Epidemiological analysis of the association between two congenital anomalies in the same child: A method for adjusting nonspecific clustering

Abstract

Most of the methods proposed to analyze patterns of congenital anomaly clusters have been criticized because large observed/expected (O/E) ratios are obtained for many defect combinations, suggesting that the association of congenital defects is a generalized and nonspecific tendency. To avoid this problem, two methods have been proposed: (1) the analysis of the association of two defects in the subgroup of infants with only multiple congenital anomaly (MCA) patterns, and (2) a method for adjusting the O/E ratio for the nonspecific tendency of defects to cluster among themselves. However, neither of these methods analyzes the association of only two particular defects in a child who apparently has no other anomalies apart from the two that are being analyzed. Here we present a method that analyzes the association of two particular defects, when they are the only defects observed in the child. The method is tested on the Spanish Collaborative Study of Congenital Malformations (ECEMC) data base. The proposed method also controls for the tendency of nonspecific defect clustering. © 1996 Wiley‐Liss, Inc.

Co‐occurring defect analysis: A platform for analyzing birth defect co‐occurrence in registries

“…Defects that co-occurred with non-syndromic spina bifida were also consistent with previous literature (Juranek & Salman, 2010;Parker, Yazdy, Mitchell, Demmer, & Werler, 2014;Stevenson, Seaver, Collins, & Dean, 2004). Three of the previous studies that have used Khoury et al's adjusted O/E ratios have included spina bifida in analyses (Castilla et al, 1998;Prieto & Martinez-Frias, 1996). Elevated adjusted O/E ratios (>5.0) were reported for the following co-occurring defects as part of a schisis association study: anencephaly, encephalocele, and omphalocele .…”

“…In the Texas Registry, we were able to rapidly analyze 2-, 3-, 4-, and 5-way combinations of 175 defects in over 200,000 infants with birth defects. Khoury et al's (1990) O/E ratio adjustment methods have been implemented previously for some 2-way combinations (Castilla, Lugarinho, da Graca Dutra, & Salgado, 1998;Martinez-Frias, 1996;Martinez-Frias, Frias, Bermejo, Rodriguez-Pinilla, & Urioste, 1997;Prieto & Martinez-Frias, 1996) and CODA now provides a platform for implementation of higher-order combinations in large- scale data sets (e.g., registries, medical claims data) for a large number (e.g., hundreds) of unique birth defect types. In our analyses, defects co-occurring with trisomy 21 were consistent with previous literature describing other defects present among infants with trisomy 21 (Buchin, Levy, & Schullinger, 1986;Cleves et al, 2007;Holmes, 2014;Källén, Mastroiacovo, & Robert, 1996;Stoll, Dott, Alembik, & Roth, 2015).…”

“…Elevated adjusted O/E ratios (>5.0) were reported for the following co-occurring defects as part of a schisis association study: anencephaly, encephalocele, and omphalocele . These older studies limited analyses to less than 30 defects to examine specific pairwise co-occurrence with polydactyly (Castilla et al, 1998) or characterize co-occurrence in a small subset of defects Prieto & Martinez-Frias, 1996). CODA can facilitate large-scale analyses of cooccurrence in contemporary databases.…”

Clinical/epidemiological analysis of congenital anomalies associated with diaphragmatic hernia

“…To analyze the data, we used the method described by Prieto and Martinez-Frias [1996]. As we study only malformed infants, the method requires the cases be distributed by presence or absence of CDH as it is shown in Table I.…”