Statistical Methods
For continuous demographic, treatment and outcome factors, means
(standard deviations (SDs)) and medians (ranges) are reported, and for
discrete factors, frequencies (percentages) are reported. The impact of
paternal and maternal BMI on infant birthweight was assessed using
quantile regressions, adjusting for baby sex (male or female),
gestational age, delivery method (vaginal or caesarean), transfer method
(IVF or ICSI), maternal age, paternal age and parental SEIFA score.
Non-linear associations using restricted cubic splines (knots at
5th, 10th 50th,
90th and 95th percentiles) were
included for gestational age and maternal BMI. An interaction between
the maternal and paternal BMI factors (both linear) was also included.
These quantile regressions were constructed for the
5th, 10th, 50th,
90th and 95th weight percentiles
based on Australian specific birthweight standards 13(Figure S2). Multiple imputation using chained equations (100 datasets
were imputed each with 100 iterations) was employed to account for the
substantial missing paternal BMI data. Analyses were performed in R
(version 3.6.3) using the mice and rms packages. P-values
of 0.05 were considered statistically significant.