The 131st Annual Meeting (November 15-19, 2003) of APHA

Abstract #67952

Shihfen Tu, PhD¹, Quansheng Song, MS², and Craig A. Mason, PhD¹. (1) College of Education and Human Development/UCEDD, University of Maine, 5717 Corbett Hall, Room 1, Orono, ME 04469, (207) 581-9064, shihfen.tu@umit.maine.edu, (2) UCEDD, University of Maine, 5717 Corbett Hall, Room 14, Orono, ME 04469

This poster describes a system in which a virtual population can be used to test the quality of a data-linkage/integration system. Specifically, users can examine the nature and extent of linkage problems due to either specific types of missing or inaccurate information, or specific characteristics or processes occurring in their population.

The system operates by a user inputting multiple parameters that are used to define both the initial characteristics of a population, and to define interactions and life-events within that population over time (e.g., birth, divorce, and in-migration rates). The virtual population is then allowed to “live” and interact over multiple generations, resulting in a final complex simulated population where the user knows all relationships between all individuals. “Official” birth, school, screening, etc., records are then made for each member of the population, and users can define error rates within these records. For example, father information could be eliminated from 40% of all birth records of children born to unmarried mothers. Researchers and/or public health officials could then apply their own linkage algorithms to this data in order to assess exactly how and to what extent these issues result in missing or inaccurate linkages in their own data.

For example, a child may be born to an unmarried mother who later marries and changes her name. How often are such cases lost in the linkage, or linked to the wrong individual? Using this system, one could accurately evaluate the degree to which their own linkage protocol results in such errors.

Learning Objectives:

Based upon material presented, individuals should be able to:
• Identify factors that lead to errors when linking data from multiple Child Health Information Systems
• Describe several common strategies for evaluating the extent of missed linkages or incorrect linkages.
• Articulate benefits obtained throught the use of simulated population data sets when evaluating linkage algorithms.

Keywords: Information System Integration, Methodology

Presenting author's disclosure statement:
I do not have any significant financial interest/arrangement or affiliation with any organization/institution whose products or services are being discussed in this session.