Bleeding related to Oral Anticoagulants: Trends in US Emergency Department Visits, 2016–2020

Abstract

Background:

Clinical trials suggest lower rates of major bleeding with direct-acting oral anticoagulants (DOACs) than with warfarin, but anticoagulant-related bleeding remains one of the most common outpatient adverse drug events.

Methods:

We estimated the number of ED visits and subsequent hospitalizations for oral anticoagulant-related bleeding in 2016–2020 based on active surveillance in a nationally-representative, size-stratified probability sample of 60 U.S. hospitals. We estimated rates of ED visits using a nationally-projected retail prescription dispensing database.

Results:

Based on 19,557 cases, oral anticoagulant-related bleeding resulted in an estimated 1,270,259 (95% Confidence Interval [CI], 644,686–1,895,832) ED visits for the five years 2016–2020, of which 47.8%; (95% CI, 40.6%−55.0%) resulted in hospitalization. Oral anticoagulant-related bleeding resulted in an estimated 230,163 (109,598–350,728) ED visits in 2016 and 301,433 (138,363–464,503) in 2020. During 2016–2020, ED visits for DOAC-related bleeding increased by an average of 27.9% (95% CI 24.0%−32.0%; p<.001) per year, while ED visits for warfarin-related bleeding decreased by an average of 8.8% (95% CI −10.7– −7.0; p=.001). The estimated rate of bleeding visits per 100 patients dispensed oral anticoagulants at least once in 2016–2020 was highest for patients aged ≥80 years (13.1; 95% CI,6.2–20.0) and lowest for those aged <45 years (4.0; 95% CI,2.6–5.5); it was 5.9 visits per 100 patients dispensed DOACs [95% CI, 2.5–9.2] and 13.0 visits per 100 patients dispensed warfarin [95% CI,7.4–18.7].

Conclusions:

Although the rates of ED visits for anticoagulant-related bleeding may be lower for DOACs than for warfarin, persistently large numbers of patients requiring ED visits for anticoagulant-related bleeding despite increased use of DOACs and declining use of warfarin suggest that efforts to improve appropriate prescribing and monitoring of anticoagulants remain important.

BACKGROUND

Anticoagulants are key components in treatment and prevention of various hematologic and cardiovascular conditions, including venous thromboembolism (VTE) and atrial fibrillation (AF). However, anticoagulants commonly cause adverse events. For over a decade, anticoagulants have been among the most commonly implicated medications in Emergency Department (ED) visits for adverse drug events, with population rates as high as 4.5 ED visits per 1,000 population annually among older adults (aged ≥ 65 years).^1–3 Warfarin was the cornerstone of outpatient antithrombotic management for decades, but in 2010, direct-acting oral anticoagulants (DOACs) were introduced with fixed dosing, without requirements for laboratory monitoring, and lower rates of major bleeding than warfarin.^4–7 As the number of patients treated with DOACs increased and surpassed the number of patients treated with warfarin, there has been limited data describing the effect of this shift in prescribing on the national burden of ED visits for anticoagulant-related bleeding.^8,9

We used data from a nationally representative public health surveillance system to provide estimated numbers, rates, and time trends for ED visits for oral anticoagulant-related bleeding by patient characteristics, type of anticoagulant (DOACs or warfarin), and type of bleeding from 2016 to 2020.

METHODS

Data Sources

To estimate emergency department (ED) visits and subsequent hospitalizations for oral anticoagulant-related bleeding, data were obtained from the National Electronic Injury Surveillance System-Cooperative Adverse Drug Event Surveillance (NEISS-CADES) project, a joint collaboration of the US Centers for Disease Control and Prevention (CDC), the US Consumer Product Safety Commission, and the US Food and Drug Administration. NEISS-CADES is an active public health surveillance system based on a nationally-representative, size-stratified probability sample of 60 US hospitals and a 24-hour ED.^10–12

As described previously, trained abstractors review medical records of every ED visit to identify adverse drug events based on clinician documentation, including adverse reactions, medication errors, overdoses, and secondary effects (e.g., choking on a pill). Abstractors record up to four implicated medications and intent of use, clinician diagnoses, free-text narratives of the event (including clinical manifestations, relevant preceding events, concurrent medication, illicit drug or alcohol use), laboratory testing, treatments administered, and discharge disposition (including whether the patient was hospitalized [inpatient admission or transfer to another hospital], observed, or not hospitalized [treated and released, or left against medical advice/without being seen]). For every adverse drug event case, two clinician reviewers at CDC review free-text narratives and code clinical manifestations (and documented medication errors) based on the Medical Dictionary for Regulatory Activities (MedDRA), version 24.0.

To assess the estimated number of ED visits relative to oral anticoagulant exposure, national estimates of oral anticoagulant prescriptions dispensed at outpatient retail pharmacies were obtained from IQVIA Total Patient Tracker (TPT). IQVIA tracks national prescription trends and activity for pharmaceutical products, aggregates prescription data from participating U.S. pharmacies, and uses a proprietary algorithm to project TPT national-level estimates of unique patients receiving dispensed prescriptions. TPT data have previously been used to estimate the number of users of outpatient dispensed prescriptions nationally.^13–15 The IQVIA sample included pharmacy records from nearly 49,000 retail pharmacies across the U.S., representing approximately 92% of all retail prescription activity from independent and chain pharmacies and pharmacies in food and mass merchandise stores.

Because adding antiplatelet therapy to oral anticoagulants has been associated with increased bleeding risk,¹⁶ we also estimated the number of ED visits for oral anticoagulant-related bleeding which also involved antiplatelet therapy by examining the implicated and concomitant medications documented to identify antiplatelet medications (aspirin, aspirin/dipyridamole, clopidogrel, prasugrel, ticagrelor, anagrelide, and cilostazol).

Definitions

Cases of ED visits in which an oral anticoagulant was implicated from January 1, 2016, through December 31, 2020, were included. Oral anticoagulants included warfarin or DOACs (apixaban, dabigatran, edoxaban, and rivaroxaban). The MedDRA-coded narrative for each case was used to categorize primary manifestation of bleeding in a mutually exclusive and hierarchical manner based on severity; for example, an ED visit in which a patient experienced both hematemesis and epistaxis while on an anticoagulant would be categorized as gastrointestinal bleeding. Cases not involving bleeding (e.g., ED visits involving warfarin but with INR elevation alone) were excluded. Cases involving medication misuse or abuse, medication self-harm, overdoses without documentation of patient intent, and unsupervised ingestions by children aged ≤10 years were excluded. NEISS-CADES does not include therapeutic failures, nonadherence, withdrawal, or deaths while hospitalized or en route to the ED.

Statistical analysis

Each case is assigned a sample weight based on hospital sampling design and inverse probability of selection. We used the SAS (v9.4) SURVEYMEANS procedure to calculate 5-year and annual national estimates of ED visits in 2016–2020, with corresponding 95% confidence intervals (CIs), accounting for sample weights and complex sample design. Estimates based on <20 cases or total estimates of <1200 were considered statistically unstable and are not shown. Estimates with a coefficient of variation of >30% might be statistically unstable and are noted.

Rates of ED visits for oral anticoagulant-related bleeding were calculated by dividing the ED visit estimate (from NEISS-CADES) by dispensed prescription recipient estimates (from TPT). TPT’s nationally-projected estimates identify the number of patients receiving at least one dispensed prescription over the time period. Thus, 2016–2020 TPT oral anticoagulant estimates identify the number of patients receiving a dispensed oral anticoagulant prescription at least once over the 5-year period. A single year TPT estimate during this period identifies the number of patients receiving a dispensed oral anticoagulant prescription at least once during that corresponding year. Because of IQVIA’s large sample size (3.5 billion retail transactions annually), variance of TPT estimates was considered negligible. We used piecewise (segmented) regression (Joinpoint v4.9, National Cancer Institute, Bethesda, Maryland) to calculate changes in estimated annual rates of ED visits for warfarin- and DOAC-related bleeding. This activity was reviewed by CDC and conducted consistent with applicable federal law and CDC policy.¹⁷

RESULTS

Based on 19,557 cases in NEISS-CADES from 2016–2020, there were an estimated 1,270,259 (95% confidence interval [CI], 644,686–1,895,832) ED visits for oral anticoagulant-related bleeding, an average of 254,052 (95% CI, 128,937–379,166) ED visits annually. These visits accounted for 87.0% (95% CI, 83.7%−90.3%) of visits for oral anticoagulant adverse drug events overall (bleeding and non-bleeding) during this period.

Of these anticoagulant-related bleeding ED visits, an estimated 80.1% (95% CI, 77.3%−83.0%) of visits were made by adults aged ≥65 years, half of which (48.7%; 95% CI, 46.1%−51.3%) were visits made by adults aged ≥80 years. Adults aged ≥80 years had the highest five-year rate of ED visits (13.1 estimated visits per 100 recipients of oral anticoagulants), over 3-fold higher than adults aged <45 years (Table 1). Most estimated ED visits for oral anticoagulant-related bleeding were made by males (54.5%) but the estimated rate of ED visits by males was not different from that for females (9.3 visits per 100 male recipients [95% CI, 4.2–14.3] versus 8.3 ED visits per 100 female recipients [95% CI, 4.7–12.0]). Hospitalization (inpatient admissions and transfers to other hospitals) was required for nearly one-half (47.8%) of estimated ED visits for oral anticoagulant-related bleeding.

Table 1.

Numbers of Cases and National Estimates of Emergency Department (ED) Visits for Oral Anticoagulant-related Bleeding, by Case Characteristics—United States, 2016–2020^a

Case Characteristics	Emergency Department Visits for Bleeding
	Cases	National Estimate (2016–2020)
	No.	No. (95% CI)	% (95% CI)	Rate per 100 Recipientsb of Dispensed Oral Anticoagulants (95% CI)
Age c
<45	773	44,897 (28,820–60,975)	3.5 (2.8–4.3)	4.0 (2.6–5.5)
45–64	3,481	207,435 (122,153–292,717)	16.3 (14.1–18.5)	5.0 (2.9–7.0)
≥65	15,302	1,017,903 (489,371–1,546,436)	80.1 (77.3–83.0)	10.7 (5.1–16.2)
65–79	7,609	522,385 (252,204–792,566)	41.1 (38.4–43.8)	8.3 (4.0–12.5)
≥80	7,693	495,519 (234,483–756,554)	39.0 (36.6–41.4)	13.1 (6.2–20.0)
Sex d
Female	9,158	577,566 (324,132–831,001)	45.5 (41.1–49.8)	8.3 (4.7–12.0)
Male	10,397	692,467 (314,326–1,070,609)	54.5 (50.2–58.8)	9.3 (4.2–14.3)
Discharge Disposition
Hospitalizede	9,700	607,380 (238,335–976,425)	47.8 (40.6–55.0)	4.3 (1.7–6.8)
Observedf	365	21,162 (9,582–32,742)	1.7 (0.6–2.8)	0.1 (0.1–0.2)
Not Hospitalizedg	9,492	641,717 (374,664–908,769)	50.5 (44.0–57.1)	4.5 (2.6–6.4)
Total	19,557	1,270,259 (644,686–1,895,832)	87.0 (83.7–90.3)	8.9 (4.5–13.3)

^aCase counts and estimates are from the National Electronic Injury Surveillance System–Cooperative Adverse Drug Event Surveillance project, CDC. Estimates of recipients of dispensed oral anticoagulants are from IQVIA Total Patient Tracker, 2016–2020. ‘Oral Anticoagulants’ includes warfarin or direct-acting oral anticoagulants (apixaban, dabigatran, edoxaban, and rivaroxaban). Excludes cases involving medication misuse or abuse, medication self-harm, overdoses without documentation of patient intent, and unsupervised ingestions by children. CI, Confidence Interval.

^bRecipients of at least one dispensed oral anticoagulant prescription in 2016–2020.

^cAge not documented for 1 case of bleeding from direct-acting oral anticoagulants (DOACs).

^dSex not documented for 2 cases of bleeding from DOACs.

^eHospital admission or transfer to another hospital.

^fTime-limited assessment, treatment, and reassessment, nearly always lasting <48 hours (usually <24 hours).

^gTreated and released, or left against medical advice/without being seen.

Over the five-year period of 2016–2020, DOACs were implicated in 48.0% (95% CI, 43.0%−52.9%) of estimated ED visits for oral anticoagulant-related bleeding with a nominally lower rate compared with that for warfarin (5.9 ED visits per 100 patients dispensed DOACs [95% CI, 2.5–9.2] versus 13.0 ED visits per 100 patients dispensed warfarin [95% CI,7.4–18.7]). The proportions of visits by patient age and sex, and the proportion of patients hospitalized were similar for visits attributed to DOACs and warfarin (Table S1). During the most recent calendar year analyzed, 2020, among adults aged ≥65 years, the estimated annual rate of ED visits for bleeding was 4.4 per 100 DOAC recipients (95% CI, 1.7–7.0) and 6.7 per 100 warfarin recipients (95% CI% 3.1–10.3).

The number of patients prescribed any oral anticoagulant increased 17.8% between 2016 and 2020 (6,031,759 to 7,107,688), with the number of patients prescribed DOACs increasing by 83.6% (2,854,098 to 5,239,405) and the number of patients prescribed warfarin decreasing by 41.2% (3,177,661 to 1,868,283). Overall, the estimated number of ED visits for oral anticoagulant-related bleeding was 230,163 (109,598–350,728) in 2016 and 301,433 (138,363–464,503) in 2020 (Figure and Table S2). During 2016–2020, the number of ED visits for DOAC-related bleeding increased an average of 27.9% (95% CI 24.0%−32.0%; p<.001) per year, and the number of ED visits for warfarin-related bleeding decreased by an average of 8.8% (95% CI −10.7– −7.0; p=.001). Estimated rates of ED visits for DOAC- and warfarin-related bleeding increased in 2016–2020 for both DOACs (Annual percentage change [APC] 9.4%, 95% CI 5.9–12.9; p=.003) and warfarin (APC 4.0%; 95% CI, 1.9%−6.2%; p=.009) (Table S2).

Figure 1.

National estimates of Emergency Department (ED) visits for Oral Anticoagulant-related Bleeding and Patients Dispensed Oral Anticoagulants, by Year—United States, 2016–2020.

Estimates of ED visits for oral anticoagulant-related bleeding are from the National Electronic Injury Surveillance System-Cooperative Adverse Drug Event Surveillance project, CDC; estimates of unique recipients of dispensed oral anticoagulants (direct-acting oral anticoagulants [DOACs] or warfarin) from outpatient retail pharmacies are from IQVIA Total Patient Tracker (2016–2020). Error bars display 95% confidence intervals for ED visit estimates. Top panel: national estimates of ED visits for oral anticoagulant-related bleeding (solid line and primary vertical axis) and unique recipients of dispensed oral anticoagulants (dashed line and secondary vertical axis). Middle panel: national estimates of ED visits for DOAC-related bleeding (solid line and primary vertical axis) and unique recipients of dispensed DOAC prescriptions (dashed line and secondary vertical axis). Bottom panel: national estimates of ED visits for warfarin-related bleeding (solid line and primary vertical axis) and unique recipients of dispensed warfarin prescriptions (dashed line and secondary vertical axis).

Note: coefficient of variation for the 2016 DOAC-related bleeding estimate exceeds 30% and may therefore be considered statistically unstable.

In 2016–2020, DOAC-related bleeding visits most commonly involved apixaban (52.4% [95% CI, 48.0%−56.7%]), followed by rivaroxaban (41.0% [95% CI, 36.8%−45.2%]) and dabigatran (6.0% [95% CI, 3.1%−9.0%]). There were too few cases of edoxaban-related bleeding to calculate a stable national estimate. The estimated five-year rate of bleeding visits was similar between DOACs: 5.0 (95% CI, 2.1–7.9) per 100 apixaban recipients, 5.8 (95% CI, 2.7–8.9) per 100 rivaroxaban recipients, and 7.6 (95% CI, 0.4–14.9) per 100 dabigatran recipients (CV>30%).

Nearly one-third (31.2%; 95% CI, 27.2%−35.1%) of oral anticoagulant-related bleeding ED visits also documented concomitant use of aspirin (26.3%; 95% CI, 22.0%−30.5%) and 7.5% documented the use of another antiplatelet agent (95% CI, 6.6%−8.5%). Most oral anticoagulant-related bleeding ED visits which also documented concomitant use of antiplatelet agents involved men (62.5% [95% CI, 58.3%−66.7%]); ED visits for anticoagulant-related bleeding without documented use of antiplatelet agents were evenly divided between males and females (50.9% [95% CI, 46.7%−55.1%] versus 49.1% [95% CI, 44.9%−53.3%]) (Table S3). Concomitant antiplatelet agent use was documented in a nearly identical proportion of anticoagulant-related bleeding visits attributed to DOACs and warfarin (31.5% [27.4%−35.6%] DOAC visits versus 30.8% [95% CI, 26.7%−35.0%] warfarin visits). The proportion of ED visits for oral anticoagulant-related bleeding resulting in hospital admission or transfer to another facility was similar between ED visits documenting concomitant antiplatelet use (49.1%; 95% CI, 40.6%−57.6%) and those that did not document antiplatelet use (47.2%; 95% CI, 40.7%−53.8%).

Gastrointestinal bleeding was the most commonly documented manifestation in ED visits for oral anticoagulant-related bleeding (34.5%) (Table 2) and was involved in a similar proportion of DOAC-related bleeding visits compared with warfarin (39.3% [95% CI, 29.3%−49.4%] versus 30.1% [95% CI, 24.7%−35.4%]). The proportion of visits was also similar for DOACs and warfarin for central nervous system bleeding (4.0% [95% CI, 2.8%−5.3% versus 5.0% [95% CI, 3.8%−6.2%]), pulmonary bleeding (2.4% [95% CI, 1.7%−3.1%] versus 2.8% [95% CI, 1.8%−3.8%]), genitourinary bleeding (11.9% [95% CI, 8.4%−15.4%] versus 10.4% [7.1%−13.7%]), epistaxis (18.0% [95% CI, 12.6%−23.3%] versus 20.0% [95% CI, 15.5%−24.5%]), and skin/wound or other minor bleeding (19.5% [95% CI, 10.2%−28.8%] versus 25.1% [95% CI, 18.2%−31.9%]).

Table 2.

National Estimates of Emergency Department (ED) Visits for Oral Anticoagulant-related Bleeding, by Bleeding Manifestation—United States, 2016–2020^a

Bleeding Manifestationb	Emergency Department Visits for Bleeding: National Estimate (2016–2020)
	Any Oral Anticoagulant		Direct-Acting Oral Anticoagulants (DOACs)		Warfarin
	No. (95% CI)	% (95% CI)	No. (95% CI)	% (95% CI)	No. (95% CI)	% (95% CI)
Central nervous system bleedingc	57,642 (19,555–95,728)	4.5 (3.4–5.7)	24,249 (5,496–43,002)d	4.0 (2.8–5.3)	33,393 (13,445–53,341)	5.0 (3.8–6.2)
Pulmonary bleeding	33,140 (10,236–56,045)d	2.6 (1.8–3.4)	14,623 (4,207–25,039)d	2.4 (1.7–3.1)	18,518 (5,820–31,215)d	2.8 (1.8–3.8)
Gastrointestinal bleeding	438,092 (165,922–710,262)	34.5 (26.8–42.2)	238,258 (67,134–409,381)d	39.3 (29.3–49.4)	199,834 (96,740–302,929)	30.1 (24.7–35.4)
Genitourinary bleeding	140,896 (38,127–243,664)d	11.1 (7.7–14.5)	71,971 (15,113–128,829)d	11.9 (8.4–15.4)	68,924 (22,404–115,445)d	10.4 (7.1–13.7)
Epistaxis	241,683 (160,593–322,772)	19.0 (14.1–23.9)	108,812 (69,885–147,738)	18.0 (12.6–23.3)	132,871 (88,410–177,332)	20.0 (15.5–24.5)
Skin/wound or other minor bleeding	284,637 (143,934–425,340)	22.4 (14.4–30.5)	118,014 (46,618–189,411)	19.5 (10.2–28.8)	166,623 (94,396–238,850)	25.1 (18.2–31.9)
Other types of hemorrhage	74,170 (26,672–121,667)	5.8 (3.7–8.0)	29,675 (6,439–52,911)d	4.9 (2.3–7.5)	44,495 (19,305–69,684)d	6.7 (4.9–8.5)
Total	1,270,259 (644,686–1,895,832) e	100	609,116 (264,324–953,909)	48.0 (43.0–52.9)	664,658 (375,437–953,879)	52.3 (47.3–57.4)

^aEstimates are from the National Electronic Injury Surveillance System–Cooperative Adverse Drug Event Surveillance project, CDC. ‘Oral Anticoagulants’ includes warfarin or direct-acting oral anticoagulants (apixaban, dabigatran, edoxaban, and rivaroxaban). Excludes cases involving medication misuse or abuse, medication self-harm, overdoses without documentation of patient intent, and unsupervised ingestions by children. CI, Confidence Interval.

^bBleeding manifestations are mutually exclusive and were assigned hierarchically. For example, an ED visit in which a patient experienced both hematemesis and epistaxis while on an anticoagulant would be categorized as gastrointestinal bleeding.

^cIncludes epidural or subdural hematoma, hemorrhagic stroke, and intracerebral or subarachnoid hemorrhage.

^dCoefficient of variation >30%.

^eThe total estimate for ED visits for bleeding from any oral anticoagulant is less than the sum of ED visit estimates for bleeding from DOACs and warfarin because use of both a DOAC and warfarin was implicated in some cases.

DISCUSSION

The estimated number of ED visits for oral anticoagulant bleeding did not decline from 2016 to 2020, despite declining use of warfarin and the introduction of agents with lower bleeding rates. Although DOACs are associated with fewer major bleeds and require less frequent monitoring, with increasing use of DOACs the number of ED visits for DOAC-related bleeding also increased, with the proportion hospitalized (50.0%) similar to that for warfarin-related bleeding (45.7%).

Increased use of DOACs has been supported by clinical practice guidelines favoring their use over warfarin in many situations, for example nonvalvular atrial fibrillation in the 2019 American College of Cardiology guidelines^18,19 and venous thromboembolism/pulmonary embolism in the 2020 American Society of Hematology guidelines.²⁰ Additionally, evidence continues to suggest there continues to be under-prescribing of anticoagulation for patients for whom anticoagulation is recommended, as demonstrated by global geographic variability in initiating oral anticoagulants among patients recently diagnosed with atrial fibrillation.²¹ Yet the nearly 200,000 estimated DOAC-related bleeding ED visits in 2020 – with a rate of 3.7 per 100 or 1 ED visit for bleeding per 27 patients prescribed DOACs – suggests further bleeding prevention efforts are warranted alongside efforts to expand appropriate use.

This need to both increase uptake when indicated and minimize risk for bleeding complications is particularly true for older adults. Older adults have not only the highest incidence of atrial fibrillation complications but also the highest bleeding risk.²² In this study, in 2020 the estimated rate of ED visits for oral anticoagulant-related bleeding was 6.7 per 100 or 1 in 15 adults aged ≥65 years prescribed warfarin, and 4.4 per 100 or 1 in 23 adults aged ≥65 years prescribed DOACs. Older adults may have been under-represented in clinical trials of DOACs, which provided initial estimates of bleeding risk; some patients may have unmet clinical needs that warrant particular attention to optimize safety and effectiveness.^23–27

Nearly one-third (31.2%) of estimated bleeding visits included documentation of concomitant aspirin or antiplatelet use, which may provide an opportunity for reducing bleeding risk in some patients. Increased rates of nonmajor bleeding have previously been demonstrated when aspirin is added to DOACs²⁸ or warfarin^16,29, with thrombosis rates similar to those for monotherapy. A nationwide cohort study of major bleeding among patients in Denmark with atrial fibrillation found concomitant antiplatelet therapy increased major bleeding by over 25% for DOAC-treated patients and approximately 80% for warfarin-treated patients.³⁰ A multi-center cohort study in Michigan found nearly 1 out of every 3 DOAC-treated patients with AF and/or VTE received aspirin without a clear indication, which suggests opportunities for deprescribing are needed, even for patients receiving DOACs.²⁸

Anticoagulation management systems, which have demonstrated cost effectiveness mainly in studies of warfarin-treated patients,^31,32 are adapting to serve the expanding DOAC-treated patient population.^33,34 Programs such as the National Quality Forum’s Anticoagulation Stewardship Playbook guide efforts to optimize anticoagulant outcomes based on lessons from improving prescribing and minimizing adverse effects of antimicrobials.^35,36

This study is subject to several limitations. First, it underestimates the burden of oral anticoagulant-related bleeding because it does not include episodes treated outside of EDs or those resulting in death (due to variability in how deaths are tracked in ED records systems). ED visits for nonbleeding complications of anticoagulant use (e.g., warfarin-associated international normalized ratio level elevation, DOAC-associated hepatic enzyme elevation³⁷) were also not included. Second, these data should not be used for direct comparisons between DOACs and warfarin, or among DOACs, or to guide the choice of anticoagulant prescribing. Additional patient data needed for risk adjustment (e.g., indication for prescribing, patient age, functional status, and current medical comorbidities) are important to differentiate agent-specific bleeding risk based upon patient characteristics and underlying diagnoses. Additional research is needed to overcome anticoagulant prescribing hesitancy,²² inform clinician prescribing discussions with patients, and optimize choice of anticoagulant for the right patient for the right indication. Third, IQVIA TPT estimates only include prescriptions dispensed from outpatient retail pharmacies. Prescriptions dispensed from other settings (e.g., hospital, clinic, or long-term care pharmacies, mail order pharmacies) are not included and it may be possible that the relative frequency of DOAC versus warfarin dispensing is different in these settings, although these settings comprise a small proportion of total prescribing and a similar increase in DOAC prescribing has been observed among Medicare beneficiaries.³⁸ Fourth, oral anticoagulant bleeding ED visit data could not be analyzed by race/ethnicity because of variability in how each NEISS-CADES participating hospital collects and records patient race/ethnicity.

CONCLUSIONS

Oral anticoagulant-related bleeding causes a significant population burden with high numbers of ED visits and hospitalizations for adverse events, especially among older adults. Although more patients are dispensed DOACs than warfarin and the rates of ED visits for anticoagulant-related bleeding may be lower for DOACs than for warfarin, persistently large numbers of patients requiring ED visits for anticoagulant-related bleeding suggest that efforts to improve appropriate prescribing and monitoring of anticoagulant remain important in the era of the DOACs. Renewed quality improvement efforts to optimize appropriate anticoagulant therapy and minimize adverse events should be evaluated for measurably improving patient safety.

Abbreviations:

ED

emergency department

DOAC

direct-acting oral anticoagulant

NEISS-CADES

National Electronic Injury Surveillance System-Cooperative Adverse Drug Event Surveillance

TPT

Total Patient Tracker

CI

confidence interval