Statistical analysis
Statistical analyses included all available stillbirth cases (n=9,404),
controls (n=18,808) that were 2:1 matched with cases, and relatives of
the parents of cases and controls. We compared maternal demographic
characteristics of stillbirth cases to those of live birth controls
using t-tests for continuous variables and chi-square tests for
categorical variables.
Determination of stillbirth familial aggregation
We used the genealogical index,6 which considers all
relationships in the UPDB, to identify pedigrees of stillbirth index
cases (i.e., probands). To capture the maternally and paternally
inherited risk of stillbirth, we designated the offspring rather than
mothers as probands. We estimated the Familial Standardized Incidence
Ratio (FSIR) to determine the level of stillbirth familial aggregation
and identify high-risk pedigrees.23 The FSIR compares
the observed incidence of stillbirth in a pedigree to its expected
incidence in the UPDB population, weighting the familial risk
contribution of each relative in a pedigree by the kinship coefficient
(i.e., the probability that a relative shares an allele with the proband
through a common ancestor).24 Our a priorispecified criteria for familial aggregation and a high-risk pedigree
were pedigrees with FSIR ≥2 and P-value
<0.05.23 We evaluated the skewness of FSIR
among high-risk pedigrees by examining its distribution. We presented
the top 10 high-risk pedigrees sorted by their FSIR and founders to
eliminate any multiple founders claiming the same descendants through
marriage.25 To avoid potential biased estimates of
stillbirth familial aggregation due to multiple partnerships, we
excluded multiple partnerships.