Association of Direct Oral Anticoagulation Management Strategies With Clinical Outcomes for Adults With Atrial Fibrillation

Abstract

Importance: Anticoagulation management services (AMSs; ie, warfarin clinics) have evolved to include patients treated with direct oral anticoagulants (DOACs), but it is unknown whether DOAC therapy management services improve outcomes for patients with atrial fibrillation (AF).

Objective: To compare outcomes associated with 3 DOAC care models for preventing adverse anticoagulation-related outcomes among patients with AF.

Design, setting, and participants: This retrospective cohort study included 44 746 adult patients with a diagnosis of AF who initiated oral anticoagulation (DOAC or warfarin) between August 1, 2016, and December 31, 2019, in 3 Kaiser Permanente (KP) regions. Statistical analysis was conducted from August 2021 through May 2023.

Exposures: Each KP region used an AMS to manage warfarin but used distinct approaches to DOAC care: (1) usual care (UC) by the prescribing clinician, (2) UC plus an automated population management tool (PMT), or (3) pharmacist-managed AMS care. Propensity scores and inverse probability of treatment weights (IPTWs) were estimated. Direct oral anticoagulant care models were first indirectly compared using warfarin as a common comparator within each region and then directly compared across regions.

Main outcomes and measures: Patients were followed up until the first occurrence of an outcome (composite of thromboembolic stroke, intracranial hemorrhage, other major bleeding, or death), discontinuation of KP membership, or December 31, 2020.

Results: Overall, 44 746 patients were included: 6182 in the UC care model (3297 DOAC; 2885 warfarin), 33 625 in the UC plus PMT care model (21 891 DOAC; 11 734 warfarin), and 4939 in the AMS care model (2089 DOAC; 2850 warfarin). Baseline characteristics (mean [SD] age, 73.1 [10.6] years, 56.1% male, 67.2% non-Hispanic White, median CHA2DS2-VASc [congestive heart failure, hypertension, age ≥75 years, diabetes, stroke, vascular disease, age 65-74 years, female sex] score of 3 [IQR, 2-5]) were well balanced after IPTW. Over a median follow-up of 2 years, patients who received the UC plus PMT or AMS care model did not have significantly better outcomes than those who received UC. The incidence rate of the composite outcome was 5.4% per year for DOAC and 9.1% per year for warfarin for those in the UC group, 6.1% per year for DOAC and 10.5% per year for those in the UC plus PMT group, and 5.1% per year for DOAC and 8.0% per year for those in the AMS group. The IPTW-adjusted hazard ratios (HRs) for the composite outcome comparing DOAC vs warfarin were 0.91 (95% CI, 0.79-1.05) in the UC group, 0.85 (95% CI, 0.79-0.90) in the UC plus PMT group, and 0.84 (95% CI, 0.72-0.99) in the AMS group (P = .62 for heterogeneity across care models). When directly comparing patients receiving DOAC, the IPTW-adjusted HR was 1.06 (95% CI, 0.85-1.34) for the UC plus PMT group vs the UC group and 0.85 (95% CI, 0.71-1.02) for the AMS group vs the UC group.

Conclusions and relevance: This cohort study did not find appreciably better outcomes for patients receiving DOAC who were managed by either a UC plus PMT or AMS care model compared with UC.

Figure 1.. Association of Direct Oral Anticoagulant (DOAC) vs Warfarin Use With Major Clinical Outcomes by DOAC Management Model

Event rates are derived from each site independently and are unweighted; hazard ratios (HRs) are weighted by the inverse of the propensity score. AMS indicates anticoagulation management service; IP, inverse probability; PMT, population management tool; and UC, usual care. ^aComposite end point of thromboembolic stroke, intracranial hemorrhage, gastrointestinal bleeding, extracranial major bleeding, or death.

Figure 2.. Kaplan-Meier Graph for the Association of Direct Oral Anticoagulant (DOAC) vs Warfarin Use With the Net Clinical Benefit Outcome by DOAC Care Model

A-C, Unweighted curves for the net clinical benefit outcome in each DOAC care model. D-F, Weighted curves for the net clinical benefit outcome generated from inverse probability-weighted Cox proportional hazards regression models. The composite outcome was thromboembolic stroke, intracranial hemorrhage, gastrointestinal bleeding, extracranial major bleeding, or death. Patients were censored at the first occurrence of an outcome, loss of Kaiser Permanente membership, or December 31, 2020. Shaded regions around each line indicate the 95% CIs. Numbers at risk for the weighted analyses are omitted because the curves were generated from a model using weighted samples. AMS indicates anticoagulation management service; IPTW, inverse probability of treatment weighted; PMT, population management tool; and UC, usual care.

Figure 3.. Kaplan-Meier Graph for the Association of Direct Oral Anticoagulant (DOAC) Use Between Care Models With the Net Clinical Benefit Outcome

A and B, Unweighted curves for the net clinical benefit outcome between DOAC users in the usual care (UC) plus population management tool (PMT) vs UC care models, and anticoagulation management service (AMS) vs UC care models, respectively. C and D, Weighted curves for the net clinical benefit outcome generated from inverse probability-weighted Cox proportional hazards regression models. The composite outcome was thromboembolic stroke, intracranial hemorrhage, gastrointestinal bleeding, extracranial major bleeding, or death. Patients were censored at the first occurrence of an outcome, loss of Kaiser Permanente membership, or December 31, 2020. Shaded regions around each line indicate the 95% CIs. Numbers at risk for the weighted analyses are omitted because the curves were generated from a model using weighted samples. IPTW indicates inverse probability of treatment weighted.