Statistical analysis
SAS version 9.4 (SAS Institute Inc) was used for all statistical analyses. Differences were considered statistically significant when a 2-sided P < 0.05. The general demographic characteristics of the participating women are presented as percentages (%), and the distributions of the characteristics were compared between women with a history of HA-treated IUAs and matched women with no history of IUAs.
Log-binomial models using generalized estimating equations methods to account for the paired nature of the matched sample were conducted to estimate the relative risks (RRs) and 95% confidence intervals by comparing the proportions of pregnancy complications, placental abnormalities, PPH or adverse birth outcomes in exposed and matched control pregnancies. The final model included the matched cohorts for analyses of each pregnancy complication, placental abnormality, PPH and adverse birth outcome, respectively, with no additional adjustment. Women with no history of IUAs served as the reference.
In a secondary analysis, women with a history of HA-treated IUAs were further divided into three subgroups: women who received one surgery, two surgeries and more than three surgeries before this pregnancy. RRs and 95% CIs for the associations between the number of surgeries and risk of adverse obstetrical outcomes were estimated by unadjusted log-binomial analysis using generalized estimating equations with women with no history of IUAs serving as the reference. P-values for trend were estimated by including the number of hysteroscopic surgeries as continuous variables in the log-binomial models.
Based on the time intervals from the last hysteroscopic surgery to the date of conception, all women with a history of HA-treated IUAs were further divided into two subgroups: ≤ 3 months and > 3 months since the date of the last hysteroscopic surgery. The associations of the time intervals with pregnancy complications, placental abnormalities, PPH or adverse birth outcomes were further examined. The estimated RRs and 95% CIs were reported with women with no history of IUAs serving as the reference.
Finally, to investigate whether the associations of history of HA-treated IUAs with risk of adverse obstetric outcomes differed according to maternal age (< 35, ≥ 35), mode of conception (natural conception, conceived through ART) and maternal parity (nulliparous, multiparous), stratified analyses were conducted and the RRs and 95 % CIs were calculated in each subgroup.