Background:
Warfarin and dabigatran etexilate (DE) are oral anticoagulants (OAC) used to reduce the risk of stroke among patients with nonvalvular atrial fibrillation (AF). However, DE does not require titration and INR monitoring. This study examined whether hospital length of stay (LOS) and total hospital costs differed between the two therapies among treatment-naive, newly-diagnosed AF patients.
Methods:
LOS and total hospital costs were evaluated for hospitalizations with a primary or secondary discharge diagnosis of atrial fibrillation (AF) between 1/1/2011-3/31/2012, with DE or warfarin administered during hospitalization, and excluding hospitalizations of patients with valvular AF, previously diagnosed with AF, or previously treated with OAC. Hospitalizations were identified from a Charge Detail Masters database containing 397 qualified hospitals. Samples were propensity score matched using nearest neighbor within a caliper of 0.20 standard deviations of the logit, without replacement and a 2:1 match. Differences in LOS and hospital cost were then estimated using generalized linear models, fitted by generalized estimating equations (clustered by hospital) to account for possible correlation between observations. The hospitalization’s charged amount was multiplied by the hospital’s inpatient cost-to-charge ratio to estimate the total hospital cost. Covariates estimating the propensity score, LOS, and costs included patient age, payer type, CHADS
2
and HAS-BLED scores, use of bridging agents, comorbid conditions, and hospital attributes. As a sensitivity analysis, LOS and costs were estimated with the same parameters and covariates among the raw, unbalanced sample.
Results:
Matched samples included 1,292 warfarin and 646 DE hospitalizations of treatment-naive, newly diagnosed patients out of 4,619 and 715 hospitalizations, respectively. No covariates used in matching had standardized mean differences > 10% after matching. Two comorbidities (thromboembolism, coronary artery disease) had statistically different distributions after matching (DE: 3% vs. warfarin: 8%, p<0.001 and DE: 40% vs. warfarin: 45%, p=0.048); these were included as model covariates. Among the sample, DE had an estimated 0.7 days shorter stay compared to warfarin (DE: 4.8 days vs. warfarin: 5.5 days, p<0.01) and a $2,031 lower estimated total cost (DE: $14,794 vs warfarin: $16,826, p=0.007). Sensitivity analysis confirmed a shorter DE LOS (DE: 5.5 days vs. warfarin: 6.6 days, delta=1.1 days, p<0.01) and a lower DE hospital cost (DE: $18,362 vs. warfarin: $22,602, delta=$4,240, p<0.01).
Conclusions:
Among hospitalizations of treatment-naive patients newly diagnosed with nonvalvular AF, the hospitalizations during which DE was administered had a shorter LOS and at least a 12% lower total hospital cost compared to hospitalizations where warfarin was administered.
Dabigatran etexilate for nonvalvular atrial fibrillation in real practice and promises for its use to prevent stroke