Jack K Leiss1. 1. Center for Health Research, Constella Group, LLC, 2605 Meridian Parkway, Suite 200, Durham, NC 27713, USA. jleiss@constellagroup.com
Abstract
OBJECTIVES: Numerous studies have used maternally linked birth records to investigate perinatal outcomes, maternal behaviors, and the quality of vital records birth data. Little attention has been given to assessing errors in the linkages and to understanding how such errors affect estimates derived from the linked data. The author developed a framework for conceptualizing maternal linkage error and measures for quantifying it, and examined the behavior of the new measures in a maternally linked file. METHODS: Linkage errors were conceptualized as misclassification, with the classes being the maternal sets (records classified as representing different births to the same woman). The true linkage proportion, analogous to sensitivity, was used to capture the degree to which all of a woman's births were assigned to a single maternal set; the false linkage proportion, analogous to specificity, was used to capture the degree to which the assigned maternal sets combined births from different women. The behavior of the two proportions was examined by introducing increasing degrees of linkage error into a maternally linked file. RESULTS: Both measures indicated greater misclassification with increasing simulated linkage errors. CONCLUSIONS: The new measures may be a useful tool for assessing the quality of maternally linked data, as well as other types of linked records where the linkages are within a single file. This is a necessary step towards developing methods for addressing misclassification bias in studies of maternally linked records through sensitivity analysis, adjustment, and other means.
OBJECTIVES: Numerous studies have used maternally linked birth records to investigate perinatal outcomes, maternal behaviors, and the quality of vital records birth data. Little attention has been given to assessing errors in the linkages and to understanding how such errors affect estimates derived from the linked data. The author developed a framework for conceptualizing maternal linkage error and measures for quantifying it, and examined the behavior of the new measures in a maternally linked file. METHODS: Linkage errors were conceptualized as misclassification, with the classes being the maternal sets (records classified as representing different births to the same woman). The true linkage proportion, analogous to sensitivity, was used to capture the degree to which all of a woman's births were assigned to a single maternal set; the false linkage proportion, analogous to specificity, was used to capture the degree to which the assigned maternal sets combined births from different women. The behavior of the two proportions was examined by introducing increasing degrees of linkage error into a maternally linked file. RESULTS: Both measures indicated greater misclassification with increasing simulated linkage errors. CONCLUSIONS: The new measures may be a useful tool for assessing the quality of maternally linked data, as well as other types of linked records where the linkages are within a single file. This is a necessary step towards developing methods for addressing misclassification bias in studies of maternally linked records through sensitivity analysis, adjustment, and other means.
Authors: A A Herman; B J McCarthy; J M Bakewell; R H Ward; B A Mueller; N E Maconochie; A W Read; P Zadka; R Skjaerven Journal: Paediatr Perinat Epidemiol Date: 1997-01 Impact factor: 3.980
Authors: A B Miller; G R Howe; G J Sherman; J P Lindsay; M J Yaffe; P J Dinner; H A Risch; D L Preston Journal: N Engl J Med Date: 1989-11-09 Impact factor: 91.245
Authors: Jack K Leiss; Denise Giles; Kristin M Sullivan; Rahel Mathews; Glenda Sentelle; Kay M Tomashek Journal: Ann Epidemiol Date: 2010-01 Impact factor: 3.797