Tips for matching disease databases

New York's extensive experience with matching computerized databases for enhancing AIDS surveillance has yielded a few tips that other health departments might want to follow, says Susan Forlenza, PhD, director of the department's office of AIDS surveillance. They include:

· After identifying potential databases, make sure the database partner understands what information is going to be imported and whether the information will be shared one-way or two-way. When dealing with agencies outside the health department, you may need to have a written memorandum of understanding.

· To ensure accuracy, partners must assess the purpose of the match and understand the criteria for designing data elements you plan to share. Conflicts over case definitions must be identified and gold standards should be established for each database.

· Validation studies should be implemented to determine the accuracy of information, particularly when case definitions differ.

· Various matching keys may need to be developed beyond a patient's name in order to ensure matches. Other keys that may be useful include social security number, date of birth, zip code, and reversal of first and last name.

· Accurate case identifier matches require complex algorithms and careful attention to details. Database formats may differ, complicating the matching process.

· A cost-effectiveness study is a useful way to determine whether the effort was worthwhile.