Initiation of oral anticoagulation in US older adults newly diagnosed with atrial fibrillation during hospitalization

Abstract

Background: Atrial fibrillation is a common cause of stroke among older adults and is often first detected during hospitalization, given frequent use of cardiac telemetry.

Methods: In a 20% national sample of Medicare fee-for-service beneficiaries, we identified patients aged 65-or-older newly diagnosed with atrial fibrillation while hospitalized in 2016. Our primary outcome was an oral anticoagulant claim within 7-days of discharge. Multivariable logistic regression analyses assessed relationships between anticoagulation initiation and thromboembolic and bleeding risk scores while controlling for demographics, frailty, comorbidities, and hospitalization characteristics.

Results: Among 38,379 older adults newly diagnosed with atrial fibrillation while hospitalized (mean age 78.2 [SD 8.4]; 51.8% female; 83.3% white), 36,633 (95.4%) had an indication for anticoagulation and 24.6% (9011) of those initiated an oral anticoagulant following discharge. Higher CHA₂ DS₂ -VASc score was associated with a small increase in oral anticoagulant initiation (predicted probability 20.5% [95% CI, 18.7%-22.3%] for scores <2 and 24.9% [CI, 24.4%-25.4%] for ≥4). Elevated HAS-BLED score was associated with a small decrease in probability of anticoagulant initiation (25.4% [CI, 24.4%-26.4%] for score <2 and 23.1% [CI, 22.5%-23.8%] for ≥3). Frailty was associated with decreased likelihood of oral anticoagulant initiation (24.7% [CI, 23.2%-26.2%] for non-frail and 18.1% [CI, 16.6%-19.6%] for moderately-severely frail). Anticoagulant initiation varied by primary reason for hospitalization, with predicted probability highest among patients with a primary diagnosis of atrial fibrillation (46.1% [CI, 45.0%-47.3%]) and lowest among those with non-cardiovascular conditions (13.8% [CI, 13.3%-14.3%]) and bleeds (3.6% [CI, 2.4%-4.8%]).

Conclusions: Oral anticoagulant initiation is uncommon among older adults newly diagnosed with atrial fibrillation during hospitalization, even among patients hospitalized primarily for atrial fibrillation and patients with high thromboembolic risk. Clinicians should discuss risks and benefits of oral anticoagulants with all inpatients found to have atrial fibrillation.

Keywords: anticoagulation; atrial fibrillation; frailty; geriatric cardiology; inpatient.

Figure 1.

Unadjusted rates of anticoagulation initiation (bottom row) by primary discharge diagnosis (atrial fibrillation vs. other) and frailty status

Figure 2.

Predicted probability of anticoagulation initiation, by thromboembolic and bleeding risk category ^a For thromboembolic risk, CHA₂DS₂-VASc score <2 was considered low, 2–3 moderate, and ≥4 high. Compared with patients at low thromboembolic risk, those with moderate and high thromboembolic risk had higher predicted probability of initiating anticoagulation (differences: +3.3% [CI +1.3% to +5.4%] for moderate thromboembolic risk, +4.4% [CI 2.5% to 6.3%] for high thromboembolic risk). ^b For bleeding risk, HAS-BLED score <2 was considered low, 2 moderate and ≥3 high. Compared with patients at low bleeding risk, those with high bleeding risk had lower predicted probability of initiating anticoagulation, while those with moderate bleeding risk did not (differences: −0.26% [CI −0.92% to +1.4%] for moderate bleeding risk, −2.2% [CI −3.4% to −1.0%] for high bleeding risk).

Figure 3.

Predicted probability of oral anticoagulant initiation by frailty and reason for hospitalization among patients with guideline-based indications for anticoagulation ^a Compared with non-frail patients, those with mild frailty had a 1.8% lower predicted probability of initiating anticoagulation (CI −3.6% to −0.04%) and those with moderate-severe frailty had 6.6% lower predicted probability of anticoagulation (CI −8.9% to −4.4%). ^b Compared to patients hospitalized for atrial fibrillation, those hospitalized for all other reasons had lower predicted probability of initiating anticoagulation (differences: −4.0% [CI −7.1% to −0.83%] for cardiac surgery patients, −20.8% [CI −22.2% to −19.4%] for patients with other cardiovascular conditions, −32.4% [CI −33.6 to −31.1%] for patients with non-cardiovascular conditions, and −42.6% [CI −44.2% to −40.9%] for patients with bleeding).