Background While hemophilia is traditionally considered an X-linked recessive disorder mainly affecting males, female hemophilia carriers (FHCs), are increasingly recognized to experience abnormal bleeding, especially when factor VIII (FVIII) or factor IX (FIX) levels below 50 IU/dL. These individuals are now classified as women and girls with hemophilia (WGH) by the ISTH/SSC. However, in China, clinical recognition and testing for FHCs and WGH remain limited due to the high cost and inaccessibility of genetic testing. There is an unmet need for affordable, accessible screening tools to identify high-risk women, guide further diagnosis, and enable timely prevention and management in resource-limited regions.

Objectives To develop and validate a low-cost prediction model incorporating ISTH bleeding assessment tool (ISTH-BAT) scores, FVIII:C or FIX:C levels, and age, aiming to identify FHCs and classify WGH among females from families affected by hemophilia A and B.

Methods We prospectively enrolled 318 females aged 5–80 years from families with hemophilia A or B in central China, primarily from Jiangxi Province. and categorized them into three groups:

  • Group A: confirmed carriers (n=129; 100 with hemophilia A, 29 with hemophilia B)

  • Group B: femaleswith a family history of hemophilia but unknown carrier status (n=104)

  • Group C: healthy non-carrier controls (n=85)

All participants completed the ISTH-BAT, and coagulation factor VIII (FVIII:C) or IX (FIX:C) activity was measured using standard one-stage clotting assays. Age, family history, and other clinical data were collected via structured questionnaires. A multivariable logistic regression model was constructed using 80% of the dataset for training and 20% for testing, with 5-fold cross-validation. Predictor variables included clotting factor level (FVIII:C or FIX:C), total BAT score, and age. Model discrimination was assessed using area under the ROC curve (AUC), and performance metrics included sensitivity, specificity, and overall accuracy. Calibration was evaluated using calibration plots and Hosmer–Lemeshow tests.

Results The final model achieved an AUC of 0.94 (95% CI: 0.89–0.98) in the test dataset, with an overall accuracy of 94.7%, sensitivity of 94.4%, and specificity of 100%. Factor activity was the strongest independent predictor (OR per 10 IU/dL = 0.32; 95% CI: 0.18–0.55; p<0.001), followed by BAT score (OR per point = 1.41; 95% CI: 1.18–1.68; p=0.002) and younger age (OR per 10-year increase = 0.77; 95% CI: 0.61–0.96; p=0.021). The tool requires only basic clinical data and inexpensive lab tests, offering a practical screening method before genetic testing.

Conclusions We developed a low-cost prediction model combining BAT scores, age and coagulation factor levels for identifying FHCs and WGH in central China. The tool helps identify at-risk women, supports shared decision-making, facilitates early identification and encourages appropriate genetic testing, particularly in under-resourced areas. To our knowledge, this represents the first prediction model specifically tailored for female hemophilia screening in a Chinese population. This study offers a scalable framework for broader clinical implementation and may support the development of national strategies for carrier screening and counseling. Future work will focus on multi-centre validation and development of a digital tool for clinical implementation.

Disclosures Topics: Hemophilia, Female Carriers, BAT Score, Coagulation Factors, Screening Tool, Bleeding Risk, Prediction Model

Author notes *Asterisk with author names denotes non-ASH members.

This content is only available as a PDF.
2025
Sign in via your Institution