Taking blood thinning drugs for long periods can increase the risk of internal bleeding in some patients. Now, scientists at KAIMRC have developed a new model for predicting the risk of bleeding in Arab patients taking direct oral anticoagulants (DOACs).

Existing bleeding risk models developed in other countries are not directly transferrable to Arab populations. Along with co-workers across Saudi Arabia, Maha Al Ammari and Khizra Sultana at KAIMRC have developed a new model called the GAMARLSTC score based on 15 variables. They validated it using data from a cohort of 1722 patients with non-valvular atrial fibrillation who were receiving DOACs in hospitals and clinics in Riyadh.     

“Most previous studies assessing bleeding risk used models like HAS-BLED score, which was originally validated in patients using warfarin, but not in patients using the newer anticoagulants,” says Sultana. “Our new model is necessary for the Arab population because factors such as eating habits, family relationships, and lifestyle are very different to Western cultures. Many of these factors are proven to be associated with diseases that lead to the use of DOACs.”

DOACs can limit strokes and blood clots in patients with non-valvular atrial fibrillation, with some DOACs showing superior action to warfarin. Bleeding risk can vary with physiological factors such as age, gender, ethnicity, comorbidities and uncontrolled blood pressure, alongside lifestyle choices. The team determined the best-fit parameters for their model using a resampling approach and a five-fold cross validation process. This analysis helped them fit the model to accurately predict bleeding risk in an Arab population. 

The GAMARLSTC model is encoded in an Excel score calculator with a simple interface where users can enter patient data and produce an instant risk score — a green colour code indicates the use of DOACs is safe in that patient, yellow indicates a moderate or borderline risk, and red indicates a high risk of bleeding. This means physicians get an instant notification of an individual’s risk score.

“The score applies to people who do not consume alcohol,” says Sultana. “The model works from easily measured and inputted variables and does not require genetic details or laboratory data.”

The team aims to use this model widely in the Saudi population to confirm that it works on a wider scale. “This will give us valuable feedback to improve the model if required, and we hope that this model will also be validated in other Arab populations,” says Sultana.

References