Collecting and evaluating adverse events is an essential aspect of weighing the risk-benefit ratio of a drug. By providing a mechanism for classifying adverse events, dictionary coding is integral to evaluating safety data. In this paper, coding philosophy is discussed in terms of organizational structure (centralized versus decentralized) and level of automation (manual versus computer-assisted). Then a strategy for implementing a computer-assisted coding methodology that maximizes coding consistency and accuracy using the Astra Merck coding system as a case study is presented. Finally, techniques for evaluating and improving coding systems are described.
International monitoring of adverse reactions to drugs.Adverse reactions terminology. Uppsala, Sweden: WHO Collaborating Centre for International Drug Monitoring; 1993.
2.
Coding Symbols for a Thesaurus of Adverse Reaction Terms. 3rd ed.Rockville MD: US Food and Drug Administration; 1989.
3.
WoodKL. The Medical Dictionary for Drug Regulatory Affairs (MEDDRA) Project. Pharmacoepidemiol Drug Safety.1994;3:7–13.
4.
WestlandM. Coding: The Mortar in the Bricks of Data Analysis. Drug Inf J.1991;25:197–200.
5.
StephensMDB. The Management of Adverse Drug Reaction and Adverse Event Data Through Collection, Storage and Retrieval. In: Detection of New Adverse Drug Reaction.3rd ed.New York: Van Nostrand Reinhold; 1992:244–262.
6.
MatsumotoC. Implementation and Use of the COSTART Dictionary at Lilly: Would We Do It Again?. Drug Inf J.1991;25:201–203.
7.
GillumTLeitmeyerJ. Consistency in Adverse Event Terminology: Merck's Modified CLINTERM. Drug Inf J.1993;27:431–435.
8.
JosephMCSchoefflerKPeggyAD. An Automated COSTART Coding Scheme. Drug Inf J.1991;25:97–108.
