Reflections on modern methods: linkage error bias.

James C Doidge,Katie L Harron

doi:10.1093/ije/dyz203

Abstract

Linked data are increasingly being used for epidemiological research, to enhance primary research, and in planning, monitoring and evaluating public policy and services. Linkage error (missed links between records that relate to the same person or false links between unrelated records) can manifest in many ways: as missing data, measurement error and misclassification, unrepresentative sampling, or as a special combination of these that is specific to analysis of linked data: the merging and splitting of people that can occur when two hospital admission records are counted as one person admitted twice if linked and two people admitted once if not. Through these mechanisms, linkage error can ultimately lead to information bias and selection bias; so identifying relevant mechanisms is key in quantitative bias analysis. In this article we introduce five key concepts and a study classification system for identifying which mechanisms are relevant to any given analysis. We provide examples and discuss options for estimating parameters for bias analysis. This conceptual framework provides the ‘links’ between linkage error, information bias and selection bias, and lays the groundwork for quantitative bias analysis for linkage error.

Highlights

With advances in computing technology and increasing use of secondary data for research, there has been rapid growth in analysis of linked data but little corresponding acknowledgement of the statistical problems introduced by linkage error: missed links between records relating to the same entity and false links between records relating to different entities (Figure 1)
Linkage error can manifest in many ways: as missing data, measurement error and misclassification, unrepresentative sampling, or as a special combination of these that is specific to analysis of linked data: the merging and splitting of people that can occur when two hospital admission records are counted as one person admitted twice if linked and two people admitted once if not
Linkage error can lead to information bias and selection bias; so identifying relevant mechanisms is key in quantitative bias analysis

Summary

Introduction

With advances in computing technology and increasing use of secondary data for research, there has been rapid growth in analysis of linked data but little corresponding acknowledgement of the statistical problems introduced by linkage error: missed links between records relating to the same entity (usually a person) and false links between records relating to different entities (Figure 1). Splitting and merging can be operationalized as a combination of varied probabilities of selection and misclassification or measurement error in variables that are meaningfully interpreted or otherwise derived from multiple records. A set of hospital admission records, for example, may contain a uniquely identified sample if the unit of observation is admissions, but not if the unit is people, and not if the unit is sequenced events such as people’s first admission (because first admissions cannot be identified until they have been linked to any other relevant admissions) These three key concepts provide three questions in identifying the manifestations of linkage error: (i) are links being meaningfully interpreted? The previous section explored qualitative differences in the way that linkage error can manifest in different analyses: as misclassification and measurement error, varied probabilities of selection into an analysis, missing data and splitting and merging. Each describes a different linkage structure for combining two data sources, and highlights the relevance of the caveats described above and of any internal linkage within each data source that may be required

Discussion

Findings

Limitations