Event Rates and Risk Factors for Recurrent Cardiovascular Events and Mortality in a Contemporary Post Acute Coronary Syndrome Population Representing 239 234 Patients During 2005 to 2018 in the United States.

Background Patients with acute coronary syndrome (ACS) are recognized by guidelines as remaining at high risk for adverse outcomes. Evidence from contemporary, representative ACS populations in a clinical practice setting is necessary to identify subgroups and strategies for improving patient outcomes. We aimed to describe event rates and risk factors in an ACS population over prolonged follow-up for cardiovascular end points. Methods and Results We identified 239 234 patients in the Optum Research Database (57.2% men; mean [standard deviation] age, 69.2 [12.2] years) with evidence of an ACS hospitalization (index ACS) during January 1, 2005 through December 30, 2018. Subgroups were based on index ACS event (myocardial infarction/unstable angina and revascularization status) and the Thrombolysis In Myocardial Infarction Risk Score for Secondary Prevention. The 5-year event rate for the primary end point representing nonfatal myocardial infarction, nonfatal ischemic stroke, and cardiovascular death was 33.4% (95% CI, 33.1%-33.7%; P<0.001). The risk of experiencing the primary end point was ≈6-fold higher immediately after discharge (≈40.9% annualized risk) as compared with the period 1+ years after hospitalization (≈6.4% annualized risk). Among subgroups, the 5-year primary end point event rate was highest for myocardial infarction without revascularization and a Thrombolysis In Myocardial Infarction Risk Score for Secondary Prevention ≥4, at 47.9% (95% CI, 47.3%-48.4%; P<0.001) and 56.7% (95% CI, 55.9%-57.4%; P<0.001), respectively. Conclusions Patients with ACS remain at very high risk of experiencing recurrent cardiovascular events, particularly early after discharge, with identifiable subgroups at multifold higher risk of specific clinical end points.

American College of Cardiology

American Heart Association

Thrombolysis In Myocardial Infarction Risk Score for Secondary Prevention

unstable angina

Clinical Perspective

What Is New?

Characteristics, event rates, and predictors of acute coronary syndromes (ACS) were evaluated in a large, contemporary US clinical practice population of 239 234 patients with an ACS hospitalization, from the time of hospital discharge through long‐term follow‐up.

Subgroups based on the index ACS type, index revascularization, and the Thrombolysis In Myocardial Infarction Risk Score for Secondary Prevention were analyzed separately to identify those who would derive the greatest clinical benefit from guideline‐based therapies.

The Thrombolysis In Myocardial Infarction Risk Score for Secondary Prevention was applied to determine its usefulness in risk prognostication in a usual clinical practice population representing patients hospitalized for ACS.

What Are the Clinical Implications?

This ACS analysis may prove useful to payers, regulatory authorities, clinicians, and clinical guidelines committees.

Our findings support the claim that patients with ACS remain at very high risk of experiencing recurrent cardiovascular events, particularly early after discharge, with subgroups such as those who experience myocardial infarction without revascularization at index and those with Thrombolysis In Myocardial Infarction Risk Score for Secondary Prevention ≥4 at multifold higher risk of specific cardiovascular outcomes.

The Thrombolysis In Myocardial Infarction Risk Score for Secondary Prevention demonstrated usefulness in risk stratification when applied to a large health care administrative claims database, potentially expanding its use to investigators working with similar databases and to inform decision making in clinical practice.

In the current era, health care databases are becoming larger, the coding has become more specific, and increased linkage with other patient‐level data has become possible. Better definition of patient populations, subgroups, and outcomes, both over immediate and long‐term follow‐up now provides greater value from these data. Although new therapies are brought into practice through clinical trials, these observational clinical practice databases provide additional information on existing treatment patterns, patient subgroups not represented in clinical trials, and the potential impact of therapies over extended follow‐up.

Patients with an acute coronary syndrome (ACS) are recognized by guidelines as remaining at high risk for future adverse outcomes. , , , , , Numerous medical therapies have not only been validated for secondary cardiovascular prevention, but specifically for patients after ACS. A refined understanding of the characteristics of this population, current treatment patterns, and risk over time is therefore necessary to understand current practice patterns and outcomes, but also which subgroups and strategies are optimal for additional risk mitigation. The primary aim of this study was to evaluate the characteristics, events rates, and risk predictors of ACS in a large and contemporary US clinical practice population, from the time of hospital discharge through long‐term follow‐up. Specific objectives included: (1) comparison of the characteristics and event rates of the myocardial infarction (MI) versus unstable angina (UA) population and whether these differed by treatment of the index ACS with revascularization; (2) comparison of the characteristics by sex; (3) evaluation of the Thrombolysis In Myocardial Infarction Risk Score for Secondary Prevention (TRS 2oP) , for risk stratification in a usual clinical practice population with an ACS event; (4) comparison of the similarity (and potential applicability) of our study findings with contemporary randomized controlled trials (RCTs) of ACS populations; and (5) identification of independent risk predictors for trial‐related end points, such as cardiovascular death, MI, ischemic stroke, UA, and elective coronary revascularization after discharge.

Methods

The data underlying the results presented in this study contain proprietary elements owned by Optum Research Database and therefore cannot be broadly disclosed or made publicly available. These underlying data can, however, be made available to a third party via the corresponding author upon reasonable request, assuming certain data security and privacy protocols are in place and that the third party has executed Optum’s standard license agreement, which includes restrictive covenants governing the use of the data.

The study cohort was developed from the Optum Research Database, one of the largest anonymized health care research databases in the United States, representing over 70 million patients with data from administrative claims and additional information from enrollment records. In 2018, the Optum Research Database represented 19% of the US population enrolled in commercial health plans and 21% enrolled in Medicare Advantage. Medicare Advantage represents health plans offered by a private company that contracts with Medicare to provide patients with hospital and medical insurance benefits. The study was submitted to the New England Institutional Review Board, received a Waiver of Authorization, and was determined to be exempt from institutional review board review because it did not meet the definition of human subject research (New England Institutional Review Board number 1‐9234‐1); informed consent was therefore not required.

Cohort Development

Patients with evidence of an ACS hospitalization between January 1, 2005 and December 30, 2018 were identified (Figure S1). For individuals experiencing multiple qualifying ACS hospitalizations, 1 event was randomly selected, which is subsequently referred to as the index ACS. This strategy was used because a patient’s characteristics and future risk will differ depending on whether the first, second, or another subsequent ACS event is chosen. Randomly selecting an ACS event helped ensure representativeness (see Data S1 for supporting sensitivity analysis). The index date was the date of hospital discharge and defined the initiation of follow‐up. Additional inclusion criteria were age ≥20 years, availability of 1‐year baseline data before the index date, and discharge codes indicative of alive status. A 1‐year baseline period was chosen because it represented an optimal tradeoff between a more accurate ascertainment of preexisting conditions and sample size (Table S1). To define baseline characteristics, a single claim was sufficient if it was identified from an inpatient stay; otherwise, 2 medical claims on 2 different dates of service were required. Diagnosis and procedure codes for ascertaining clinical conditions were thoroughly reviewed by coauthors, with support provided by subspecialty experts. Patients with a history of heart transplantation were excluded (Figure 1). Medication use at index ACS was determined by medication availability via a filled prescription immediately before the index ACS (Figure S2).

Figure 1

Consort diagram for patient selection.

Patients with an ACS hospitalization during January 1, 2005 to December 30, 2018 were initially selected. Sequential application of the study inclusion and exclusion criteria resulted in a final study population of N=239 234. ACS indicates acute coronary syndrome; and NDI, National Death Index.

End Points

Patients were followed after index ACS for recurrent cardiovascular events. The primary composite end point was defined as the first occurrence of cardiovascular death, nonfatal MI, or nonfatal ischemic stroke. The secondary end points were defined as the first occurrence of cardiovascular death, nonfatal MI, nonfatal ischemic stroke, nonfatal UA hospitalization, and nonfatal elective coronary revascularization, separately and as a composite. The estimation of event rates for these end points were based on a Kaplan‐Meier approach. Separately, we also estimated the event rates for the individual components that contributed to each composite using a competing risk analysis. , Diagnosis and procedure codes used for identifying these events were based on algorithms used and validated in previous reports (Table S2). , , , , , , , MI, ischemic stroke, and UA required inpatient hospitalization, whereas criteria for elective coronary revascularization was applied to either the inpatient or outpatient setting. The definition of elective coronary revascularization required that codes for MI, ischemic stroke, or UA not be present on the day of the procedure and included a 30‐day lookback period free of other events defined in the composite (eg, MI) to ensure the revascularization procedure was not related to these events. To facilitate an accurate determination of mortality status during follow‐up and the subsequent cause of death, the study population was integrated with the National Death Index. Deaths were classified as being cardiovascular‐related if the International Classification of Diseases, Tenth Revision, Clinical Modification (ICD‐10‐CM) diagnosis codes included I00‐I99 in the underlying cause of death field.

Cohort and Subgroup Characteristics

Baseline characteristics included demographics, atherosclerotic cardiovascular disease (ASCVD)‐related comorbidities, non‐ASCVD comorbidities, revascularization procedures at index ACS, medications, and categories based on the TRS 2oP, which is a 9‐point integer‐based risk score derived from clinical trials. , TRS 2oP assigns 1 point each (scale 0–9) for heart failure, hypertension, age ≥75 years, diabetes, prior stroke, prior coronary artery bypass grafting, peripheral arterial disease, estimated glomerular filtration rate <60, and smoking. TRS 2oP categories were: 0 to 1 point (low), 2 to 3 points (intermediate), or ≥4 points (high).

Baseline characteristics for subgroups were also described. These subgroups were: (1) presentations with MI or UA at index ACS; (2) revascularization (coronary artery bypass grafting or percutaneous coronary intervention [PCI]) at or during the 30 days following the index ACS; (3) MI with revascularization, MI without revascularization, UA with revascularization, or UA without revascularization; and (4) sex. Finally, the characteristics of the study cohort were descriptively compared with those of recent RCTs enrolling ACS populations to gain insights into the similarity and potential applicability of trial populations to our cohort(s). An informal search strategy was conducted based on the following criteria to select the trials for comparison: contemporary ACS trials conducted during our cohort identification period from 2005 to 2018 enrolling at least 4000 patients for whom reported baseline characteristics facilitated comparison with our study population (eg, index revascularization status).

Risk Estimation

The proportion of the study cohort defined by guidelines as very high risk for atherosclerotic events was estimated. The first definition used was from the 2018 American College of Cardiology (ACC)/American Heart Association (AHA) Guidelines for the Management of Blood Cholesterol, which defines patients at very high risk of ASCVD events as those with a history of multiple major ASCVD events or 1 major ASCVD event and multiple high‐risk conditions. In our study cohort, the index ACS qualified as a major ASCVD event, and further assessment for identifying patients at an increased risk of adverse outcomes was conducted by identifying patients presenting with multiple major ASCVD events or additional high‐risk factors.

Next, event rates for the composite end points and individual components of the composite were estimated via Kaplan‐Meier analyses for the overall cohort and subgroups. For the overall cohort, we also estimated the instantaneous hazard rates over time. In contrast to the more common concept of risk over a specified time horizon (eg, 30 days, 1 year, or 5 years), the instantaneous hazard rate is a measure of risk at a moment in time (see Data S2).

Statistical Analysis

Cox proportional hazards models were used to investigate the association of baseline characteristics with the risk of cardiovascular events during follow‐up. Separate models were fitted for each of the composite end points as well as individual components of the composite. Model parameters were estimated via a backward selection strategy with P>0.05 as the criteria for filtering nonsignificant variables. We retained index ACS event categories (MI/UA and revascularization status), age, and sex in the model, regardless of statistical significance. For the purpose of qualitative interpretation, we defined a relatively stricter criterion based on hazard ratio (HR) >1.5 (indicating a relatively high impact on risk) together with P<0.001 (indicating a high statistical significance) to identify the key factors influencing risk. All statistical analyses were performed with SAS software version 9.4 (SAS Institute, Cary, NC).

Results

The overall cohort included 239 234 patients. Median follow‐up duration was 1.2 years (interquartile range, 0.4–2.8 years), with follow‐up of ≥5 years available for 24 648 patients (10.3%) (Table S3). The mean age was 69 years, 57% were men, and 72% had Medicare Advantage (Table 1). According to the 2018 ACC/AHA guidelines, ≈97% of the cohort was classified as very high risk. Specifically, 87% had a history of at least 2 high‐risk conditions, and an additional 10% had a history of 1 of these high‐risk conditions.

Table 1

Baseline Characteristics: Overall Cohort and Subgroups of Type of Index ACS and Use of Revascularization

	Myocardial infarction, n=169 220	Unstable angina, n=70 014	ACS with revascularization, n=101 476	ACS without revascularization, n=137 758	Total, N=239 234
Demographic characteristics
Age, y, mean (SD)	70.2 (12.2)	66.9 (11.8)	66.5 (11.3)	71.3 (12.4)	69.2 (12.2)
Age ≥75 y, %	41.6	29.5	26.4	46.6	38.0
Men, %	56.3	59.2	68.7	48.7	57.2
Medicare advantage, %*	74.7	64.3	61.2	79.3	71.6
United States region, %
South	44.1	51.4	47.9	45.0	46.3
Midwest	30.8	27.7	31.5	28.7	29.9
Northeast	16.7	14.7	12.5	18.8	16.1
West/other	8.3	6.1	8.0	7.4	7.7
Atherosclerotic cardiovascular disease characteristics, %
ACS hospital past 12 mo †	8.6	8.6	5.9	10.6	8.6
History of CHD †	37.8	54.7	39.9	44.9	42.7
CABG past 12 mo ‡	1.2	1.5	1.4	1.3	1.3
History of CABG §	10.1	14.1	10.2	12.0	11.2
PCI past 12 mo ‡	4.8	7.7	6.7	4.9	5.7
History of PCI §	12.2	19.6	15.7	13.4	14.4
IS hospital past 12 mo †	3.3	2.3	1.4	4.2	3.0
History of IS	8.2	4.6	3.6	9.7	7.1
ICBVD without stroke	7.4	8.0	7.4	7.8	7.6
PAD (lower extremities)	3.6	2.8	2.6	3.9	3.4
PAD (aorta)	3.4	2.9	2.7	3.6	3.2
PAD (other, unspecified) ‖	12.9	11.4	9.8	14.5	12.5
Other comorbidities, %
Atrial fibrillation/flutter	23.7	18.7	15.0	27.6	22.2
Hypertension	83.8	86.3	81.5	86.8	84.5
Heart failure	39.5	26.4	24.0	44.3	35.7
Renal disease stage III	14.1	10.9	9.4	16.0	13.2
Renal disease stages IV–V ¶	7.5	4.6	3.7	8.9	6.7
COPD	26.4	22.9	17.9	30.9	25.4
Moderate/severe liver disease	3.4	3.0	2.1	4.1	3.3
Diabetes	41.8	42.9	40.2	43.6	42.1
Diabetes on insulin	15.8	14.9	13.7	17.0	15.6
Treatment status before index ACS (point‐in‐time), % #
Lipid‐lowering therapy**	40.2	48.3	41.4	43.5	42.6
ß‐Blockers	38.3	46.0	36.5	43.5	40.6
ACEi/ARBs	40.8	44.7	40.5	42.9	41.9
P2Y12 inhibitors ††	14.4	21.8	16.0	16.9	16.5
Treatment status at 1 y after discharge (point‐in‐time), % ‡‡
Lipid‐lowering therapy**	63.1	60.1	73.3	52.1	62.1
ß‐Blockers	64.2	57.5	69.6	55.2	62.0
ACEi/ARBs	49.4	46.6	53.2	44.3	48.5
P2Y12 inhibitors ††	42.0	35.5	61.2	20.8	39.9
Treatment status at 1 y after discharge (cumulative), % ‡‡
Lipid‐lowering therapy**	79.5	75.6	90.2	67.5	78.2
ß‐Blockers	82.3	74.3	88.2	72.0	79.7
ACEi/ARBs	65.7	60.1	69.0	59.3	63.9
P2Y12 inhibitors ††	55.6	47.8	78.9	30.0	53.0
Procedures at index ACS or 30 d after discharge
CABG	9.9	18.7	29.4	0.0	12.5
PCI	33.2	27.7	74.5	0.0	31.6
TRS 2oP categories
0–1	17.9	22.0	25.0	14.8	19.1
2–3	47.2	52.0	52.7	45.7	48.6
4+	34.8	26.0	22.4	39.5	32.3

Point‐in‐time assessment indicates medication on hand at 1‐year follow‐up (see Figure S2), and cumulative assessment indicates medication on hand at any time during the 1‐year follow‐up. ACEi indicates angiotensin‐converting enzyme inhibitor; ACS, acute coronary syndrome; ARBs, angiotensin receptor blockers; CABG, coronary artery bypass grafting; CHD, coronary heart disease; COPD, chronic obstructive pulmonary disease; ICBVD, ischemic cerebrovascular disease; IS, ischemic stroke; PAD, peripheral arterial disease; PCI, percutaneous coronary intervention; and TRS 2oP, Thrombolysis In Myocardial Infarction Risk Score for Secondary Prevention.

Represents health plan offered by a private company that contracts with Medicare to provide patients with hospital and medical insurance benefits.

Measured in the 1‐year period before the start of the index hospitalization.

Measured before the start of the index hospitalization using procedure codes only.

Measured 1 year before admission of index hospitalization, during index hospitalization, or in the 30 days after hospitalization, using both diagnosis and procedure codes.

Includes any type of disease that is noncoronary, noncerebrovascular, or does not involve the lower extremities or aorta.

Includes dialysis or renal transplant.

Medication on hand at index ACS (see Figure S2).

Includes statins, ezetimibe, and proprotein convertase subtilisin/kexin type 9 inhibitors.

Includes clopidogrel, ticagrelor, and prasugrel.

Based on 132 489 patients at 1 year follow‐up (myocardial infarction=89 043; unstable angina=43 446; ACS with revascularization=62 508; ACS without revascularization=69 981).

Demographic and Clinical Characteristics of Subgroups

In Table 1, the subgroup presenting with MI was older, had more comorbidities, but was less likely to have had preexisting coronary heart disease, prior coronary revascularization, or treatment with standard cardiac medications, compared with the subgroup presenting with UA. In both subgroups, 8.6% were hospitalized for ACS within the previous 12 months. The subgroup treated at index ACS with revascularization was younger, more likely to be men, and was less likely to have comorbidities (including prior coronary heart disease), compared with the subgroup not treated with revascularization. Prior ACS within 12 months was lower in the revascularization subgroup, 5.9% versus 10.6%, respectively. Characteristics of the study population stratified by sex are summarized in Table S4.

Event Rates and Instantaneous Risk Over Time

The 5‐year rate for the primary end point representing first occurrence of nonfatal MI, nonfatal ischemic stroke, or cardiovascular death was 33.4% (95% CI, 33.1%–33.7%; P<0.001) (Figure 2) and for the secondary end point representing first occurrence of nonfatal MI, nonfatal ischemic stroke, cardiovascular death, UA hospitalization, or elective revascularization was 41.8% (95% CI, 41.4%–42.1%; P<0.001) (Figure S3). The rates were higher for the individual cardiovascular end points representing nonfatal MI, nonfatal ischemic stroke, and cardiovascular death as compared with UA hospitalization and elective revascularization. The instantaneous risk for the primary end point was highest immediately after discharge at 40.9%. This means that if this rate were to be continued, ≈40.9% would be expected to have had an event at 1 year. The risk gradually declined from discharge to an instantaneous risk of ≈6.7% at 1 year of follow‐up. Risk remained stable over time after 1 year (instantaneous risk of ≈6.4%). Table 2 illustrates the event rates for the primary and secondary composite end points and individual components at 5 years.

Figure 2

Event rates and instantaneous risk over time for the primary and individual end points.

Shown is the cumulative incidence of the primary end point (first occurrence of nonfatal MI, nonfatal ischemic stroke, or CV death) and individual end points (represented by components of the composite). The KM rates for the primary end point at 1, 3, and 5 years were 14.5% (95% CI, 14.3%–14.6%; P<0.001), 25.2% (95% CI, 25.0%–25.5%; P<0.001), and 33.4% (95% CI, 33.1%–33.7%; P<0.001), respectively. The instantaneous risks (annualized) for the primary end point at 0, 1, and 3 years were 40.9%, 6.7%, and 6.4%, respectively. Below the graphs capturing the cumulative incidence of the primary and individual end points are data on the number of patients at risk of experiencing the corresponding end point. CV indicates cardiovascular; IS, ischemic stroke; KM, Kaplan‐Meier; and MI, myocardial infarction.

Table 2

Event Rates for the Primary and Secondary Composite End Points and Individual Components at 5 Years

	No. at risk at 5 y	Adjusted cumulative count of events at 5 y*	Event rate at 5 y, % (95% CI)
Primary composite end point	21 193	79 923	33.4 (33.1–33.7)
Nonfatal MI †	21 193	30 139	12.6 (12.4–12.8)
Nonfatal IS †	21 193	22 462	9.4 (9.2–9.6)
Cardiovascular death †	21 193	27 323	11.4 (11.2–11.6)
Secondary composite end point	18 604	99 927	41.8 (41.1–42.1)
Nonfatal MI ‡	18 604	27 061	11.3 (11.1–11.5)
Nonfatal IS ‡	18 604	20 827	8.7 (8.5–8.9)
Cardiovascular death ‡	18 604	25 518	10.7 (10.5–10.9)
Nonfatal UA requiring hospitalization ‡	18 604	10 310	4.3 (4.2–4.4)
Nonfatal elective coronary revascularization ‡	18 604	16 211	6.8 (6.6–6.9)
Individual end points
Nonfatal MI §	22 596	34 594	14.5 (14.2–14.7)
Nonfatal IS §	23 030	27 436	11.5 (11.2–11.7)
Cardiovascular death §	24 648	37 888	15.8 (15.6–16.1)
Nonfatal UA requiring hospitalization §	23 089	16 251	6.8 (6.6–7.0)
Nonfatal elective coronary revascularization §	22 545	21 234	8.9 (8.7–9.1)

IS indicates ischemic stroke; MI, myocardial infarction; and UA, unstable angina.

Estimated by multiplying the 5‐year event rate by the initial cohort size. Counts of events for individual components of the composite may not add up to the count of the composite because of rounding.

Represent events contributing to the primary composite end point. Estimates are based on competing risk analysis.

Represent events contributing to the secondary composite end point. Estimates are based on competing risk analysis.

Represent first events regardless of whether they contribute to the primary or the secondary composite end point. Estimates are based on Kaplan‐Meier analysis.

Next, 4 index ACS subgroups were analyzed (baseline characteristics in Table S5): (1) MI treated with revascularization, (2) MI not treated with revascularization, (3) UA treated with revascularization, and (4) UA not treated with revascularization. Among the subgroups, the rates of the primary and secondary end points were highest for the MI without revascularization subgroup (Figure 3 and Figure S4). Rates of nonfatal MI and UA hospitalization were highest among patients who had already presented with an MI or UA, respectively. Rates for elective revascularization were higher for the subgroup of patients who had already had a revascularization. Finally, higher rates of the primary and secondary composite end points, as well as their individual components, were associated with more points on the TRS 2oP (Figure 4 and Figure S5).

Figure 3

Event rates for the primary and individual end points by type of index acute coronary syndrome and use of revascularization.

Shown is the cumulative incidence of the primary end point (first occurrence of nonfatal MI, nonfatal ischemic stroke, or CV death) and individual end points (represented by components of the composite) by index ACS (MI/UA and revascularization status). Below each graph are data on the number of patients in each index ACS subgroup who are at risk of experiencing the corresponding end point. ACS indicates acute coronary syndrome; CV, cardiovascular; IS, ischemic stroke; MI, myocardial infarction; Revasc, revascularization; and UA, unstable angina.

Figure 4

Event rates for the primary and individual end points by categories of TIMI risk score for secondary prevention.

Shown is the cumulative incidence of the primary end point (first occurrence of nonfatal MI, nonfatal ischemic stroke, or CV death) and individual end points (represented by components of the composite) by TRS 2oP categories. Below each graph are data on the number of patients in each TRS 2oP category who are at risk of experiencing the corresponding end point. CV indicates cardiovascular; IS, ischemic stroke; MI, myocardial infarction; TIMI, Thrombolysis In Myocardial Infarction; and TRS 2oP, Thrombolysis in Myocardial Infarction Risk Score for Secondary Prevention.

The current study population was older, more likely to be women, and less likely to be revascularized for an ACS as compared with the RCT populations (Table S6). , , , , , When the current study subgroup with revascularization at index ACS was compared with RCT populations, the baseline characteristics and the 1‐year event rates for the primary end point were much more similar (Table S6).

Risk Factors Associated With the Primary and Individual End Points

The adjusted associations between baseline characteristics and the primary composite end point are shown in Table 3. Age ≥65 years, heart failure, renal disease stages IV to V, ACS hospitalization during the past 12 months, history of ischemic stroke, and an index MI without revascularization were most strongly associated with risk of experiencing the primary end point (HR, >1.5; P<0.001). Risks of the individual end points of nonfatal MI and nonfatal ischemic stroke were highly associated with prior events of these types. Key factors most strongly associated with the risk of nonfatal MI were an ACS within 12 months of the index hospitalization or an MI as the index ACS. The strongest factor associated with risk of nonfatal ischemic stroke was a prior ischemic stroke. The strongest associations with risk of cardiovascular death were age ≥65 years and heart failure. Although statin treatment at baseline was only 41.9% (Table S7), it was associated with lower risk of the primary composite end point and its individual components. Evaluating the individual end points of the secondary composite end point, hospitalization for UA was most strongly associated with an ACS within 12 months of the index hospitalization. The factor most strongly associated with risk of elective revascularization was revascularization for either an index UA or MI (Table S8).

Table 3

Patient Characteristics Associated With the Primary and Individual End Points: Cox Proportional Hazard Model

	Primary end point*	Nonfatal myocardial infarction †	Nonfatal ischemic stroke ‡	Cardiovascular death §
	Hazard ratio (95% CI)	Hazard ratio (95% CI)	Hazard ratio (95% CI)	Hazard ratio (95% CI)
Index ACS type and revascularization status
MI with revascularization	1.16 (1.13–1.20)	2.08 (1.97–2.19)	0.74 (0.70–0.79)	0.85 (0.81–0.89)
MI without revascularization	1.81 (1.76–1.86)	2.46 (2.34–2.59)	1.15 (1.10–1.21)	1.86 (1.78–1.94)
UA with revascularization	0.69 (0.66–0.72)	0.85 (0.79–0.92)	0.65 (0.61–0.70)	0.59 (0.55–0.63)
UA without revascularization	Reference	Reference	Reference	Reference
Demographics
Age, y
20–44	0.68 (0.62–0.75)	0.79 (0.70–0.89)	0.61 (0.51–0.74)	0.52 (0.42–0.64)
45–64	Reference	Reference	Reference	Reference
≥65	1.62 (1.58–1.66)	1.28 (1.23–1.32)	1.47 (1.41–1.55)	2.56 (2.44–2.69)
Men	1.00 (0.98–1.02)	1.00 (0.97–1.03)	0.87 (0.84–0.90)	1.09 (1.06–1.12)
Cardiovascular comorbidities and risk factors
Atrial fibrillation/flutter	1.21 (1.18–1.23)	…	1.23 (1.19–1.28)	1.40 (1.36–1.44)
Hypertension	1.21 (1.17–1.25)	1.19 (1.14–1.26)	1.45 (1.36–1.56)	1.08 (1.03–1.14)
Heart failure	1.74 (1.70–1.77)	1.49 (1.44–1.54)	1.26 (1.21–1.31)	2.51 (2.43–2.60)
Renal disease
Stage III	1.25 (1.22–1.29)	1.22 (1.17–1.27)	1.19 (1.14–1.25)	1.33 (1.28–1.38)
Stage IV	1.61 (1.54–1.68)	1.51 (1.40–1.62)	1.36 (1.24–1.49)	1.87 (1.76–1.98)
Stage V	1.58 (1.52–1.64)	1.58 (1.49–1.68)	1.63 (1.52–1.76)	1.59 (1.50–1.68)
Stage ≤II	Reference	Reference	Reference	Reference
COPD	1.20 (1.17–1.22)	1.23 (1.19–1.28)	1.16 (1.11–1.20)	1.20 (1.16–1.24)
Moderate/severe liver disease	…	1.13 (1.05–1.22)	…	0.87 (0.80–0.95)
Diabetes
Not receiving insulin	1.11 (1.09–1.14)	1.19 (1.15–1.23)	1.28 (1.23–1.33)	0.96 (0.92–0.99)
Receiving insulin	1.33 (1.29–1.36)	1.49 (1.44–1.55)	1.64 (1.57–1.72)	1.04 (1.00–1.09)
No diabetes	reference	Reference	Reference	Reference
Tobacco use	0.91 (0.89–0.93)	1.07 (1.04–1.11)	…	0.70 (0.68–0.72)
Baseline atherosclerotic cardiovascular disease
History of CHD
ACS hospitalization past 12 mo	1.58 (1.53–1.63)	2.33 (2.23–2.44)	1.10 (1.03–1.17)	1.31 (1.25–1.37)
No ACS hospitalization past 12 mo	1.24 (1.21–1.26)	1.34 (1.29–1.38)	1.12 (1.07–1.16)	1.23 (1.19–1.27)
No history of CHD	Reference	Reference	Reference	Reference
History of ICBVD
IS hospitalization past 12 mo	1.81 (1.74–1.89)	1.05 (0.98–1.13)	4.70 (4.43–4.99)	1.45 (1.36–1.54)
IS without hospitalization past 12 mo	2.00 (1.94–2.07)	0.89 (0.83–0.96)	4.75 (4.52–4.99)	1.78 (1.69–1.87)
ICBVD without stroke	1.11 (1.07–1.15)	1.01 (0.95–1.07)	1.44 (1.35–1.55)	1.07 (1.01–1.13)
No history of ICBVD	Reference	Reference	Reference	Reference
PAD, lower extremities	1.12 (1.07–1.17)	1.15 (1.08–1.23)	…	1.16 (1.08–1.24)
PAD, aorta	1.11 (1.06–1.16)	1.09 (1.01–1.17)	1.16 (1.06–1.27)	1.13 (1.05–1.21)
Concurrent cardiovascular medications
Any statin	0.86 (0.84–0.87)	0.90 (0.87–0.93)	0.89 (0.86–0.92)	0.79 (0.77–0.81)
β‐Blockers	…	…	…	1.05 (1.01–1.08)
ACEi/ARB	0.94 (0.92–0.96)	0.96 (0.93–0.99)	0.94 (0.91–0.98)	0.92 (0.89–0.95)
Antiplatelets	1.16 (1.14–1.19)	1.20 (1.16–1.25)	1.18 (1.13–1.24)	1.11 (1.07–1.15)

Ellipses (…) indicate the covariate was determined to be nonsignificant in the model for that end point. ACEi indicates angiotensin‐converting enzyme inhibitor; ACS, acute coronary syndrome; ARB, angiotensin II receptor blocker; CHD, coronary heart disease; COPD, chronic obstructive pulmonary disease; ICBVD, ischemic cerebrovascular disease; ICD‐9/ICD‐10, International Classification of Diseases, Ninth Revision and Tenth Revision; IS, ischemic stroke; MI, myocardial infarction; NDI, National Death Index; PAD, peripheral arterial disease; and UA, unstable angina.

A major cardiovascular event for the primary end point is the NDI‐based date of cardiovascular death or the last date of an inpatient encounter with at least 1 ICD‐9/ICD‐10 code diagnosis for acute myocardial infarction or ischemic cerebrovascular disease with stroke.

A nonfatal myocardial infarction is defined as the last date of an inpatient encounter with at least 1 ICD‐9/ICD‐10 diagnosis for acute myocardial infarction.

A nonfatal ischemic stroke is defined as the last date of an inpatient encounter with at least 1 ICD‐9/ICD‐10 diagnosis for ischemic cerebrovascular disease with stroke.

A cardiovascular‐related death is defined as the NDI‐based date of cardiovascular death.

Discussion

We used several approaches to create a large, representative, well‐characterized cohort of patients hospitalized with ACS to better understand recurrent risk and risk predictors in a US clinical practice population. According to the 2018 ACC/AHA guidelines, ≈97% of our cohort was classified as very high risk. The 2020 European Society of Cardiology Guidelines for the Management of ACS in Patients Presenting Without Persistent ST‐Segment Elevation consider all patients with ACS as very high‐risk. Consistent with these classifications, we found the 5‐year event rate for the primary composite representing first occurrence of nonfatal MI, nonfatal ischemic stroke, or cardiovascular death was 33.4% (95% CI, 33.1%–33.7%; P<0.001).

The risk of experiencing the primary end point (Figure 2) was ≈6‐fold higher immediately following discharge for the index ACS, as compared with the period ≥1 year after hospital discharge. The extremely high risk immediately after ACS hospitalization underscores the usefulness of risk mitigation strategies when started during, or immediately after, ACS hospitalization. , , , , , Risk during the first year after an ACS was much more pronounced for the nonfatal MI and cardiovascular death end points as compared with the other individual end points. This observation further supports the high expected clinical benefit when guideline‐based treatments are initiated during hospitalization for an ACS. The risk of the primary end point was consistently 6.4% after 1 year, illustrating a high residual risk during chronic follow‐up.

In general, baseline atherothrombotic disease, index ACS presentation, and use of revascularization were strongly associated with future adverse outcomes (eg, recurrent MI in those with index MI, recurrent UA in those with index UA, recurrent ischemic stroke in those with prior stroke, repeat revascularization in those revascularized). Rates of ischemic stroke and cardiovascular death were higher in those not revascularized for index ACS. There are many potential explanations suggested by the data including greater burden of noncardiovascular comorbidities, higher rates of myocardial injury and type 2 MI for which the benefit of revascularization may not be indicated or is less clear, perception of less favorable risk:benefit of revascularization, and greater reluctance of the patient to proceed with revascularization (eg, because of advanced age, renal failure).

Our subgroup analyses yielded further insights. With the increasing use of highly sensitive cardiac biomarkers, separately analyzing MI versus UA subgroups in health care databases will help better separate those who may have an atherothrombotic event versus those who do not. This was apparent in our findings, because the MI subgroups had much higher event rates for the primary composite end point and individual components of the composite when compared with the UA subgroups. In terms of revascularization at index ACS, especially in a US population, our data suggest this criterion may enhance the representation of patients with true acute coronary atherothrombosis (eg, as opposed to type 2 MI). For some patients with ACS presentations, however, a strategy of no angiography nor PCI is also influenced by patients’ age and comorbidities. This likely explains the findings in the subgroup without revascularization, which was relatively older, with a greater number of comorbidities, and had a much higher rate for the primary composite end point and individual components of the composite. A much lower proportion of women received revascularization at index ACS (Table S4). In our study population, women were both older and somewhat more comorbid. Older women with ACS have a higher prevalence of age‐related functional impairment, which could also have contributed to a lower rate of revascularization. Other studies have indicated a lower likelihood of revascularization in women even after adjusting for other factors, which could also help explain this observation. , There are, however, no differences in the expected benefit of treatment with revascularization by sex based on evidence from meta‐analysis. In regard to the subgroup with revascularization, a comparison of its baseline characteristics with RCTs enrolling patients with ACS revealed these populations to be more similar as compared with the subpopulation without revascularization, which supports the hypothesis that the subpopulation with revascularization more closely represents those experiencing a true acute atherothrombotic event (Table S6).

Application of TRS 2oP resulted in a substantial stratification of the risk for the primary end point and individual components of the composite (Figure 4). TRS 2oP was originally developed from a population representing a stable post‐MI profile in the TRA 2oP–TIMI 50 (Thrombin Receptor Antagonist in Secondary Prevention of Atherothrombotic Ischemic Events–Thrombolysis In Myocardial Infarction 50) trial. The TRS 2oP was demonstrated to effectively stratify the risk of recurrent cardiovascular events in the same trial, as well in IMPROVE‐IT (Improved Reduction of Outcomes: Vytorin Efficacy International Trial), an RCT representing a recent ACS population. , Our study contributes additional evidence on risk stratification via TRS 2oP in a post‐ACS population representing usual clinical practice. For the primary end point, the risk at 5 years for the subgroup with the highest TRS 2oP category (≥4) was 56.7% (95% CI, 55.9%–57.4%; P<0.001) as compared with 12.9% (95% CI, 12.4%–13.4%; P<0.001) for the subgroup representing the lowest TRS 2oP category (0–1), over a 4‐fold difference. Because the TRS 2oP is based on measures commonly available in routine clinical practice (Table S9), our findings support the potential application of the TRS 2oP for informed decision making on clinical management of patients with ACS, identification of patients at highest risk, and development of future guidelines. TRS 2oP captured the main risk factors identified in the multivariable Cox model (Table 3) except for the history of coronary heart disease and a prior ACS event, which are not represented in TRS 2oP, because it was originally developed in a population with a history of MI. An implication of the potential improvement in TRS 2oP for risk stratification in a post‐ACS population could thus be inclusion of the history of coronary heart disease and a prior ACS event.

Our study indicated absence of treatment with revascularization at index ACS is an independent predictor of risk after adjustment of other patient characteristics and comorbidities, including those represented in TRS 2oP. Evidence suggests improved outcomes with routine invasive as compared with medical therapy alone in a population with ACS. This observation combined with our study findings indicate that though revascularization may improve patient outcomes, it could be a less viable option in clinical practice for patients presenting with multiple comorbidities and/or advanced age. In Figure 5 and Figure S6, we presented an analysis that combines TRS 2oP categories and revascularization status, which offers further improvement on a framework for identifying those at a higher risk and indicates that individuals with TRS 2oP ≥4 without revascularization have a 5‐fold higher risk as compared with those with TRS 2oP ≤1 and revascularization.

Figure 5

Event rates for the primary and individual end points by categories of TIMI risk score for secondary prevention and use of revascularization.

Shown is the cumulative incidence of the primary end point (first occurrence of nonfatal MI, nonfatal ischemic stroke, or CV death) and individual end points (represented by components of the composite) by TRS 2oP categories and use of revascularization. Below each graph are data on the number of patients in each TRS 2oP category with or without revascularization who are at risk of experiencing the corresponding end point. CV indicates cardiovascular; IS, ischemic stroke; MI, myocardial infarction; Revasc, revascularization; TIMI, Thrombolysis In Myocardial Infarction; and TRS 2oP, Thrombolysis In Myocardial Infarction Risk Score for Secondary Prevention.

The proportion of patients receiving PCI at index ACS was only 33.2% and 27.7% for the MI and UA subgroups, respectively, which is lower than the rates typically reported in RCTs (Table S6). The 2021 ACC/AHA Guidelines for Coronary Artery Revascularization recommends the decision to treat with revascularization be based on clinical indications, regardless of sex, race, or ethnicity. In our study, the following factors were more prevalent in the subgroup with ACS without revascularization (Table 1): age ≥75 years, women, atrial fibrillation, heart failure, renal disease, chronic obstructive pulmonary disease, and diabetes. In contrast, prior ASCVD comorbidities generally were numerically higher in the ACS with revascularization subgroup, which indicates that non‐ASCVD comorbidities and patient suitability for PCI procedure likely contributed to the lower rates of PCI. Other criteria mentioned in the 2021 ACC/AHA guidelines for consideration of revascularization are disease complexity, technical feasibility, and patient preference, all of which could also have influenced rates of PCI, especially in those with comorbidities and advanced age. In addition, in typical clinical practice, chest pain caused by acute myocardial injury (eg, pulmonary embolism) or chronic myocardial injury (eg, renal failure) can be coded as myocardial infarction, because of symptoms and elevated biomarkers. Because such patients with myocardial injury (in contrast to acute MI) and type 2 MI less frequently meet an indication for revascularization, the category of ACS without revascularization was enriched with these patients. Of note, this subgroup also had greater rates of noncardiovascular comorbidities, which may have altered the risk:benefit assessment, thus contributing to the lower rate of revascularization. An important avenue for additional research on this topic could be investigating disparities in treatment with revascularization, representing factors such as sex, race, and ethnicity.

The goal of guideline‐based therapies in a population with ASCVD or an ACS event is to reduce the risk of recurrent cardiovascular events. , The absolute risk reduction with treatment depends on the patient’s baseline risk representing characteristics before treatment (ie, presentation at index ACS in the current study) and the choice of treatment, meaning patients at higher predicted risk at index ACS are expected to derive greater absolute benefit from a given treatment. In our study, the use of guideline‐based therapies was relatively low at baseline, which could be because of low prescription rates, nonadherence, or that many patients experiencing an ACS event may not have prior clinical ASCVD or other treatment indications. The use of guideline‐based therapies increased during 1 year after index ACS (Tables S10 and S11). The use of lipid‐lowering therapies, for example, was 41.4% and 43.5%, before index ACS in subgroups with and without revascularization, respectively (based on a point‐in‐time evaluation), increasing to 73.3% and 52.1%, respectively, at 1 year (the same measure being 90.2% and 67.5%, respectively, based on cumulative evaluation) (Table 1). The pattern was similar for P2Y12 inhibitors, where before the index ACS, ≈16.0% and 16.9% were receiving a P2Y12 inhibitor in subgroups with and without revascularization, respectively. At 1 year after discharge, the rate increased to ≈61.2% and 20.8%, respectively (78.9% and 30.0%, respectively, based on cumulative evaluation). The potential explanations noted previously about the differences in the subgroups with and without revascularization, likely also explains the marked difference in the rates of treatment with lipid‐lowering therapies and P2Y12 inhibitors in these subgroups, with the subgroup without revascularization heavily influencing the low overall use of guideline‐recommended medical therapy at 1 year after discharge. Because the guidelines strongly endorse medical therapies such as lipid‐lowering therapies and P2Y12 inhibitors after ACS (the latter at least for 1 year in most patients), it is apparent that the guideline‐recommended use of these therapies can be improved. Thus, a broad implication from our study on improving patient outcomes in clinical practice is to identify patients at higher risk (eg, based on combination of TRS 2oP and revascularization status) and permit timely initiation, appropriate intensification, and continuation of guideline‐based therapies.

Key strengths of our study include a large (N=239 234) and contemporary (2005–2018) population, with ACS hospitalization representing a usual clinical practice setting, from the time of hospital discharge through long‐term follow‐up (N=24 648 patients with ≥5 years follow‐up). Another key strength of this study as compared with prior investigations was that determination of mortality status was supplemented with National Death Index data to provide a more reliable ascertainment of cardiovascular mortality. The ACS population characterization, subgroup analyses, event rate patterns, and risk predictor analyses may prove useful to multiple stakeholders including payers, regulatory authorities, clinicians, and clinical guidelines committees for the purpose of identifying those at highest risk of adverse events and who would derive the highest clinical benefit from evidence‐based therapies. In addition, these data may help inform academic and governmental organizations that track data on heart disease and stroke.

Limitations

The Optum Research Database represents individuals enrolled in commercial and Medicare Advantage health plans in the United States. Thus, the analysis may not be representative of the noninsured population, those covered under a public insurance program other than Medicare Advantage (eg, Medicaid and other Medicare plans), or international populations. Evidence suggests that patients with ACS without insurance or with Medicaid are less likely to receive evidence‐based therapies and have worse outcomes compared with those with private insurance. , , Claims databases capture medications received via coverage under a health plan. As such, medications obtained over the counter (eg, aspirin) or those that are paid entirely out of pocket (eg, statin medication paid at list price) will not be captured. Studies investigating concordance between claims databases and patient self‐report or medication inventory have reported an agreement of ≈87.5% for statins, indicating a possibility for a slight underestimation of treatment with statins in our study. We used claims‐based discharge diagnoses from inpatient hospitalizations for defining nonfatal cardiovascular events. Studies evaluating validation of this methodology as adjudicated by investigators have concluded positive predictive values for MI within the range of 88.4% to 96.9%. , , , , Similar studies have reported the range of positive predictive values for ischemic stroke as 80.4% to 91.1%. , , , , A recent study that evaluated this methodology for trial end points adjudicated by a clinical events committee using standardized end point definitions concluded a positive predictive value of 85.7% for MI but only 47.1% for stroke (authors recommended caution in interpretation because of the low number of stroke events). In addition, unlike clinical trials, which typically classify patients into ST versus non–ST‐segment–elevation MI by electrocardiographic review, this cannot be reliably done using claims data. Our study mainly focused on the demographic, clinical, and treatment‐related factors as predictors of recurrent cardiovascular event risk in a population with ACS. Though biomarkers including high‐sensitivity troponin T, N‐terminal pro‐B‐type natriuretic peptide, growth‐differentiation factor‐15, and high‐sensitivity C‐reactive protein have shown to offer useful prognostication on top of clinical risk predictors (eg, TRS 2oP), these biomarkers are not typically collected in a manner that enables their incorporation in our analysis. Furthermore, during the identification period, 2005 to 2018, the United States moved from MI definitions based on less‐sensitive biomarkers to highly sensitive biomarkers (high‐sensitivity troponin), which likely affected the classification of MI versus UA in clinical practice over this period. Although tobacco use was captured for the study, assessment was limited to instances where it is recorded in relation to patient care and does not adequately measure frequency and amount of tobacco use. Our reported measures of tobacco use are thus likely to be underestimated. Finally, the RCTs used for comparing our study population with trial populations represented global trials, whereas our study population represents only patients in the United States.

Conclusions

In a large, representative, and well‐characterized US population followed after discharge from ACS hospitalization, the 5‐year event rate for the primary composite end point representing nonfatal MI, nonfatal ischemic stroke, or cardiovascular death was 33.4% (95% CI, 33.1%–33.7%; P<0.001). The risk of experiencing the primary end point was ≈6‐fold higher immediately following the index ACS, as compared with ≥1 year after hospital discharge. For health care stakeholders interested in understanding the applicability of RCT data to clinical practice populations, the subgroup and risk predictor analysis identified a methodology to better identify a trial‐like population and characteristics that predict specific clinical end points. In addition to these new insights, the TRS 2oP demonstrated use in risk stratification when applied to a large health care administrative claims database, potentially expanding its use to investigators working with similar databases.

Sources of Funding

This study was funded by Sanofi.

Disclosures

Dr Steen reports receiving consultant fees from Sanofi and Regeneron. Dr Khan is an employee and stockholder of Sanofi. K. Andrade is an employee of Optum. A. Koumas does not have any disclosures to report. Dr Giugliano reports receiving consultant fees from Amarin, American College of Cardiology, Amgen, Astra Zeneca, Boehringer Ingelheim Pharmaceuticals, Inc., Bristol‐Myers‐Squibb Company, CryoLife, Inc., CVS Caremark, Daiichi Sankyo, Dr Reddy’s Laboratories, Eli Lilly and Company, Esperion, Gilead Sciences, Inc., GlaxoSmithKline, Janssen Pharmaceuticals, Inc., Lexicon, Merck, Pfizer, St. Luke’s Hospital System, Kansas City, SAJA Pharmaceuticals, Samsung, Servier; research grants from Amgen, Anthos Therapeutics, and Daiichi‐Sankyo; and honoraria for CME lectures from Amgen, Daiichi Sankyo, Merck, SAJA Pharmaceuticals, and Servier.

Data S1–S2

Tables S1–S11

Figures S1–S6

Reference 44

Click here for additional data file.