Recent Trends in Oral Anticoagulant Use and Post-Discharge Complications Among Atrial Fibrillation Patients with Acute Myocardial Infarction

Abstract

Background

Atrial fibrillation (AF) is a common complication of acute myocardial infarction (AMI).The CHA2DS2VAScand CHADS2risk scoresare used to identifypatients with AF at risk for strokeand to guide oral anticoagulants (OAC) use, including patients with AMI. However, the epidemiology of AF, further stratifiedaccording to patients’ risk of stroke, has not been wellcharacterized among those hospitalized for AMI.

Methods

We examined trends in the frequency of AF, rates of discharge OAC use, and post-discharge outcomes among 6,627 residents of the Worcester, Massachusetts area who survived hospitalization for AMI at 11 medical centers between 1997 and 2011.

Results

A total of 1,050AMI patients had AF (16%) andthe majority (91%)had a CHA2DS2VAScscore >2.AF rates were highest among patients in the highest stroke risk group.In comparison to patients without AF, patients with AMI and AF in the highest stroke risk category had higher rates of post-discharge complications, including higher 30-day re-hospitalization [27 % vs. 17 %], 30-day post-discharge death [10 % vs. 5%], and 1-year post-discharge death [46 % vs. 18 %] (p < 0.001 for all). Notably, fewerthan half of guideline-eligible AF patientsreceived an OACprescription at discharge. Usage rates for other evidence-based therapiessuch as statins and beta-blockers,lagged in comparison to AMI patients free from AF.

Conclusions

Our findings highlight the need to enhance efforts towards stroke prevention among AMI survivors with AF

Introduction

Atrial fibrillation (AF) is a cardiac arrhythmia that affects thousands of hospitalized Americans, with an increasing prevalence in the United States1,2. Atrial fibrillation is frequently observed as a complication of acute myocardial infarction (AMI), affecting 5-13%of all patients suffering an AMI3-7. New-onset and prevalent AF is not only associated with higher rates of in-hospital complicationssuch as heart failure, stroke, anddeath7-10, but also a higher likelihood of re-hospitalization among hospital survivors of an AMI7.

The CHADS2 and CHA2DS2VAScscoring systemsare used to stratify patients with AF at risk for developing strokeand identify patients who maybenefit from oral anticoagulation (OAC), including patients with AMI8,11,12.Although antiplatelet drugs prevent thromboembolic complications in patients with AMI, particularly those receiving percutaneous interventions (PCI), these agents have minimal impact on stroke risk in patients with AF11. Therefore,prescription ofOAC in addition to antiplatelettherapy isrecommended for AMI patients with AF consistently in the ESC and ACC/AHA/HRS guidelines11,13.However, practice patterns vary widely, particularly when dual antiplatelet therapy is prescribed among AMI patients,largely due to concerns about the bleedingrisk from “triple therapy,”but also because many providers view new-onset AF following AMI as being associated with lower risk for stroke14. Observational studies have lent credibility to these concerns by showing an elevated bleeding risk among patients treated with dual antiplatelet agents and OAC15.

Few investigations have described the epidemiology of AF among patients with AMI, further stratified according to stroke risk, or trends in OAC prescription and post-discharge complications among guideline-eligible AF patients with AMI. Therefore, we analyzeddata from the multi-hospital, population-based, Worcester Heart Attack Study (WHAS)7,8,16,17.

Methods

The WHAS is an ongoing, population-based, observational study of patients hospitalized at all 11 medical centers in central Massachusetts, documentinglong-term trends in the incidence, morbidity, mortality, and complications of AMI7,8,1,17. Our analyses focusedon patientswho were hospitalized with a discharge diagnosis of AMI at all Worcester Standard Metropolitan Statistical Area (SMSA) hospitals during 8 biennial years between 1997 and 2011. We selected these study years based on the availability of data on AF status, CHADS2 and CHA2DS2VAScscores, and systematic collection of OAC prescription rates and post-discharge outcomes.Since this study primarily focused on post-discharge outcomes and OAC prescription, patients who died during hospitalization were excluded. Medical records of patients admitted for possible AMI at all metropolitan Worcester medical centers were reviewed and validated. Diagnosis of AMI was confirmed using pre-established criteria18.

The methods used for the identification of patients hospitalized with AMI and various comorbidities and complications have been described in prior publications7,8,16,17. The data collected included patients’ age, gender, race, comorbidities (hypertension, heart failure, stroke, hyperlipidemia, chronic obstructive pulmonary disease [COPD], chronic kidney disease [CKD], diabetes), type of AMI,physiologic parameters (admission heart rate, blood pressure, respiratory rate, serum creatinine and peak troponin), in-hospital and discharge medications, in-hospital complicationsand post-discharge outcomes. Rates of warfarin, aspirin, and clopidogrel prescription were determined based on the discharge medication list. Rates of direct oral anticoagulant prescriptionwere not examined in this study, as these agents were not approved for use in the US until 2010, nor collected in this study until after 2011. Rigorous methods have been used to identify deaths and cases of readmission to area hospitals7.

Patient’s AF status was determined based on available clinical data7,16. A review of ambulance transport records, emergency admission notes, progress notes, and in-hospital 12-lead ECGs was conducted to identify patients with AF.The classification of prevalent AF was assigned if a patient had prior AF noted in their hospital admission note or any progress note. A case of incident, or newly diagnosed, AF was defined according to the following criteria: No documentation of pre-existing AF anda) AF on any 12-lead ECG obtained during the index hospitalization8,17, or b) new-onset AF documented in any clinical note during the index hospitalization. For purposes of focusing on post-discharge outcomes and discharge prescription practices, we included hospital survivors with both incident and prevalent AF. Patients who underwent coronary artery bypass grafting (CABG) during hospitalizationwere excluded from the study due to the different mechanisms and natural historyof surgical patients with AF19.

Patients with confirmed AMI were further stratified according to their CHA2DS2VAScscore based on documentation of risk factors in the medical record [age, sex, history of heart failure, hypertension, diabetes mellitus, stroke or transient ischemic attack, vascular disease (including myocardial infarction and peripheral artery disease)]11,13.Patients were grouped into 3 categories according to their CHA2DS2VASc score (Group 1 =CHA2DS2VASc score 1-2, Group 2 = CHA2DS2VASc 3-6, Group 3 = CHA2DS2VASc 7-9) to distinguish between those at lower, high, and very high risk of thromboembolism20-23. Although use of the CHA2DS2VASc stroke risk score was introduced in 2009, the CHADS2 scoring system was in use for most of the period under study. Therefore, we also stratified the cohort according to CHADS2score (congestive heart failure, hypertension, age, diabetes, and stroke)using similar methods,but divided the population into two groups – those with CHADS2 score less than two and those with a CHADS2 score of two or more, in order to identify the proportion of the population eligible for OAC prescription based on contemporary guidelines11,13.

We examined differences in the frequency and outcomes of patients withAF compared to those withoutAF through the use of chi-square tests and one-way analysis of variance (ANOVA) fordiscrete and continuous variables, respectively. We examined differences in the characteristics and use of different evidence-based AMI and AF therapies according to stroke risk classifications through the Cochran-Armitage test for trends. Differences in post-discharge complications, including hospital re-admission and mortality, as well as the associated crude and multivariable adjusted odds ratios (OR) with 95% confidence intervals (CI), were assessed. The odds of developing these complications were adjusted for confounding demographic and clinical factors.

We restricted our analyses on post-discharge prescription rates of OAC to the study years 2003-2011 because data on OAC was not collected in this study until 2003. HAS-BLED scores were calculatedto characterize participants’ risk of bleeding on OAC using the following factors documented in the medical record - age,hypertension, abnormal renal/liver function, stroke, bleeding history, drugs or alcohol use (1 pointeach)11,21].Since INRs prior to admission were not abstracted as part of this population-based study, we were not able to determine time in therapeutic range over the preceding 4 weeks among individuals treated with vitamin K antagonists. Hospital re-admission data was collected at only 3 of the major Worcester hospitals, and was reported for the years 1999-2011. All analyses were performed using SAS version 9.3 (SAS Institute Inc., Cary, NC).

Results

The baseline characteristics of 6,627 patients discharged from all participating medical centers in central Massachusetts with an independently confirmed AMI during the years under studyare shown in Table 1. The mean age of this population sample was 69 years, 40% were women, and approximately two-thirds presented with an initial AMI. A total of 1,050patients (16%) had AF. Of these, the majoritywereat a high (76 %) or very high risk (15 %) for stroke according to their CHA2DS2VAScscore. Patients in the “lower” stroke risk category were more likely to have new-onset AF compared to prevalent AF, whereas patients with AF in the highest stroke risk group were older, more likely to be male, have higher serum troponin levels, have a history of CKD, and have higher average HAS-BLED scores. Patients with AF and higher CHA2DS2VAScscores were also more likely to have their AMI hospitalization complicated by development of heart failure or stroke.We also observed that in patients with AMI and AF, bleeding risk (based on HAS-BLED score) tracked closely with stroke risk (Table 1)

Table 1. Demographic and Clinical Characteristics of Hospital Survivors of Acute Myocardial Infarction (AMI), further stratified according to Atrial Fibrillation (AF) Status.

* Complicated AMI included the development of cardiogenic shock, heart failure, stroke, recurrent AMI, or heart block.

Characteristics	No AF	AF-	AF-	AF-	p-value
	[n=5577]	CHA2DS2V	CHA2DS2V	CHA2DS2V
		ASc =1-2	ASc=3-6	ASc=7-9
		[n=96]	[n=798]	[n=156]
Age (mean,	67.9	58.4 [8.6]	78.3 [9.78]	81.8 [6.1]	<. 001
years)[SD]	[14.3]
Age <65 y (%)	39.2	68.7	8.1	1.3	<. 001
Age 65-74 y	21.3	16.6	21.5	8.3	<. 001
Age 75-84	24.1	0	37.8	53.2	<. 001
Age >= 85 y	15.3	14.6	32.4	37.2	<. 001
Sex, % Men	57.7	87.5	52.7	32.7	<. 001
White Race	89.4	89.6	92.5	94.2	0.01
AF Type
Incident	0	87.5	58.4	48.1	<. 001
Prevalent	0	12.5	41.6	51.9	<. 001
Medical History (%)
Angina Pectoris	16.2	10.4	16.8	21.1	0.15
Hypertension	69.1	26.1	77.9	96.1	<. 001
Heart Failure	20.9	5.2	29.8	64.1	<. 001
Stroke	10.6	0	7.0	64.1	<. 001
Diabetes	32.9	6.2	30.9	67.3	<. 001
Hyperlipidemia	54.9	44.8	51.2	56.4	0.06
COPD	15.9	17.7	22.7	19.2	<. 001
Prior bleeding	22.4	19.8	33.4	28.8	<. 001
CKD	16.2	13.5	22.9	35.9	<. 001
STEMI	36.6	57.3	28.8	27.5	<. 001
NSTEMI	63.4	42.7	71.2	72.5	<. 001
Initial AMI	65.0	82.3	62.4	58.3	<. 001
Creatinine in mg/dl	1.3 [1.0]	1.2 [0.7]	1.5[1.1]	1.6 [0.8]	<. 001
- Mean [SD]
Troponin I Peak in	16.7	16.6 [49.6]	20.7 [57.9]	24.9 [75.6]	.029
ng/mL – Mean [SD]	[65.5]
Mean HASBLED	1.6 [1.0]	0.7 [0.8]	2.0 [0.8]	3.0 [0.7]	<. 001
score [SD]
Mean CHADS2VASc	3.8 [1.9]	1.6 [0.4]	4.5 [1.0]	7.4[0.6]	<. 001
In-Hospital Complications (%)
Complicated AMI*	32.6	39.6	58.3	64.7	<. 001
Cardiogenic Shock	2.1	9.3	7.7	2.5	<. 001
Heart Failure	30.3	33.3	54.2	60.9	<. 001
Stroke	1.1	1.1	2.5	5.8	<. 001
Heart Block	1.2	6.3	3.1	2.5	<. 001
Recurrent AMI	0.7	0	0.5	0	0.5
Hospital Procedures (%)
Cardiac	59.2	71.9	43.7	30.8	<. 001
Catheterization
PCI	42.8	63.5	31.2	18.6	<. 001

Rates of AF increased significantly between 1997 and 2003, after which time AF rates remained relatively stable, and patterns did not appear to differ significantly according to CHA2DS2VAScstroke risk category(Figure 1).Rates were highest among patients with AF in the highest stroke risk group, but the relative proportion of individuals in the lowest stroke risk category appeared to increase during the most recent study years. Similar results were seen in AF rates stratified by CHADS2 scores, with higher rates observed among patients with AF and a CHADS2score≥2(Figure 2).

Figure 1. Trends in the Rates of AFby CHA2DS2VASc Score between 1997 and 2011.

Figure 2. Trends in the Rates of AF by CHADS2 Score between 1997 and 2011.

In comparison to AMI patients who remained free from AF during their hospitalization, individuals with AF in the highest stroke risk category had significantly higher rates of post-discharge complications, including higher 30-day re-hospitalizationrates [27 % vs. 17 %], 30-day post-discharge deathrates [10 % vs. 5%], and 1-year post-discharge deathrates [46 % vs. 18%] (p < 0.001 for all). Notably, we observed a direct relationship between CHA2DS2VASccategory and rates of readmission or death within 30 days, as well as death rates at 1-year (Table 2). Furthermore, after adjustment for several demographic factors, comorbid conditions, and AMI-associated characteristics, AF patients in the highest stroke risk group had2-fold higher oddsof dying at 1-year after discharge, compared to patients free from AF.However, adjustment for the same set of key covariates attenuated crude associations observed between stroke risk and readmission, as well as mortality 30-days after discharge (Table 2).

Table 2. Post-Discharge Outcomes* among Hospital Survivors of Acute Myocardial Infarction (AMI), further stratifiedaccording to Atrial Fibrillation (AF) Status.

* Outcomes presented as Percentage along with Adjusted Odds Ratio (OR) with 95 % Confidence Intervals, compared to No AF (Referent Group). Adjusted for sex, race, history of angina, hyperlipidemia, Chronic Obstructive Pulmonary Disease, Chronic Kidney Disease, ST-segment Elevation Myocardial Infarction, Initial Acute Myocardial Infarction, bleeding and other complications.

Outcome	No AF	AF- CHA2DS2VASc	AF-	AF-
		1-2	CHA2DS2VASc	CHA2DS2VASc
			3-6	7-9
30 - day Re-	17.1	16.2 %	27.1 %	27 %
hospitalization**		OR 0.9 (0.5-1.7)	OR 1.5 (1.2-1.8)	1.2 (0.7-1.7)
30-day Mortality	4.9	7.3 %	8.8 %	10.3 %
		OR 1.5 (0.7-3.4)	OR 1.3 (1-1.7)	OR 1.3 (0.7-2.3)
1- year Mortality	18	14.6 %	33.9 %	46.1 %
		OR 0.8 (0.4-1.5)	OR 1.7 (1.4-2.1)	OR 2.2 (1.6-3.2)

We also examinedthe use of evidence-based medications for AMI or AF at discharge, further stratified by stroke risk. Rates of OAC use were highest among patients with highest CHA2DS2VASc scores (38% in the very high stroke risk, 35% in the high stroke risk, and 31% in the lowest stroke risk), but less than half of patients at high or very-high stroke risk received an OAC prescription at discharge. Interestingly, aspirin and clopidogrel use were also lower among AMI patients with AF at highest stroke risk,compared to those at lower risk (88.5 % and 58.3% among those with AF in the lowest stroke risk category vs. 78.2 % and 33.9% among those with AF highest stroke risk, respectively). In contrast, prescription of beta-blockers and lipid-lowering agents did not vary among AF patients based on stroke risk (Table 3).

Table 3. Discharge Medication Prescription (%) among Hospital Survivors of Acute Myocardial Infarction with Atrial Fibrillation (AF), further stratified according to CHA2DS2VASc Score.

Data on Warfarin and Triple Therapy Rates collected from 2003-2011 only. n=3910.

Medication	No AF	AF-	AF-	AF-	p-value
		CHA2DS2VASc	CHA2DS2VA	CHA2DS2V
		1-2	Sc 3-6	ASc 7-9
Warfarin*	7.5	31.1	34.6	37.9	< .001
Dual Antiplatelet	44.7	43.7	26.5	25	< .001
Therapy
Aspirin only	86.4	88.5	80.7	78.2	< .001
Clopidogrel only	50.1	58.3	38.8	33.9	< .001
Beta Blockers	83.9	79.1	77.9	80.7	< .001
ACE/ARB	59.5	62.5	58	64.7	0.9
Lipid Lowering Agents	63.8	65.6	55.3	62.1	< .001
Triple Therapy *	4.4	19.7	15.2	10.3	< .001

Notably, only 1 in 4 AMI patients with AF and a high or very high stroke risk score received dual antiplatelet therapy at hospital discharge. Low rates of OAC prescription at discharge were also observed when stratifying the patients based on their CHADS2 score(Table 4). We did not observe any significant change in the rates of OAC prescription over time among individuals with AF (Table 5).Lastly, in AF patients at elevated stroke risk, we observed no significant relations between bleeding risk and rates of OAC prescription at discharge (Table 6).

Table 4. Post-Discharge Prescription Rates of Warfarin, Antiplatelet Agents and Triple Therapy from 2003-2011 in Patients with Acute Myocardial Infarction (AMI) and Atrial Fibrillation (AF), further stratified by CHADS2 Scores.

DAPT/Dual Antiplatelet Therapy = Aspirin + Clopidogrel

Triple Therapy = Aspirin + Clopidogrel+Warfarin

Note: Data available from 2003-2011 only (n=3910).

		CHADS2≤1					CHADS2≥2
Year	Warfa	Aspirin	Clopidog	DAPT	Triple	Warfa	Aspirin	Clopido	DAPT	Triple
	rin	Alone	rel Alone	(%)	Therapy	rin	Alone	grel	(%)	Therapy
	(%)	(%)	(%)			(%)	(%)	Alone		(%)
								(%)
2003	35.3	29.4	5.9	51.0	17.7	32.5	40.8	3.2	36.3	7.6
2005	21.2	12.1	3.0	66.7	6.1	37.0	38.7	3.4	47.9	16.0
2007	30.6	38.9	2.8	50.0	16.7	38.8	38.8	0.0	50.5	14.6
2009	39.1	30.4	8.7	56.5	17.4	39.7	33.3	1.6	58.7	25.4
2011	36.4	30.3	0.0	60.6	24.2	32.8	27.7	3.1	54.7	15.6

Table 5. Yearly Trends in Discharge Anticoagulation Prescription (%) in Patients with Acute Myocardial Infarction (AMI) and Atrial Fibrillation (AF), further stratified by CHA2DS2VASc scores.

Triple Therapy = Aspirin + Clopidogrel+Warfarin

Data collected from 2003-2011 only (n=3910).

Medication	AF Status	2003	2005	2007	2009	2011
Warfarin	No AF	6.2	6.6	8.3	8	9.1
	AF-CHA2DS2VASc	29.4	9.1	28.6	70	25
	1-2
	AF- CHA2DS2VASc	33.7	38.2	33	31.8	35.8
	3-6
	AF- CHA2DS2VASc	32.3	23.1	51.7	53.8	35.7
	7-9
Triple Therapy	No AF	2.3	4.4	5.5	5.5	4.9
	AF- CHA2DS2VASc	17.6	0	28.6	40	18.8
	1-2
	AF- CHA2DS2VASc	10.9	16.5	12.6	20.6	22.4
	3-6
	AF- CHA2DS2VASc	2.9	7.7	20.7	23.1	0
	7-9

Table 6. Discharge Medication Prescription (%) among Hospital Survivors of Acute Myocardial Infarction (AMI) with Atrial Fibrillation (AF), further stratified according to Risk of Bleeding.

DAPT/Dual Antiplatelet Therapy = Aspirin + Clopidogrel

Triple Therapy = Aspirin + Clopidogrel+Warfarin

Low HASBLED= Score <3

High HASBLED = Score >=3

* Data on Warfarin and Triple Therapy Rates collected from 2003-2011 only (n=3,910).

Medication	Overall	AF-CHADS2VASc	AF-CHADS2VASc	p- value
	N=1050	1-2	<2
		N=96	N=954
	Low	High	Low	High	Low	High
	HASBL	HASBL	HASBL	HASBL	HASBL	HASBL
	ED	ED	ED	D	ED	D
Warfarin*	36.4	32	31.6	25	37.1	32.1	0.2
DAPT	28.7	25.9	43.4	50	26.6	25.6	0.3
Aspirin only	82.2	78.2	89.1	75	81.2	78.2	0.1
Clopidogrel only	41.3	36.5	58.7	50	38.8	36.3	0.1
Triple Therapy *	17.5	9.7	21	50	0	17	9.8	0.005

Discussion

The CHA2DS2VAScscoring system is a well-validated stroke risk stratification tool for patients with non-valvular AF24. Components of the CHA2DS2VAScscoreare readily available prognostic risk factors for outcomes of ischemicheart disease25.The CHA2DS2VAScscore has been shown to successfully estimate the risk of adverse events in patients with AMI24 and alsoidentify post-STEMI patients at high risk of developing new onset AF and stroke26.

There arelimited studies that have evaluated the prognostic utility of stroke risk classification schemes in predicting the risk of adverse outcomes in patients with concomitant AMI and AF.In a study ofmore than 15,000 patients admitted with AMI tohospitals in Korea, the investigators assessed the utility of CHA2DS2VAScscores in predicting the risk of dying and/or recurrent MIin patients with and without AF. They found that higher CHA2DS2VAScscoreswereassociated with an increased risk of adverse cardiovascular outcomes in AMI patients at 1,6,12, and 24 months, irrespective of the presence of AF27. In our community-wide study involving almost 7,000hospital survivors, patients with AF and a higher CHA2DS2VAScscores were at greater risk for both short- and long- term adverse events than both AF patients with lower CHA2DS2VAScscores, and patients without AF. In particular, weobserved a direct relationship between increasingCHA2DS2VASc scores and odds of dying at 1-year after discharge in patients with AF, compared to patientsfree from AF. This was not seen in AF patients with a low CHA2DS2VAScscore (Figure 3).Our findings expand on the results of the previousKorean study27, and suggest that the CHA2DS2VAScsystem can be used as a toolto predict survival among patients with AMI and AF.

Figure 3. One-Year Mortality (Odds Ratio with 95 % Confidence Intervals) among AMI Patients with AF stratified by CHA2DS2VASc scores, compared to AMI Patients without AF.

Current AF and AMI management guidelines support the use of dual antiplatelet therapy plus OAC when independent indications for AMI and AF exist28,29. However, institution of “triple therapy” poses risks for bleeding. Inasmuch, alternative strategies have been proposed for patients with AMI and AF, such as discontinuing antiplatelet drugs among patients who are greater than 12 months out from undergoing a PCI30. In clinical practice, there is significant individualization of treatment with respect to balancingthe risks of stroke and bleeding.In our study, approximately 90% of patients with AMI and AF had a high or very high risk for thromboembolism based on having a CHA2DS2VAScscore>2. However, only 1 out of4 AMI patients with AF received dual antiplatelet therapyat the time of hospital discharge andless than half received OAC at discharge, suggesting thata treatment gap between guideline and real-world rates of prescription exists.

Reasons for the low rates of OAC usage among AMI patients with AF at hospital discharge may include lack of awareness of guidelines, provider or patient concerns over risk for bleeding with the use of OAC and antiplatelet drugs, or under-estimation of stroke risk from AF complicating AMI. Since the FDA approved the first target specific OAC agent, dabigatran, for use in 2010 and our study ended in 2011, we do not believe that the low rates of OAC seen in our study relate to the use ofvitamin-K antagonist alternatives. Furthermore, we have generally observed high rates of adherence to AMI-specific guideline recommendations for effective patient management in this community-wide study, with the majority of eligible patients having been treated with aspirin, statins, beta-blockers and prompt percutaneous revascularization for STEMI31. This was corroborated in our analysis as the vast majority of AMI patients with AF received guideline-directed antiplatelet, lipid-lowering agents, and beta-blocker therapies (Table 3). Therefore, we do not believe that the low rate of observed OAC prescription reflects a general disregard among providers at participating hospitals for guideline recommendations. An alternative explanation for our findings is that, despite evidence to the contrary7, hospital providers may consider AF complicating an AMIto be of lesser clinical significance than AF in the community, irrespective of their CHADS2 or CHA2DS2VAScscores.

Neither treatment nor outcomes of AFare mandatedpublicallyreported quality metrics.It has not been until recently that the American Heart Association created and distributed a ‘Get with the Guidelines’ for AF, a mechanism to enhance adherence to guideline-recommendations for AF treatment, including use of OAC for patients at intermediate and high stroke risk32. It is certainly possiblethat clinicians managing patients with AMI during the period under study may not have been aware of the importance of stroke risk estimation and OAC prescription for at-risk patients. It is our hope that, through increased public and clinician AF awareness, adherence to guideline recommendations for OAC among eligible AMI patients with AF will increase.

Strengths of the present study includethe population-based design involving all patients hospitalized with AMI across central Massachusetts. Inclusion of all regional hospitals in Worcester, the large number of patients hospitalized with AMI and AF, and the use of standardized criteria for diagnosing AMI and AF were additional strengths. We were also able to control for a variety of potential confounding variables outside of the CHA2DS2VAScscoring system that could have contributedtopost-discharge outcomes.

Limitations

Limitations of this study include the lack of data on type of AF (paroxysmal, persistent, or permanent) and the inability to determine the time in therapeutic INR range among patients with AF prior to admission. Our study population likely included a number of patients at high risk for bleeding, and we did not systematically assess components of frailty, such as falls, weight loss, or grip strength, which may influence decisions to withhold OAC. We also did not assess for patient-level factors, such as patient’s attitudes about OAC, allergies or prior adverse reactions that may have led to the withholding of OAC’s at the time of hospital discharge.Data on prescription rates of antiarrhythmic medications in patients with AF was not collected. Since this study was conducted in a single geographic region, there was potential for under-estimation of re-hospitalization rates, as patients may have presented outside of the hospital network for subsequent hospitalizations.Moreover, hospital re-admission data was reported at only 3 of the major Worcester hospitals, although these comprise the majority of all AMI admissions in this population.Data on the etiology of death after hospital discharge was not available.

Conclusions

We observed thata large proportion of patients hospitalized for AMI over a 14-year period had AF,and that the proportion of AF patients at high risk for strokeremained elevated throughout the study. Our study suggests that the CHA2DS2VASc tool can be effectively used to identify AMI patients with AF who are at greater risk for post-discharge complications, particularly mortality and hospital re-admission. Another interesting finding was that less than half the patients with a high CHA2DS2VASc were discharged on anticoagulation therapy. Our findings suggest significant deviation from guideline-based AF management practices in the context of AMI, and that patients with higher CHA2DS2VASc scores may benefit from greater monitoring and/or more aggressive treatment in the peri-discharge period. Future studies are needed to see if the CHA2DS2VASc score can truly be validated as tool for not only predicting stroke risk, but also post-discharge outcomes (including mortality) in patients with AMI and AF.