A key finding in the study was that the stent size selected intraoperatively based on the angle of the cornus was clearly associated with postoperative pregnancy outcomes. Researchers analyzed nine stent sizes (from XXXS to XXXL) into ≤ XS and > XS groups. Results showed that patients in the ≤ XS group (i.e., extremely small stent size, corresponding to ICDs about 20-22 mm) had a significantly lower clinical pregnancy rate than the > XS group.

After accounting for multiple confounding factors such as age, history of uterine scarring, and previous adhesiolysis procedures, multivariate logistic regression analysis confirmed that the adjusted odds ratio (aOR) for clinical pregnancy in ≤ XS group was 0.447 (95% confidence interval 0.255-0.786, P=0.005). This means that compared to the > XS group, patients in the ≤ XS group had a 55.3% lower chance of clinical pregnancy within one year postoperatively.

Figure: Evaluating candidate thresholds based on stent size categories for predicting clinical pregnancy within one year after the last hysteroscopy

Although incorporating stent size classification into traditional predictive models results in limited improvement in model discrimination (AUC only increases by 0.010), its clinical value is reflected in significant differences in absolute risk. Data show that between the ≤ XS group and the > XS group, the absolute risk difference (ARD) for clinical pregnancy within one year after the last hysteroscopy was 21.2% (95% confidence interval 9.6%-32.2%, P<0.001). This means that compared to the > XS group (clinical pregnancy rate 64.6%), the ≤ XS group (clinical pregnancy rate 43.4%) had an absolute 21.2% reduction in pregnancy chances. This value corresponds to a number of people needing treatment (NNB) of 4.7, indicating that this indicator has clear clinical risk stratification value.

Figure: Absolute difference in clinical pregnancy probability within one year after the last hysteroscopy between the ≤ XS group and the > XS group

The study further conducted subgroup analyses based on pregnancy mode (natural pregnancy and assisted reproductive technology (ART). The results showed that whether patients attempted natural pregnancy or received ART, the association between stent size classification (≤XS vs >XS) and clinical pregnancy was statistically significant. In the natural pregnancy subgroup, the aOR was 0.394 (95% CI 0.183-0.846, P=0.017); In the ART subgroup, the aOR was 0.451 (95% CI 0.205-0.994, P=0.048). This consistent correlation suggests that the impact of ICDs may be independent of pregnancy mode.

To eliminate potential confounding bias, researchers employed various sensitivity analysis methods such as Propensity Score Matching (PSM) and Inverse Probability Processing Weighting (IPTW). All analyses consistently showed that the clinical pregnancy risk in the ≤ XS group was significantly lower than in the > XS group.

Figure: The forest diagram shows the association between stent size classification (≤XS vs >XS) and clinical pregnancy within one year after the last hysteroscopy

The study also evaluated the consistency between intraoperative stent size classification and postoperative ultrasound measurements (ICD). The results showed that 42.9% of the two methods were completely consistent, and 84.1% were consistent within ± category (weighted κ=0.539, P<0.001), indicating moderate consistency between the two methods. This supports the rationale of using stent size as an intraoperative approximate value for ICD.

For secondary outcomes, the study found a significant association between stent size classification (≤XS vs >XS) and live birth after first clinical pregnancy within one year after the last hysteroscopy (aOR=0.500, 95% CI 0.280-0.893, P=0.019). However, the association with cumulative live birth rate over 2 years was not statistically significant (aOR=0.606, 95% CI 0.339-1.083, P=0.091). Researchers pointed out that since live birth outcomes are influenced by more downstream factors (such as obstetric management and follow-up treatment decisions), these findings should be interpreted as exploratory results.

In an analysis of 608 women with complete gestational time data, the Kaplan-Meier survival curve showed that the probability of cumulative clinical pregnancy within one year after the last hysteroscopy in the ≤ XS group was significantly lower than in the > XS group (log-rank P=0.002). Multivariate Cox regression analysis also confirmed that the risk ratio (aHR) for achieving clinical pregnancy in the ≤ XS group was 0.630 (95% CI 0.439-0.903, P=0.012).

The authors clearly point out that this is a single-center retrospective study and may have selection bias and residual confounding, with limited extrapolation of results. Additionally, the XS threshold for stent size classification was exploratory determined in this dataset and may be overfitted, requiring validation in independent external cohorts. The study also did not collect detailed ART-related variables (such as embryo quality, male factors, etc.), which may affect interpretation of ART subgroup results.