INTRODUCTION:
About 10% of annual births are preterm, and up to 34% of infant deaths are attributable to prematurity. Although preterm birth (PTB) is heritable, our understanding of the genetic basis of PTB remains limited. We hypothesize that characterizing comorbidities will yield insights into PTB genetics.
METHODS:
Using international classification of diseases (ICD9/10) codes, we identified 53,063 females with deliveries in Vanderbilt's electronic health records (EHRs). 298 codes were enriched for PTB by chi-squared test after multiple testing correction (P < 3.2*10^-5) and grouped using hierarchical clustering. To explore the genotype-phenotype relationships, a phenome wide association study (PheWAS) was conducted for single nucleotide polymorphisms (SNPs) associated with gestational age. These SNPs were also queried in the UK Biobank (UKBB) using GeneATLAS for phenotype associations.
RESULTS:
ICD9 codes enriched for PTB reproduced known (twin pregnancy, cervical insufficiency) and unexpected associations across multiple systems (hypertension, renal failure, sickle cell). PheWAS resulted in significant association with hematemesis (rs1735537, P=5.4*10^-7, OR=1.79) and schizophrenia (rs6556350, P=2.8*10^-6, OR=0.48). Attention deficit/hyperactive disorder (P=7.9*10^-6) was nominally significant. In the UKBB, height (rs3820282, P>1*10^-40), adiposity traits, and female reproductive tract disorders were significantly associated (P>1*10^-8) with gestational length.
CONCLUSION:
Using billing codes identified a large cohort of females with deliveries within the EHRs. ICD codes enriched for PTB captured several unexpected phenotypes that warrant further study. The PheWAS analysis suggests a shared genetic basis between several traits and gestational length. In particular, height is known to influence birth timing. SNP rs3820282 suggests a pleiotropic locus influence both traits.