A multidimensional threshold model for multiple congenital abnormalities

Analysis of the data of 2.762 infants with "unidentified" (i.e., no identified as dysmorphological entities) multiple congenital abnormalities from the Hungarian Congenital Malformation Registry, 1970-1976 showed that the majority of congenital abnormalities in newborns with multiple congenital abnormalities was not due to random combinations. The ratio of expected random combination and observed causal effect was 1: 6, 1: 225-1: 14,000 and 1: 2.10(6) in two, three, four, and five or more congenital abnormalities, respectively. Our multidimensional threshold model supposes specified liabilities for each congenital abnormality, and the correlation structure of these liabilities explains nearly all associations of congenital abnormalities in unidentified multiple congenital abnormalities. According to this hypothesis, the expected and the observed occurrences do not differ significantly either in the sum or in the various possible threefold associations of congenital abnormalities. The biological reasons for the phenomenon are being studied.