Midkiff K, Johannes CB, Mohan AK. Strengths and limitations of drug abuse warning network data used in a surveillance study. Poster presented at the 23rd ICPE International Conference on Pharmacoepidemiology & Therapeutic Risk Management; August 23, 2007. Quebec City, Canada. [abstract] Pharmacoepidemiol Drug Saf. 2007 Aug; 16(S1-274):S234.

BACKGROUND: Surveillance of possible abuse and misuse of a product is an important component of pharmacovigilance activities for prescription narcotic analgesics. This requires the use of specialized data sources that can provide reliable and complete data on these events. The Drug Abuse Warning Network (DAWN), operated by the Substance Abuse and Mental Health Services Administration has been used for monitoring abuse and misuse of many narcotic analgesics. DAWN captures drug-related emergency department (ED) visits from a sample of US hospitals by retrospective review of ED records. ED visits are weighted to the US population to provide national estimates and if that number is less than 30 or if the standard error exceeds 50% of its value, the estimate is not provided.  

OBJECTIVES: To evaluate the impact of DAWN estimate suppression rules on the completeness of data obtained as part of a narcotic analgesic surveillance program.  

METHODS: Tabulations of DAWN ED visit estimates for overmedication, non-medical use, accidental ingestion and suicide attempt reported for a specific narcotic analgesic of interest and separately for all prescription narcotic analgesics for 2004 were requested. In addition, estimates of the above mentioned ED visit categories stratified on demographics were requested for the specific narcotic analgesic of interest. We examined the proportion of the total requested ED visit estimates that were available from DAWN.

RESULTS: An estimated 71.2 million patients were exposed to all prescription narcotic analgesics in 2004 and of these, about 1.2 million patients were exposed to the specific narcotic analgesic of interest. Of the 32 total ED visit estimates requested 88% were provided, including 100% of the estimates related to all prescription narcotic analgesics and 50% of estimates related to the specific narcotic analgesic of interest. Of the 100 total estimates requested to describe DAWN case demographics for the specific narcotic analgesic of interest, only 22% were returned.  

CONCLUSIONS: Based on the results of this analysis, DAWN data appear to be more complete for larger product groupings compared to a specific product and may be less useful for describing demographic characteristics related to a specific product because of a lower return of ED visit estimates.