Linked Birth/Infant Death Files available from the National Center for Health Statistics identify an infant as a twin, but do not identify twin pairs.
An algorithm based on maternal, paternal, and infant characteristics has been used to identify twin pairs, but the validity of this algorithm has never been tested.
The Missouri linked birth/infant death file from 1980 to 1990 identifies twin pairs by a sequence number.
The authors tested the rate and accuracy with which the algorithm identified true pairs in the Missouri file and whether estimates of risk and possible risk factors calculated from pairs of twins identified by the algorithm agreed with these characteristics as calculated from known twin pairs.
The algorithm identified 96% (8,273 of 8,620) of true pairs and one false pair.
Mots-clés Pascal : Mortalité, Morbidité, Nourrisson, Jumeau, Facteur risque, Epidémiologie, Méthodologie, Algorithme, Analyse mathématique, Homme
Mots-clés Pascal anglais : Mortality, Morbidity, Infant, Twin, Risk factor, Epidemiology, Methodology, Algorithm, Mathematical analysis, Human
Notice produite par :
Inist-CNRS - Institut de l'Information Scientifique et Technique
Cote : 94-0459506
Code Inist : 002B30A01A1. Création : 199406.