Pharmacists’ Bleed Risk Tool and Treatment Preferences Prior to Initiating Anticoagulation in Patients With Nonvalvular Atrial Fibrillation: A Cross-Sectional Survey

The recently published ARISTOPHANES study provides evidence that apixaban is an appropriate choice in patients with a HAS-BLED score < 3. In this retrospective observational study, more than 70% of patients received standard doses of DOACs (apixaban 5 mg, dabigatran 150 mg, or rivaroxaban 20 mg) and about 20% had a bleeding history, about 30% were on PPIs, less than 25% were on NSAIDs, and about 40% had a HAS-BLED score < 3. The study found that apixaban was more effective (reduced rates of ischemic or hemorrhagic strokes/systemic embolism) and safer (reduced rates of major GI bleed or intracranial bleed) than warfarin.²⁸ Dabigatran and rivaroxaban were also more effective than warfarin for stroke prevention and had a lower risk for major intracranial bleed risk; while the risk of major GI bleed was similar between dabigatran and warfarin, major GI bleed risk was higher for rivaroxaban. When compared with each other, the 3 DOACs were effective at stroke prevention, with apixaban more effective than dabigatran and rivaroxaban; similar efficacy was noted for dabigatran versus rivaroxaban. Apixaban was associated with fewer GI bleeds versus dabigatran and rivaroxaban, but with similar intracranial bleed risks; dabigatran was associated with fewer GI bleeds but similar intracranial bleed risks versus rivaroxaban.²⁸ Efficacy and safety findings from a subgroup analysis based on HAS-BLED scores < 3 and ≥ 3 were generally consistent with the main results.

When bleed risk was equal to stroke risk, the difficulty was determining how OACT in a patient at high stroke risk (CHA₂DS₂VASc score ≥ 2) and high bleed risk (HAS-BLED score ≥ 3) should be managed. Eight respondents chose aspirin and added gastroprotection with either a PPI or H₂RA; however, currently, aspirin is not recommended as the sole antithrombotic for patients with NVAF.²³ With the OACT, an interesting finding was that as the number of patients seen in 12 months increased, pharmacists were almost twice as likely to select a DOAC over warfarin. Moreover, pharmacists were judicious in their recommendation to add gastroprotection, and would consider doing so if there was a specific indication. At the time of our survey, several studies described DOAC-associated GI bleeds,^29-31 but data on the effectiveness of acid-suppressive therapy, specifically with PPIs, in the prevention of upper GI bleeds were sparse.^4,7,32 Respondents most likely were familiar with GI bleed risk factors and prevention strategies from various guidelines published between 2009 and 2010, which did not include DOACs.^33-35

Another important finding was pharmacists’ uncertainty as to the effectiveness of PPIs in preventing GI bleeds in combination with DOACs. The data are conflicting. A meta-analysis of older studies (2007-2015) showed that PPIs (but not H₂RAs) reduced the risk of upper GI bleeds in patients on warfarin but not for dabigatran.³⁶ A retrospective cohort study of Medicare beneficiaries on OACTs (2011-2015) showed the adjusted incidence of hospitalization for upper GI bleeds in patients on PPI co-therapy was significantly lower compared with patients not on PPI co-therapy (76 versus 115 per 10,000 person-years, respectively).⁸ Apixaban without PPI co-therapy was associated with the lowest risk of upper GI bleed hospitalizations (73/10,000 person-years), and PPI co-therapy further reduced this risk (49/10,000 person-years). Warfarin without PPI co-therapy was associated with the next lowest risk (113/10,000 person-years), followed by dabigatran (120/10,000 person-years) and rivaroxaban (144/10,000 person-years). PPI co-therapy significantly reduced the risk of upper GI bleed hospitalizations with all OACTs, but the incidence of upper GI bleed hospitalizations with rivaroxaban was significantly greater than with the other OACTs.⁸ Therefore, if there are concerns about the safety of PPIs,^37-39 or the patient is unable to tolerate a PPI, then apixaban may be the most appropriate DOAC for a patient with high bleed risk. Notably, a 2020 review of data from the PINNACLE registry (average age, 75-77 years; 31% on PPIs) found that the relative GI bleed safety advantage of apixaban and dabigatran versus warfarin was attenuated in patients ≥ 75 years.⁴⁰ Last, since the risk for lower GI bleeds is not reduced by PPIs,⁴¹ consideration of their use should be accompanied by an assessment to detect bleeds (eg, low hemoglobin/hematocrit, presence of bright red blood, hematochezia/melena, fecal occult testing), with prompt management as necessary.⁵

Limitations

Limitations of our survey included an overall low response rate, which can generate a biased sample if respondents are systematically different from nonrespondents. In addition, to maintain simplicity and reduce respondents’ time commitment, our survey did not include actual CHA₂DS₂VASc stroke risk scores, HAS-BLED bleed risk scores, or specific GI bleed risk factors when querying pharmacists about treatment options based on bleed risk. The addition of these variables would have improved the robustness of the data.

Conclusion

In addition to applying BRTs in the management of NVAF patients, pharmacists considered patient-specific variables, prescriber preferences, and evidence-based guidance when recommending OACT with or without gastroprotection. To avoid suboptimal patient management, busy pharmacists should be granted time to attend continuing education programs describing optimal OACT selection and formulation of individualized, evidence-based plans to address modifiable risk factors for bleeding, including the appropriate use of gastroprotection. Randomized, prospective, long-term studies stratified by HAS-BLED and CHA₂DS₂VASc scores are needed to further clarify efficacy, safety, and cost-effectiveness of OACT, with and without PPIs, in patients who may be at risk for upper GI bleeds.

Acknowledgments: The authors thank Robin J. Jacobs, PhD, MSW, MS, MPH, Patrick C. Hardigan, PhD, Steven Brettler, PharmD, MPH, Maria-Isabel A. Cabral, PharmD, and Reginald Gyapong, PharmD, for their participation in this project. The authors also sincerely thank Fabio Franco, BS Computer Science, who organized the database to enable efficient data management.

Corresponding author: Devada Singh-Franco, PharmD, CDE, Nova Southeastern University College of Pharmacy, 3200 S University Drive, Fort Lauderdale, FL 33328; singh@nova.edu

Disclosures: None.

Funding: The study was supported by Nova Southeastern University’s Health Professions Division Internal Research Grant.