Oral anticoagulation and health care utilization in venous thromboembolism: What can we do to do better?

Since their introduction a decade ago, direct oral anticoagulants (DOACs) have experienced remarkable adoption and integration into health care.^1,2 According to Medicare part D claims, DOACs accounted for 15.4% of anticoagulant claims in 2013 and this increased to 31% by 2015, just 5 years after their introduction. ² Randomized trials in both venous thromboembolism (VTE) and nonvalvular atrial fibrillation (NVAF) have demonstrated the safety and efficacy of DOACs compared to traditional vitamin K antagonist (VKA) anticoagulants. Emerging real-world data support a reduced risk of bleeding, including decreased rates of intracranial hemorrhage, for DOACs compared to VKAs. Most practitioners recognize that in addition to being safe and efficacious, DOACs offer ease of dosing, fewer drug and food interactions, and a wide therapeutic window without the need for drug level monitoring. Though it is well recognized that DOACs are not suitable for all patients, guidelines for their safe and practical use in VTE have been published. ³ In addition, patient preference should be considered when choosing anticoagulation. Although many practitioners may assume a preference for DOAC therapy, patients may prefer an anticoagulant that is easily reversed or one in which drug levels can be measured while finding the need for regular testing and diet adjustment less troublesome. ⁴

VTE is the third most common cardiovascular disease. Health care utilization for patients with VTE is not confined to the initial encounter, or even the initial duration of anticoagulation therapy. Encounters may extend to VTE-related readmissions for recurrent events or bleeding, monitoring, and dose adjustments for patients taking VKA, and long-term sequelae of chronic thromboembolic disease, including pulmonary hypertension and post-thrombotic syndrome. ⁵

In this issue of Vascular Medicine, Lutsey and colleagues present data on patient-centered health care utilization in patients presenting with VTE and managed with oral anticoagulation. ⁶ Given the current health care environment, health care utilization both in the short-term and in the long-term is a practical consideration. Analyzing data from two large administrative databases (IBM MarketScan Commercial Claims and Encounters, and Medicare Supplemental and Coordination of Benefits databases) that included nearly 24,000 individual patients, the authors demonstrated fewer hospitalizations, fewer days hospitalized, fewer outpatient office visits, and fewer emergency department visits in patients initially prescribed rivaroxaban or apixaban compared to warfarin. Unfortunately, patients without insurance are under-represented in this analysis. According to the American Hospital Association, in 2016, approximately 10.9% of the US population was uninsured and this has risen steadily to approximately 13.7% in 2018. ⁷ This is an important limitation of this analysis and has potential impact on the generalizability of its findings.

The distribution of anticoagulant use in this analysis is interesting and not necessarily consistent with the findings of other studies. ² In 2015, DOACs comprised 31% of all Medicare part D anticoagulant claims, with warfarin claims accounting for 69% of use. ² Although the authors made efforts to minimize bias related to VTE prescription soon after DOAC approval (i.e. the majority use of rivaroxaban followed by apixaban in this study), one may assume there is still provider bias with respect to the choice of anticoagulant used. Whether this is based on prescription coverage, familiarity of use, patient co-morbidities, and/or perceived clinical risk, the influence behind this pattern cannot be ascertained. The study population is relatively young compared to Medicare participants, but whether patient age influenced the choice of anticoagulant is speculative. Though not statistically apparent, the rivaroxaban users in this study did indeed demonstrate fewer co-morbidities and less associated medication use compared to the warfarin and apixaban-treated populations.

A strength of this manuscript lies in the authors’ thoughtful analysis of their available data. Analyzing the data with respect to the individual agents, as well as including a propensity score-matched analysis, allows the reader to garner information on relative population risk between agents. However, the lack of an age- and sex-matched control population from the database limits the analysis to inferences related to overall medical care utilization and the potential impact of choice of anticoagulant. The indication for hospital readmission and emergency department visits are not defined and may be related to co-morbidities influencing the choice of anticoagulant rather than directly related to the anticoagulant agent and its clinical effects. The authors did include 180 individuals hospitalized for bleeding, again demonstrating incidence rate ratios favoring DOACs over warfarin. However, given the bleeding rates from clinical trials and ‘real-world’ data, ⁸ it seems likely that the data related to bleeding may not be completely represented.

Another consideration for application of this study is that analysis was limited to the first anticoagulant prescribed and only covered a 6-month period from the initial prescription. Therefore, patients who may have changed drugs either toward or away from DOACs due to cost, side effects, or other considerations are not represented. Limiting the analysis to 6 months is also concerning given the trend towards extended use of anticoagulants for secondary VTE prophylaxis in many patients. ⁴ Whether the reduction in resource utilization persists after the initial 6 months’ duration of therapy is unknown and merits further investigation. Lutsey and colleagues did not assess cost of therapy and related health care utilization, but one Canadian study using drug and monitoring costs alone demonstrated apixaban to be cost-effective for therapy lasting 3–12 months; however, indefinite therapy with DOACs is likely not cost-effective in the outpatient setting. ⁹ Further cost analysis of the impact of total health care resource utilization, including drug costs over an extended duration of therapy, is warranted.

Having demonstrated increased health care utilization in patients with VTE treated with warfarin, how can we best use these data to inform anticoagulation choice in VTE? Patient and provider education regarding health care utilization may impact anticoagulation choice. VKA use increases health care utilization at least in part through the need for outpatient monitoring; however, as previously noted, some patients prefer a drug that can be monitored, and indeed find monitoring acceptable. ⁴ Anticoagulation clinics have typically been used to manage VKA therapy. The role of anticoagulation clinics in mitigating health care utilization is unknown. Barnes and colleagues recently identified a number of barriers that have limited the expansion of these services to include DOAC therapy. ¹⁰ Whether incorporating DOACs into anticoagulation clinics could further decrease health care utilization has not been investigated and warrants further study.

This analysis does not allow for the evaluation of cost savings and resource utilization related to the initial hospitalization compared to outpatient therapy. DOACs may provide for outpatient management of deep vein thrombosis and low-risk uncomplicated pulmonary embolism, but comfort and confidence with this model may be lacking. As previously noted, whether long-term therapy would still favor DOACs for health care utilization is unknown. Cost remains a barrier for some patients; however, as generic drugs become available this barrier may be reduced.

In conclusion, care utilization is every practitioner’s responsibility, as is ensuring safe and effective patient care. DOACs have demonstrated safety and efficacy. Demonstrating the impact of DOACs on decreased short-term health care utilization associated with DOAC use compared to warfarin in acute VTE is an important step forward. Studies are needed to demonstrate the long-term impact on health care utilization. The cost of pharmacotherapy, as well as the impact on long-term sequelae of post-thrombotic syndrome, should be included.