Atrial fibrillation (AF) is prevalent in long‐term care (LTC) facilities. Real‐world data demonstrate that Black patients with AF are 10% to 22% less likely to be on anticoagulation compared with White patients in the ambulatory setting.¹ It is unclear if this disparity is present in the LTC setting as well.

LTC care is highly segregated. Compared with White residents, racial and ethnic minority residents receive care in facilities with more limited financial resources and quality deficiencies.² No one has previously examined the role of facility, particularly the percentage of racial and ethnic minority residents at a given facility, and its relationship with anticoagulation use in residents with elevated risks for stroke and AF.

Considering theory‐based structural racism,³ we measured the association of resident race and ethnicity combined with facility percentage of racial and ethnic minority residents on the use of anticoagulation in US LTC facilities.

Although we are not allowed to share our data, a technical appendix of our statistical procedure can be made available by the corresponding author upon request.

We included residents with AF and CHA₂DS2‐VASc scores ≥2 living in the LTC setting (ie, >100 days in a nursing home) between July and December 2015 and continuously enrolled in Medicare Part D for 7 months up to and including their index assessment. Residents were considered to have AF if they had a record of cardiac dysrhythmia recorded in the minimum data set (MDS) or an International Classification of Diseases, Tenth Revision (ICD‐10) diagnosis code consistent with AF following a prior example in the literature.⁴ We excluded residents on hospice. The University of Massachusetts Institutional Review Board approved our analysis as exempt from federal regulatory review given its deidentified nature.

We tracked anticoagulation using Medicare Part D data. Specifically, we considered a patient on anticoagulation if the fill date for an anticoagulation medication record plus days supplied in this record overlapped with the index assessment accounting for medication accumulation and recent hospitalization.

We identified race and ethnicity with the Research Triangle Institute–Race Code variable within the Medicare Beneficiary Survey File. The variable includes categories non‐Hispanic White race (hereafter referred to as simply “White” race), Black race, Hispanic ethnicity, American Indian/Alaska Native race, Asian race, or Other race (ie, not specified by other Medicare race or ethnicity category). The Research Triangle Institute‐Race Code variable improves ascertainment of Hispanic ethnicity through surname‐based imputation. We then calculated the percentage of racial and ethnic minority residents in each facility and grouped each facility into 4 categories (roughly quartiles) based on the percentages 0% to 10%, 10% to 20%, 20% to 50%, >50%. Finally, we assigned each patient into race‐facility minority percentage composite categories.

We examined multiple resident characteristics available in the MDS to adjust for potential confounding. This included several geriatric conditions including fall history, dependency in activities of daily living, mobility impairment, cognitive impairment, low body mass index, and weight loss. We also adjusted for stroke risk (ie, CHA₂DS₂‐VASc score) and bleeding risk factors (cirrhosis, anemia, renal disease, use of antiplatelet medication).

We also examined for‐profit status (for‐profit versus not‐for‐profit), chain status (versus independent facility), staffing ratios (registered nurses and licensed practical nurses to patients), percentage of Medicaid residents, and Centers for Medicare and Medicaid (CMS) 5‐star quality rating.

We constructed a binomial logistic regression model with a generalized estimating equation to measure percentage of anticoagulation use in patient race and ethnicity‐facility minority percentage groups adjusted for resident and facility characteristics as well as clustering by facility.

We identified 199 822 eligible residents living in 6183 LTC facilities. Of these, 60 065 residents (29.8%) were on anticoagulation. The mean age of the residents was 83.1±9.7 years. Most residents were of White race (81.8%). White residents had the highest anticoagulation use (30.5%), followed by Black residents (27.1%), Hispanic residents (26.3%), and Asian residents (21.6%). When examining the resident race and ethnicity‐facility minority percentage composite groupings, we found that the anticoagulation rates were higher in facilities with a lower percentage of racial and ethnic minority residents (0%–10%) and lower in facilities with a higher percentage of racial and ethnic minority residents (>50%). This was consistent across resident racial and ethnic categories (apart from American Indian for which we had small numbers, n=731; see the Figure).

Anticoagulation usage was low across the board (28.7%) with significantly lower anticoagulation use by resident race and ethnicity. Facilities with higher percentages of racial and ethnic minority residents living within them had lower anticoagulation use for nearly every racial and ethnic category.

Our findings resemble those reported almost 20 years ago by Christian et al.⁵ They found that racial and ethnic minority residents were ≥20% less likely to receive anticoagulation compared with White residents in cases in which there was an indication for anticoagulation. We extended the prior work by analyzing facility characteristics including the percentage of racial and ethnic minority residents. Despite nursing home quality improvement efforts, including CMS quality reporting, care delivery remains inequitable for racial and ethnic minority residents in terms of anticoagulation prescription.

We identified a racial disparity in anticoagulation use even after adjusting for an extensive list of resident and facility characteristics (including quality rating and staffing ratios). Limitations to our findings include that trends may also have changed since 2015. It is also unclear what the target anticoagulation rates should be in this population given many patients are nearing end of life. Absence of information on reason for nonuse of anticoagulation and cardiology consultation are also limitations. Better education of residents and increasing cardiology referrals represent potential remedies.

We conducted this research through funding received in a research award from Bristol Myers Squibb and Pfizer as part of ARISTA‐USA (American Thrombosis Investigator Initiated Research Program).

Dr Kapoor has received research grant support from Pfizer through its Independent Grants for Learning and Change funding mechanism and from Bristol Myers Squibb for independent medical education grants. Recently, he has received research grant support through a competitive process adjudicated and funded by the alliance, which is formed by both Pfizer and Bristol Myers Squibb. He has also been awarded a grant by Pfizer to examine conversations between patients and providers. Dr McManus receives sponsored research support from Bristol Myers Squibb, Boeringher Inghelheim, Pfizer, Biotronik, and Philips Healthcare and has consulted for Bristol Myers Squibb, FlexCon, Samsung, Philips, and Pfizer. Dr McManus has equity in Mobile Sense Technologies, LLC. Dr Gurwitz is a member United HealthCare Pharmacy and Therapeutics Committee. Sadiq, Patel, Dr Zhang, and Dr Crawford have also received research grant support from Bristol Meyers Squibb in the past 3 years (coinvestigator on the grants secured by Dr Kapoor and Dr McManus as described previously).

Values above each bar represent the percentage of anticoagulation use in resident race and ethnicitiy‐facility minority percentage composite groups adjusted for resident and facility characteristics as well as clustering by facility. The type III chi‐square for the composite race‐minority facility percentage variable was 320.46 with 23 degrees of freedom corresponding to a P<0.0001.

For Sources of Funding and Disclosures, see page 3.

心房颤动 (AF) 在长期护理 (LTC) 设施中很普遍。真实世界的数据表明，与门诊环境中的白人患者相比，患有 AF 的黑人患者接受抗凝治疗的可能性低 10% 至 22%。¹目前尚不清楚 LTC 环境中是否也存在这种差异。

LTC 护理是高度隔离的。与白人居民相比，种族和少数族裔居民在财政资源更有限且质量不足的设施中接受护理。²以前没有人研究过设施的作用，特别是特定设施中种族和少数族裔居民的百分比，以及它与卒中和 AF 风险较高的居民使用抗凝药的关系。

考虑到基于理论的结构性种族主义³ ，我们测量了居民种族和族裔以及种族和少数族裔居民在美国 LTC 设施中使用抗凝药物的设施百分比。

虽然我们不允许共享我们的数据，但通讯作者可以根据要求提供我们统计程序的技术附录。

我们纳入了 2015 年 7 月至 2015 年 12 月期间居住在 LTC 环境（即在疗养院中超过 100 天）的AF 和 CHA ₂ DS2‐VASc 评分≥2 的居民，并连续参加了 7 个月的 Medicare D 部分，直至并包括他们的指数评估。如果居民在最小数据集 (MDS) 或国际疾病分类第十修订版( ICD-10 ) 诊断代码中记录了与文献中先前示例一致的 AF的心律失常记录，则认为他们患有AF。⁴我们排除了接受临终关怀的居民。鉴于其身份不明的性质，马萨诸塞大学机构审查委员会批准我们的分析免于联邦监管审查。

我们使用 Medicare D 部分数据跟踪抗凝。具体来说，如果抗凝药物记录的填写日期加上该记录中提供的天数与考虑药物积累和最近住院的指数评估重叠，我们就考虑了一名接受抗凝治疗的患者。

我们在 Medicare 受益人调查文件中使用 Research Triangle Institute–Race Code 变量确定了种族和民族。该变量包括非西班牙裔白人种族（以下简称“白人”种族）、黑人种族、西班牙裔种族、美洲印第安人/阿拉斯加原住民种族、亚洲种族或其他种族（即未由其他医疗保险种族或种族类别）。Research Triangle Institute-Race Code 变量通过基于姓氏的插补改进了西班牙裔种族的确定。然后，我们计算了每个设施中种族和少数民族居民的百分比，并将每个设施分为 4 个类别（大致四分位数），基于百分比 0% 到 10%、10% 到 20%、20% 到 50%、>50% . 最后，我们将每位患者分配到种族设施少数百分比复合类别中。

我们检查了 MDS 中可用的多个居民特征，以调整潜在的混杂因素。这包括几种老年病，包括跌倒史、对日常生活活动的依赖、行动障碍、认知障碍、低体重指数和体重减轻。我们还调整了卒中风险（即 CHA ₂ DS ₂ ‐VASc 评分）和出血风险因素（肝硬化、贫血、肾病、使用抗血小板药物）。

我们还检查了营利性状态（营利性与非营利性）、连锁状态（与独立机构）、人员配备比率（注册护士和执业护士与患者）、医疗补助居民百分比以及医疗保险和医疗中心医疗补助 (CMS) 5 星质量评级。

我们构建了一个带有广义估计方程的二项式逻辑回归模型，以测量患者种族和种族-设施少数群体中抗凝药物使用的百分比，并根据居民和设施特征以及设施聚类进行调整。

我们确定了居住在 6183 个 LTC 设施中的 199822 名符合条件的居民。其中，60 065 名居民（29.8%）接受了抗凝治疗。居民的平均年龄为83.1±9.7岁。大多数居民是白人（81.8%）。白人居民的抗凝药物使用率最高（30.5%），其次是黑人居民（27.1%）、西班牙裔居民（26.3%）和亚裔居民（21.6%）。在检查居民种族和民族-设施少数民族百分比复合分组时，我们发现在种族和少数民族居民百分比较低（0%-10%）的设施中抗凝率较高，而在具有较高百分比的设施中抗凝率较低。种族和少数族裔居民（>50%）。这在居民种族和民族类别中是一致的（除了美洲印第安人，我们的人数很少，n=731；

抗凝药使用率普遍较低（28.7%），居民种族和民族的抗凝药使用率显着降低。居住在其中的种族和少数族裔居民比例较高的设施对几乎所有种族和民族类别的抗凝剂使用率较低。

我们的发现与 Christian 等人在 20 年前报告的结果相似。⁵他们发现，在有抗凝指征的情况下，与白人居民相比，种族和少数族裔居民接受抗凝治疗的可能性要低 20% 以上。我们通过分析设施特征（包括种族和少数民族居民的百分比）扩展了先前的工作。尽管疗养院质量改进工作，包括 CMS 质量报告，但在抗凝处方方面，对种族和少数族裔居民的护理提供仍然不公平。

即使在调整了广泛的居民和设施特征列表（包括质量评级和人员配备比率）后，我们仍发现抗凝药物使用存在种族差异。我们研究结果的局限性包括，自 2015 年以来趋势也可能发生了变化。鉴于许多患者接近生命的尽头，这一人群的目标抗凝率应该是多少也不清楚。缺乏关于不使用抗凝剂的原因和心脏病学咨询的信息也是局限性。更好的居民教育和增加心脏病学转诊代表了潜在的补救措施。

我们通过 Bristol Myers Squibb 和 Pfizer 作为 ARISTA-USA（美国血栓形成研究者发起的研究计划）的一部分的研究奖获得的资金进行了这项研究。

Kapoor 博士通过其独立的学习和变革资助机制获得了辉瑞（Pfizer）的研究资助，并获得了百时美施贵宝（Bristol Myers Squibb）的独立医学教育资助。最近，他通过由辉瑞和百时美施贵宝组成的联盟裁定和资助的竞争程序获得了研究资助。他还获得了辉瑞 (Pfizer) 的一项赠款，用于检查患者和提供者之间的对话。McManus 博士获得了 Bristol Myers Squibb、Boeringher Inghelheim、Pfizer、Biotronik 和 Philips Healthcare 的赞助研究支持，并为 Bristol Myers Squibb、FlexCon、三星、飞利浦和辉瑞提供咨询。McManus 博士拥有 Mobile Sense Technologies, LLC 的股权。Gurwitz 博士是 United HealthCare Pharmacy and Therapeutics Committee 的成员。萨迪克、帕特尔、张博士、

每个条形上方的值代表居民种族和种族-设施少数民族百分比复合组中抗凝剂使用的百分比，根据居民和设施特征以及设施聚类进行调整。复合种族-少数族裔设施百分比变量的 III 型卡方为 320.46，具有 23 个自由度，对应于P <0.0001。

有关资金来源和披露信息，请参见第 3 页。