Construction of a three-level early warning model for moderate to severe ovarian hyperstimulation syndrome in assisted reproductive technology
10.3760/cma.j.cn101441-20240818-00304
- VernacularTitle:辅助生殖技术中中重度卵巢过度刺激综合征的三级风险预警模型的构建
- Author:
Zhufeng WU
1
;
Jun LIU
1
;
Chunlin LIU
1
;
Qinhong LUO
1
;
Fengxiang LI
1
;
Xiu ZHANG
1
Author Information
1. 广州医科大学附属妇女儿童医疗中心生殖医学中心，广州 510000
- Publication Type:Journal Article
- Keywords:
Reproductive technology, assisted;
Fertilization in vitro;
Embryo transfer;
Ovarian hyperstimulation syndrome;
Risk early warning;
Prediction model
- From:
Chinese Journal of Reproduction and Contraception
2025;45(2):162-171
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To construct and validate a three-level early warning model of moderate to severe ovarian hyperstimulation syndrome (OHSS) in assisted reproductive technology.Methods:A case-control study was conducted. Totally 10 181 infertility patients who underwent in vitro fertilization treatment in Reproductive Medicine Center, Women and Children's Medical Center Affiliated to Guangzhou Medical University from April 2013 to April 2023 were retrospectively analyzed and divided into modeling group (8 145 cases) and validation group (2 036 cases) by random number table method. The clinical data of the two groups were analyzed, and the risk factors affecting the occurrence of moderate and severe OHSS after oocytes retrieval were screened by multi-factor logistic regression analysis. The early warning model was established and the column diagram was drawn at three nodes which were before ovarian stimulation, before trigger and 3 d after oocyte retrieval. The receiver operating characteristic (ROC) curve and calibration curve were used to verify the models. Results:The antral follicle count (AFC, OR=1.045, 95% CI: 1.020-1.071, P<0.001), anti-Müllerian hormone (AMH)>3.36 μg/L ( OR=7.135, 95% CI: 2.084-24.432, P=0.002) and number of cycles ( OR=0.149, 95% CI: 0.022-1.026, P=0.049) were included in the pre-stimulation prediction model. AFC ( OR=1.046, 95% CI: 1.018-1.074, P=0.001), AMH>3.36 μg/L ( OR=5.780, 95% CI: 1.661-20.116, P=0.006), gonadotropin releasing hormone-agonist protocols ( OR=3.895, 95% CI=1.913-7.931, P<0.001), estrogen peak≥18 350 pmol/L ( OR=2.258, 95% CI: 1.092-4.666, P=0.028), the number of follicles with a diameter of ≥10 mm>20 ( OR=2.377, 95% CI: 1.092-5.172, P=0.029) were included in the pre-trigger prediction model. AMH>3.36 μg/L ( OR=8.374, 95% CI: 2.417-29.019, P=0.001), estrogen peak≥18 350 pmol/L ( OR=3.947, 95% CI: 1.533-10.167, P=0.004), total number of oocytes retrived ( OR=1.042, 95% CI: 0.996-1.090, P=0.025), abdominal distension ( OR=60.181, 95% CI: 22.515-160.854, P<0.001), fresh transplantation ( OR=21.766, 95% CI: 7.119-66.544, P<0.001), human chorionic gonadotropin trigger ( OR=17.752, 95% CI: 3.993-78.924, P<0.001) were included in the prediction model of 3 d after oocyte retrieval. The areas under ROC curves of the three models were 0.830 (95% CI: 0.782-0.878), 0.859 (95% CI: 0.812-0.906) and 0.948 (95% CI: 0.919-0.977), respectively. The areas under ROC curves of the validation groups of the three models were 0.922 (95% CI: 0.880-0.965), 0.936 (95% CI: 0.886-0.986), and 0.971 (95% CI: 0.938-0.999), respectively. The calibration curve indicated that the early-warning evaluation model has good stability. Conclusion:The three-level early warning model of moderate and severe OHSS has good differentiation, reliable predictability and clinical practicability, which is conducive to the dynamic and continuous assessment of the risk of moderate and severe OHSS, adjustment of treatment plan at any time, and timely adoption of effective preventive measures.
