of 1) fentanyl-containing products as a group; and 2) all prescription narcotics in an ongoing Company sponsored safety surveillance program.

METHODS: We obtained emergency department (ED) reports of overmedication and non-medical use of the drug groups of interest from the Drug Abuse Warning Network (DAWN). Denominators were estimated by obtaining population exposure data from a prescription retail pharmacy database from IMS Health. We estimated annual event rates per 100,000 persons dispensed a narcotic using the number of DAWN ED visits for overmedication and non-medical use during 2004–2005 as the numerator and the projected number of patients dispensed a narcotic of interest as the denominator. We used the delta method to obtain variance estimates for calculation of 95% confidence intervals (CI).

RESULTS: About 70 million persons had exposure to any prescription narcotic analgesic in each year from 2004–2006; of these, about 1.3 million persons each year had a fentanyl exposure. For all prescription narcotic analgesics the crude ED visit rates in 2005 and 2005 of overmedication were 70 (95% CI: 54–90) and 89 (95% CI: 69–114), and of non-medical use were 116 (95%CI: 95–141) and 143 (95% CI: 113– 182). The corresponding rates for all fentanyl products were 205 (95% CI: 136–309) and 368 (95% CI: 261–519) for overmedication and 420 (95% CI: 303–582) and 345 (95% CI: 244–487) for non-medical use.

CONCLUSIONS: These visit rates have limited usefulness as stand-alone point estimates due to the small number of cases and large variances. However, they can provide information on time trends in the possible misuse and abuse of prescription narcotic analgesics, aiding in signal detection.