Variation in Bleeding Risk Estimates Among Online Calculators: A Cross-sectional Study of Apps Used by and for Patients with Atrial Fibrillation
Can Fam Physician. 2022 Apr;68(4):e127-e135. doi: 10.46747/cfp.6804e127.
OBJECTIVE: To assess the variation in bleeding risk estimates and risk stratification between web and mobile applications for patients with atrial fibrillation.
DESIGN: Cross-sectional study.
SETTING: Simulated patient population.
PARTICIPANTS: Hypothetical patient cohorts that encompassed all possible combinations of binary risk factors for each clinical prediction model.
INTERVENTIONS: Twenty-five bleeding risk calculators (18 web apps and 7 mobile apps), each using 1 of 4 clinical prediction models to predict an individual’s 12-month bleeding risk: ATRIA (Anticoagulation and Risk Factors in Atrial Fibrillation), HAS-BLED (hypertension [systolic blood pressure >160 mm Hg]abnormal kidney or liver function, stroke [caused by bleeding]hemorrhage, labile international normalized ratio, elderly [age >65 years]medications [acetylsalicylic acid or nonsteroidal anti-inflammatory drugs] or alcohol [≥8 drinks per week]), HEMORR2HAGES (liver or kidney disease, ethanol abuse, malignancy, elderly [age >75 years]decrease in platelet count or function, risk of rebleeding [history of past bleeding]hypertension [uncontrolled]anemia, genetic factors, risk of excessive falls and stroke) and mOBRI (modified Outpatient Bleeding Risk Index).
MAIN OUTCOME MEASURES: Four simulated cohorts were constructed. The coefficient of variation, relative difference (RD), and 95% CI for annual bleeding risk estimates were calculated for all hypothetical patient cohorts. In addition, pairwise agreement between calculators in low (20%) was determined.
RESULTS: The risk estimates generated by the calculators were imprecise, with coefficients of variation ranging from 14% for HEMORR2HAGES at 64% for mOBRI. A large variation was observed in the annual risk estimates for calculators using the mOBRI (maximum DR = 4.3) and HAS-BLED (maximum DR = 3.1) models. The 95% CI of the mean annual risk of bleeding varied across models; 1 calculator using the HAS-BLED model had a 95% CI of mean annual risk estimates of 5.4% to 6.2%, while another HAS-BLED calculator reported a 95% CI of 17, 7% to 18.5%. Agreement for stratification of risk categories among calculators was high for those based on mOBRI and ATRIA (=1 for both). Poor agreement was observed in 1 calculator using HEMORR2HAGES (=0.54) and another using HAS-BLED (range=-0.11 to 0.35).
CONCLUSION: Inconsistencies and lack of precision were observed in annual risk estimates and risk stratification produced by web and mobile bleeding risk calculators for patients with atrial fibrillation. Clinicians should refer to annual bleeding risks observed in major randomized controlled trials to inform risk estimates communicated to patients.
PMID:35418403 | DOI: 10.46747/cfp.6804e127