Statistics
A descriptive analysis was performed using absolute and relative frequencies for categorical variables and mean value with standard deviation for continuous variables. Then, we performed Pearson’s Chi-Square and t-test, as appropriate, of clinical and demographic characteristics of women between the groups. We used logistic regression to evaluate an association between the mode of conception and preterm delivery in infertile women after reproductive surgery. The logistic regression model was adjusted for variables which presented as significantly different between the spontaneous and IVF/ICSI group in descriptive analysis. The association was presented as adjusted odds ratio (aOR) with 95 % confidence interval (95% CI).
For the case-control study of the effect of reproductive surgery on preterm delivery, we used the propensity score (PS) method. Controls were selected from among all fertile women (n=100765) who delivered in the same period using PS matching. In the PS model, we included maternal age as a continuous variable, BMI as a categorical variable (underweight, normal weight, obese), and dichotomous variables: previous conization, chronic diseases (pre-pregnancy hypertension, pre-pregnancy diabetes, kidney, heart, thyroid and/or mental disease), previous preterm delivery, smoking, and multiple pregnancies. We performed 1:1 matching with an optimal matching algorithm with a calliper width of the linear predictor of 0.1 standard deviations. The balance between the control and treated group was considered well when the value of the standardized difference was <0.1. The effect of reproductive surgery on preterm delivery was evaluated in the PS-matched sample by Pearson’s Chi-Square test and presented as the odds ratio (OR) with 95% confidential interval (95% CI).
Considering an increase in preterm delivery rate by 30% in infertile women, within the limit of the number of treated women, the study’s power was 0.78 at a 5% Type I error rate.
For statistical calculations, we used statistical program IBM SPSS Statistics v25 and R18 with R-package ”Matchlt”19. For all calculations, a two-sided probability (p ) value < 0.05 was considered statistically significant.