Statistical Analysis
R statistical software was used to perform the statistical analyses. Due
to non-normal distribution of the data an unpaired Wilcoxon test was
applied to compare the clinical characteristics of the two study groups.
Analysis of this study involved measurement of sHLA-G i.e., a dependent
variable over time (independent variable) during the course of pregnancy
in the pregnant mothers. It represents a classical design of repeatedly
measured longitudinal data. Linear mixed effect model was used to
analyze this longitudinal data in order to determine changes in the
sHLA-G levels during pregnancy in the mothers delivering SGA and those
delivering AGA. This method of analyzing the data was selected as it
allowed for the inclusion of multiple measurements (of sHLA-G) per
participant (i.e., one measurement at each time point they were
assessed), and also allowed the inclusion of pregnant women who did not
have all the measurements in the analysis, whether due to
nonavailability of the sera sample at postpartum (42 days to 6 months)
or because sHLA-G levels were undetectable in some mothers at certain
time-points (26 to 28 weeks and delivery) of pregnancy. This model could
also account for the subject-specific variability generated in our study
as all the mothers were not sampled at exactly the same day of
pregnancy.