A study of the non-match rates to assess the magnitude and implications of any systematic biases in the non-linkage rates.

In societies which do not have universal registration systems for monitoring demographic change it is not normally possible to accurately study the process of linking population and mortality data. The NI-LS is unique within the United Kingdom in that it affords such a possibility. The study links the 2001 Census returns to all deaths registered in Northern Ireland in the five years after the Census, with the characteristics of matched and non-matched death records compared using multivariate logistic regression. The attributes of the subjects included in the analysis are as recorded on the death certificate.


O’Reilly D, Rosato M and Connolly S. (2008) Unlinked vital events in census-based longitudinal studies can bias subsequent analysis. J Clin Epidemiol. 61(4):380-5.

Research Team: Dermot O’Reilly, Michael Rosato and Sheelah Connelly
Database: NIMS
Project Status: Complete
Organisation(s): Queen’s University Belfast