Warfarin decreases stroke risk in atrial fibrillation patients, with efficacy and safety impacted by the quality of warfarin control, as measured by time in therapeutic range (TTR). Stroke and bleed risk scores are calculated prior to commencing warfarin, so it would be beneficial if these scores also identified likely warfarin control. Some studies have investigated CHADS2, CHA2DS2VASc, and HASBLED individually for this purpose, but application of all scores to diverse ethnic populations and at sites with differing overall control has not been investigated. The aim of this study was to determine if these commonly used risk scores could identify poor warfarin control.

Retrospective data was collected for non-valvular AF patients receiving warfarin between January and June 2014 in Australia and Singapore. Patient data was used to calculate TTR and risk scores. Mean TTR was used for analysis and comparison to categorised scores.

There were 3199 patients in Australia and 1171 in Singapore. At both sites, mean TTR decreased according to HASBLED category, and there was a statistically higher percentage of patients achieving a TTR > 65% in the low HASBLED category. The association between HASBLED scores and TTR was independent of lower dosing in higher risk patients, particularly in Australia. No significant differences were found in mean TTR according to CHADS2 at either site. Time in therapeutic range significantly decreased according to high CHA2DS2VASc category in Singapore, but no differences were found in Australia.

Of the bleed and stroke risk models, HASBLED is most suitable to identify a patient’s potential TTR and ability to achieve TTR > 65%. A high HASBLED score may assist prescribers in determining potential suitability to warfarin, and assist prescribers in deciding on the most suitable anticoagulant for patients.