Prediction of Live Birth After Intrauterine Insemination Using Hunault and Revised Natural Conception Prediction Models

Authors: Lara Janssens; E. Roelant; A. Sugihara; D. De Neubourg

Intrauterine insemination (IUI) is commonly used as a first-line treatment for couples with unexplained or mild male factor infertility. Natural cycle IUI is a less invasive treatment option, and accurate prediction of live birth chances could support individualized treatment decisions. However, no prediction model has been specifically developed for live birth outcomes after natural cycle IUI. This prospective cohort study evaluated whether existing natural conception prediction models could predict live birth following natural cycle IUI.

The study used data from the ID-trial conducted at the Center for Reproductive Medicine of Antwerp University Hospital. Between October 2017 and October 2020, 120 couples who had experienced at least 12 months of infertility and were starting natural cycle IUI were recruited. Participants underwent up to four IUI cycles during follow-up. The Hunault model score and the revised Bensdorp model score, originally developed to predict natural conception, were calculated using baseline patient characteristics. Eligible women were aged 18–40 years, had regular menstrual cycles, a body mass index (BMI) of 18–30, confirmed tubal patency, and a normal uterine cavity. Male partners were aged 18–65 years, had a BMI of 18–30, and had either normal semen parameters or mild male subfertility. The association between model scores and live birth was assessed using logistic regression analyses at both cycle and patient levels. Predictive performance was evaluated using receiver operating characteristic (ROC) curves. A total of 114 patients underwent at least one IUI cycle, and 30 patients completed four cycles during follow-up. The cumulative live birth rate (CLBR) after four cycles was 28.1%. The mean Hunault score was 18.1 ± 5.5, and the mean revised Bensdorp score was 74.7 ± 8.9.

Higher Hunault scores were associated with lower odds of live birth at both the cycle level (odds ratio [OR] 0.87, 95% confidence interval [CI] 0.81–0.95; p = 0.0009) and patient level (OR 0.86, 95% CI 0.78–0.94; p = 0.001). In contrast, higher revised Bensdorp scores were associated with increased odds of live birth at the cycle level (OR 1.06, 95% CI 1.01–1.11; p = 0.01) and patient level (OR 1.08, 95% CI 1.02–1.14; p = 0.01). The ability of both models to distinguish between patients who did and did not achieve live birth was modest. The area under the curve (AUC) was 0.70 (95% CI 0.59–0.81) for the Hunault model and 0.65 (95% CI 0.47–0.81) for the revised Bensdorp model, with no significant difference between the two ROC curves (p = 0.23). The observed cumulative live birth rate of 28.1% corresponded to a Hunault score of 17.0 and a revised Bensdorp score of 75.6. A cumulative live birth probability below 10% was observed in couples with a Hunault score of ≥25.1 or a revised Bensdorp score of ≤58.1. The study was limited by its cohort size and the fact that participants were from a Belgian population similar to the populations used to develop the original models, which may limit applicability to other patient groups. Larger prospective studies are needed to further evaluate and improve prediction models for IUI outcomes.

In conclusion, existing natural conception prediction models, including the Hunault and revised Bensdorp models, may provide some guidance for estimating live birth chances after natural cycle IUI, but their overall predictive accuracy remains limited. Their simple clinical application may support prognosis-based counseling while further refinement is needed.

ESHRE 2026, July 5^th-8^th, London, UK.