Related Experiment Videos
Effects of record linkage errors on registry-based follow-up studies
H Brenner1, I Schmidtmann, C Stegmaier
1University of Ulm, Department of Epidemiology Albert-Einstein-Allee 43, Germany.
Statistics in Medicine
|January 9, 1998
Summary
Record linkage errors in disease registries can significantly impact epidemiological studies. Homonym errors underestimate patient survival, while synonym errors overestimate disease incidence, affecting data reliability.
Area of Science:
- Epidemiology
- Biostatistics
- Public Health
Background:
- High-quality population-based disease registries are crucial for epidemiological research.
- Existing literature lacks systematic methodologic analysis of record linkage errors' impact on disease registries.
- Reliable record linkage is fundamental for accurate disease registration and subsequent analysis.
Purpose of the Study:
- To develop algebraic models quantifying the effects of record linkage errors on disease registries.
- To assess the influence of linkage errors on survival monitoring and incidence follow-up.
- To provide a quantitative understanding of how linkage inaccuracies affect epidemiological data.
Main Methods:
- Development of algebraic models to describe record linkage errors.
- Analysis of two common linkage error types: homonyms and synonyms.
- Simulation and numerical examples to illustrate the magnitude of effects.
Main Results:
- Homonym errors (linking distinct individuals) lead to underestimation of patient survival.
- Homonym errors result in overestimation of disease incidence in external cohort follow-up.
- Synonym errors (failing to link the same individual) produce opposite effects on survival and incidence.
Conclusions:
- Record linkage errors introduce systematic biases in disease registry data.
- Understanding and quantifying these errors is essential for accurate epidemiological interpretation.
- Methodological advancements are needed to mitigate the impact of linkage errors on public health surveillance.