Methods for Computer Linkage of Hospital Admission-Separation Records into Cumulative Health Histories

Abstract

A study involving the design and testing of a computer system for linking hospital admission-separation records into longitudinal health histories is described. Although this study is based on a sample of the machine-readable files for a single Canadian province, essentially similar punch cards are prepared routinely across Canada as part of a system of universal health care. No data are included in the study that are not already recorded by provincial authorities.The essential step in the linkage is the recognition of whether two or more independently derived records relate to the same person. The methods are designed to extract and use the maximum amount of discriminating power inherent in items of personal identifying information, which may be incompletely reported or variant on some of the records. A statistical weighting system is employed to indicate the degree of assurance that two records should or should not be linked.The computer successfully matched over 99% of all pairs of potentially linkable records, at a rate of 3000 incoming records per minute. These methods are being used as part of an overall operation in which vital and health records are being integrated into individual and family health histories.