Methodological concerns in the feasibility study on ultrasound pelvimetry for breech birth

We commend Lia et al. [1] for their innovative study introducing transperineal ultrasound (TPU) to measure the anteroposterior mid-pelvic diameter (AMD) in breech presentation. While the results are promising, several methodological issues undermine the validity of the predictive model and its clinical implications. First, the multivariable logistic regression analysis risks severe overfitting. With only 17 events (intrapartum cesarean sections), including four variables—AMD, birthweight, head circumference, and obstetric conjugate (OC)—violates the standard of 10 events per variable. This low event-to-variable ratio (~4.25) likely inflates the effect size (adjusted odds ratio 0.25 per 1 cm AMD increase), rendering the model unreliable for clinical application, while established statistical guidelines recommend approximately 10 events per variable (EPV) to ensure the reliability of multivariable logistic regression models [2-5]. Second, the comparison of AMD's predictive performance (AUC 0.77) against traditional diameters such as the OC (AUC 0.60) is biased by study design. The OC was used as a criterion for offering vaginal birth (cutoff ≥ 12.0 cm), restricting its variability and artificially attenuating its predictive power. In contrast, AMD was not constrained, leading to an inequitable comparison that overstates AMD's superiority. Furthermore, the clinical translation remains vague. Although Figure 5 suggests risk stratification by AMD quartiles, no actionable cutoff is proposed for patient counseling. The model's development without external validation or sample size justification limits its generalizability. In conclusion, while TPU is a feasible tool, the study's statistical limitations necessitate cautious interpretation. We urge validation in a larger cohort with rigorous design to ensure robust, clinically relevant predictions before widespread adoption. Medical Science and Technology Project of Zhejiang Province (No: 2024KY755). We declare no competing interest.