OBJECTIVES: To analyze time dependent disease severity patterns in prescription and reimbursement using US real world closed claims data.

METHODS: We analyzed a subset of US Medicare claims data comprising patients who initiated apixaban treatment over 5 years (from 2015 to 2019). Without access to medical records, we created a surrogate variable for the severity of non-valvular atrial fibrillation (NVAF), the sum of paid medical claims with a diagnostic code of NVAF issued in the 100-day period that preceded the date of the first prescription of apixaban. We chose apixaban because it is the top-ranking drug on the list of drugs subjected to Medicare negotiations as part of the IRA. We analyzed the distribution of HCRU (mean, median and IQR) annually, over 5 years.

RESULTS: Between 2015 and 2019 the median amount paid increased monotonically from $ 498 to $ 965 (P-value: 0.017). The mean amount paid increased from $ 4,231 to $ 6,707. The annual HCRU means are much higher than the medians, because of outliers, but the trend remains the same. These preliminary results suggest that the proportion of severe patients being treated with apixaban for NVAF increased over time (from 2015 through 2019) post launch.

CONCLUSIONS: Further research is warranted to explain the observed disease severity trend post launch. One could compare the reimbursement trends post launch, including: rejected vs. accepted claims. One of the explanations could be that the formularies became gradually less restrictive for severe patients. Another explanation may be cautious decision-making and risk-averse behavior on the part of physicians initiating new treatment. Limitations of our analysis include studying apixaban only, which precludes generalizability to other drugs and conditions, and using an unvalidated proxy for disease severity.