Statistics
Missing data were treated by listwise deletion. Continuous variables
were tested for distributions of normality and described by means and
standard deviations. Categorical variables were described by frequencies
and proportions. We performed separate analyses for the primi- and
multiparous pregnancies. For the variables within each group, we
analyzed differences between cases and controls with t-tests,
Kruskal-Wallis or Chi2-tests. Significance level was
set at 5%. Correlation matrices demonstrated covariation between
weight, length and head circumference of the infant, and only
birthweight was used in logistic regression analyses. We used univariate
conditional logistic regression analyses when assessing associations
between exposure variables and OASI. Subsequently, we built risk-factor
models for OASI by using multivariate conditional logistic regression
analyses progressing with a stepwise procedure. In this analysis we
included variables that were significantly different between the cases
and controls in the univariate analyses. We assessed multicollinearity
by using variance inflation factor (VIF). We assessed interactions
between amniotomy and the following variables: augmentation with
oxytocin, episiotomy, and instrumental delivery by vacuum or forceps.
STATA 15.1 software (STATA, College Station, TX, United States:
StataCorp, 2017) was used for all the analyses.