Association Between Claims‐Defined Frailty and Outcomes Following 30 Versus 12 Months of Dual Antiplatelet Therapy After Percutaneous Coronary Intervention: Findings From the EXTEND‐DAPT Study

Abstract

Background

Frailty is rarely assessed in clinical trials of patients who receive dual antiplatelet therapy (DAPT) after percutaneous coronary intervention. This study investigated whether frailty defined using claims data is associated with outcomes following percutaneous coronary intervention, and if there is a differential association in patients receiving standard versus extended duration DAPT.

Methods and Results

Patients ≥65 years of age in the DAPT (Dual Antiplatelet Therapy) Study, a randomized trial comparing 30 versus 12 months of DAPT following percutaneous coronary intervention, had data linked to Medicare claims (n=1326), and a previously validated claims‐based index was used to define frailty. Net adverse clinical events, a composite of all‐cause mortality, myocardial infarction, stroke, and major bleeding, were compared between frail and nonfrail patients. Patients defined as frail using claims data (12.0% of the cohort) had higher incidence of net adverse clinical events (23.1%) compared with nonfrail patients (10.7%; P<0.001) at 18‐month follow‐up and increased risk after multivariable adjustment (adjusted hazard ratio [HR], 2.24 [95% CI, 1.38–3.63]). There were no differences in effects of extended duration DAPT on net adverse clinical events for frail (HR, 1.42 [95% CI, 0.73–2.75]) and nonfrail patients (HR, 1.18 [95% CI, 0.83–1.68]; interaction P=0.61), although analyses were underpowered. Bleeding was highest among frail patients who received extended duration DAPT.

Conclusions

Among older patients in the DAPT Study, claims‐defined frailty was associated with higher net adverse clinical events. Effects of extended duration DAPT were not different for frail patients, although comparisons were underpowered. Further investigation of how frailty influences ischemic and bleeding risks with DAPT are warranted.

Registration

URL: https://www.clinicaltrials.gov; Unique identifier: NCT00977938.

Nonstandard Abbreviations and Acronyms

CFI

Claims‐Based Frailty Indicator

DAPT

dual antiplatelet therapy

FAC

frailty‐associated characteristic

MACCE

major adverse cardiovascular and cerebrovascular events

NACE

net adverse clinical events

Clinical Perspective.

What Is New?

• Frailty is rarely evaluated in clinical trials of patients with coronary artery disease but can potentially be assessed using administrative claims when frailty data are not otherwise available.

• In the DAPT (Dual Antiplatelet Therapy) Study, a randomized trial comparing 30 versus 12 months of dual antiplatelet therapy after percutaneous coronary intervention, patients with claims‐defined frailty had increased risk of adverse events.

• Frailty did not impact effects of 30 versus 12 months of dual antiplatelet therapy in patients following percutaneous coronary intervention in this analysis, although comparisons were underpowered.

What Are the Clinical Implications?

• Further investigation of how frailty impacts ischemic and bleeding risks with antiplatelet therapy is warranted.

• This study supports supplemental use of administrative claims data in the context of cardiovascular clinical trials.

Frailty is a state of increased vulnerability to stressors and decreased physiologic reserve, ¹ , ² , ³ and is present in 1 out of 5 patients with coronary artery disease (CAD) who undergo percutaneous coronary intervention (PCI). ⁴ , ⁵ Frailty is strongly associated with adverse outcomes among patients with CAD, including up to a 4‐fold increased odds of all‐cause mortality. ³ , ⁶ Although frailty is widely acknowledged as clinically significant and is becoming more prevalent with the aging population, most clinical research studies including randomized trials do not measure or otherwise assess the impact of frailty on patient outcomes. ⁷ , ⁸ Additionally, there is no standardized method for assessing frailty, and physician assessment of frailty may not necessarily be associated with risk of adverse outcomes. ⁹ Consequently, the generalizability of research findings to clinical management of frail patients with CAD remains uncertain.

In particular, although prior studies have demonstrated that frailty is associated with increased short‐term risks of bleeding following PCI, ¹⁰ , ¹¹ the long‐term risks of adverse outcomes in this population remain unknown. In addition, there is sparse evidence on how frailty impacts the balance of ischemic and bleeding risk in patients with CAD on dual antiplatelet therapy (DAPT) following PCI. Although frailty is associated with age, and older patients generally have increased mortality associated with bleeding and may therefore benefit from shorter duration of DAPT, ¹² , ¹³ , ¹⁴ the impact of this regimen specifically among frail patients has not been clearly demonstrated.

One novel way to assess frailty when not otherwise evaluated is through use of administrative claims data, which can capture conditions associated with frailty through use of billing codes. ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ Prior observational studies have used this approach, ¹⁹ , ²⁰ , ²¹ , ²² , ²³ , ²⁴ which may have additional value in the context of clinical trials. To assess this further, we used clinical trial data from the DAPT (Dual Antiplatelet Therapy) Study linked to Medicare claims to explore how frailty, defined based on a previously validated claims‐based frailty index, is related to long‐term adverse outcomes following PCI. ¹⁷ , ²⁵ We also evaluated how claims‐defined frailty impacts treatment effects of 30 versus 12 months of DAPT following PCI.

METHODS

Overview of the EXTEND‐DAPT Study

This analysis was part of the EXTEND (Extending Trial‐Based Evaluations of Medical Therapies Using Novel Sources of Data) Study. An overview of the aims and methods of the EXTEND Study have been previously described. ²⁶ The EXTEND substudy, EXTEND‐DAPT, linked data from the DAPT Study to the American College of Cardiology's National Cardiovascular Data Registry (NCDR) CathPCI Registry and Medicare fee‐for‐service beneficiary claims.

The DAPT Study was a randomized, multicenter, placebo‐controlled clinical trial of patients who underwent PCI and received DAPT consisting of aspirin and a P2Y12 inhibitor for 1 year (ClinicalTrials.gov Identifier NCT00977938). ²⁵ At 1 year following PCI, patients without ischemic or bleeding events were randomized to either placebo (12 total months of DAPT) or continued thienopyridine for an additional 18 months (30 total months of DAPT). Patients were enrolled between August 13, 2009 and July 1, 2011. The Baim Institute for Clinical Research (formerly the Harvard Clinical Research Institute) conducted the trial, and the EXTEND Study has been approved by the institutional review board at Beth Israel Deaconess Medical Center with a waiver of informed consent. Data that support findings from this study are available from the corresponding author upon reasonable request.

Study Population

In this study we included all patients in the DAPT Study who were at least 65 years of age (the age of eligibility for Medicare), enrolled in fee‐for‐service Medicare insurance in the United States at the time of PCI, and could successfully be linked via the NCDR CathPCI Registry to the Centers for Medicare and Medicaid Services inpatient claims data. ²⁷ Deterministic algorithms based on age or date of birth, sex, PCI date, stent type, hospital discharge date, and hospital identifier were used to link DAPT Study data to the CathPCI Registry. These data were then linked to Medicare inpatient claims using direct identifiers available in the CathPCI Registry. Patients who could not be linked to the CathPCI Registry due to inexact matching variables or who were not found in Centers for Medicare and Medicaid Services fee‐for‐service claims data were excluded.

The DAPT Study randomized 11 648 patients, and of these, 3908 were in the United States and ≥65 years of age. Data from 2558 of these patients were successfully linked to the CathPCI Registry, and 1336 were ultimately linked to Medicare inpatient claims; characteristics between linked and unlinked patients were generally similar, as previously reported. ²⁸ Ten patients were excluded because they were not enrolled in Medicare at the time of index PCI, and therefore did not have at least 12 months of inpatient claims data available to assess for frailty before the start of randomization. The final study cohort therefore consisted of 1326 patients (Figure S1). The primary reasons for inability to link Medicare‐eligible patients were being insured by Medicare Advantage (n=692) or lacking sufficient information to be identified in both data sources (n=530). All baseline characteristics other than claims‐defined frailty were obtained from information collected in the DAPT Study.

Frailty Definitions

In primary analyses for this study we defined frailty based on the Johns Hopkins Claims‐Based Frailty Indicator (CFI), a previously validated index that uses 21 variables available in claims data with International Classification of Diseases, Ninth Revision, Clinical Modification (ICD‐9‐CM) diagnosis codes. ¹⁶ , ¹⁷ The Hopkins CFI is based on claims‐based variables that are predictive of the Fried definition of frailty, characterized by weight loss, exhaustion, weak grip strength, slow walk speed, and low physical activity. ²⁹ Frailty based on the Hopkins CFI has been significantly associated with adverse outcomes including incidence of falls, worsening mobility, acute hospitalization, and death. ¹⁷ , ¹⁸ , ¹⁹ , ²¹ We specifically defined frailty using a Hopkins CFI cutoff of >0.12, as suggested by the authors. ¹⁷

We performed a separate exploratory analysis to assess frailty based on a cumulative deficit approach. In this analysis, frailty was based on total number of frailty‐associated characteristics (FACs) based on the Hopkins CFI as well as the Claims‐Based Frailty Index developed by Kim et al, another validated tool that predicts frailty based on cumulative deficits using survey‐reported functional limitations. ¹⁵ This index includes>90 variables from claims data, and is similarly predictive of poor outcomes including recurrent falls, impaired mobility, health care use, and death. ¹⁵ , ¹⁹ , ²⁰ , ³⁰ ICD‐9‐CM and Healthcare Common Procedure Coding System codes were grouped based on categories previously used for these 2 claims‐based indices. We only included variables from the Claims‐Based Frailty Index by Kim et al that had a ≥0.01 increase in index value based on the relevant lasso regression model, as directly reported by the authors. ¹⁵ Incorporation of variables from both models resulted in 32 unique categories of claims‐based FACs, including neurologic and psychotic disorders, arthropathies, impaired mobility and falls, infections, and use of medical equipment (Table S1). Claims‐Based Frailty Index values by Kim et al were not assessed because we did not have access to outpatient claims, which are used in calculations of this index. Increasing frailty in this sensitivity analysis was based on having 0–3, 4–5, or ≥6 FACs, with ≥6 defined as frailty. These cutoffs were chosen to approximate the previously observed prevalence of frailty using the Fried definition in patients with coronary artery disease undergoing PCI. ⁵

Outcomes

Outcomes for this study included death, myocardial infarction (MI), stroke, major bleeding, and net adverse clinical events (NACE). NACE were defined as a composite of the individual end points of all‐cause death, MI, stroke, and major bleeding. We also assessed major adverse cardiovascular and cerebrovascular events (a composite of death, MI, and stroke), because this was the primary outcome of the DAPT Study. All outcomes were assessed based on adjudication by the DAPT Study Clinical Events Committee. The trial used prespecified definitions of MI and stroke. ²⁵ Major bleeding was defined as any adjudicated event that met criteria for either moderate or severe bleeding based on GUSTO (Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Arteries) classification, or type 3 or type 5 bleeding based on the Bleeding Academic Research Consortium (BARC) definitions. ³¹ , ³² To include minor bleeding events in our outcomes, we performed sensitivity analyses defining bleeding as BARC type 2, 3, or 5 bleeding. Cardiovascular and noncardiovascular death were evaluated as secondary end points. The study period for follow‐up was from time of randomization (12 months following PCI) to 18 months after randomization (30 months following PCI).

To determine the association of frailty with adverse clinical events, cumulative incidence and adjusted risks of outcomes were analyzed across the entire cohort. To assess whether duration of DAPT impacted outcomes based on frailty status, we performed an exploratory analysis comparing groups randomized to 30 versus 12 months of DAPT stratified by frailty.

Statistical Analysis

Baseline characteristics are presented as percentages for categorical variables and means with SDs for continuous variables. Univariate comparisons of baseline characteristics were performed using the Fisher exact test for categorical variables and t test for continuous variables. Cumulative incidence rates of outcomes were determined in nonparametric survival analyses using Kaplan‐Meier estimates. To compare survivor functions of 30 versus 12 months of DAPT within each group, log‐rank tests were performed using outcomes observed up to 18 months following randomization. Cox regressions were used to determine hazard ratio (HR) with 95% CI for each outcome. Patients were censored at the last day of follow‐up or the end of the study period. For individual nondeath outcomes, patients were also censored at time of death, and we accounted for competing risk of death using a subdistribution hazard model based on the Fine‐Gray method. ³³

To assess the association of frailty with adverse clinical events, cumulative incidence of outcomes was determined. Risks of clinical events were analyzed using Cox regression with adjustment for age, sex, total number of lifestyle‐related cardiovascular risk factors (diabetes, hypertension, smoking, or obesity), presence of an additional cardiovascular risk factor (any 1 of the following: prior PCI, prior coronary artery bypass graft, prior stroke or transient ischemic attack, prior MI, or MI at the time of index PCI), and DAPT study arm (30 or 12 months of DAPT). In sensitivity analyses using a cumulative deficit approach, patients with 0 to 3 FACs were used as the reference group, and adjusted analyses were repeated with FACs evaluated as a continuous variable.

To assess the impact of frailty on treatment effects of 30 versus 12 months of DAPT after PCI, outcomes of randomized groups were analyzed using an intention‐to‐treat framework with stratification by the Hopkins CFI. Cumulative incidence and risks of outcomes were assessed separately for frail and nonfrail groups, and tests for interaction were performed using Cox regression. Similar sensitivity analyses were performed using frailty defined based on having ≥6 FACs. Adjustment was not performed in these analyses, because patient characteristics were similar between randomized groups. All analyses were conducted in SAS version 9.4 (SAS Institute, Cary, NC) using a 2‐tailed P<0.05 to define significance.

RESULTS

Baseline Characteristics

The mean age (SD) of this cohort was 71.9 (SD, 5.5) years, and 32.9% were women. Using a Hopkins CFI >0.12, 159 (12.0%) patients were defined as frail, and 1167 (88.0%) were considered nonfrail. Baseline characteristics of patients stratified by frailty are shown in Table 1. Frail patients were older and more likely to be women and have prior history of stroke.

Table 1.

Baseline Clinical Characteristics Based on Hopkins CFI in the Dual Antiplatelet Therapy Study

Characteristic	Frail (CFI>0.12; N=159)	Nonfrail (CFI≤0.12; N=1167)	P value
Age, y, mean, SD	79.8, 5.5	70.8, 4.5	<0.001
Women	88 (55.3)	348 (29.8)	<0.001
Non‐White race	20 (12.7)	67 (5.8)	<0.001
Hispanic/Latino ethnicity	6 (3.8)	27 (2.3)	0.27
BMI, mean, SD	28.3, 5.0	29.8, 5.3	<0.001
Diabetes	57 (35.8)	384 (33.0)	0.47
Hypertension	142 (89.9)	985 (84.6)	0.08
Current or recent smoking	6 (3.8)	142 (12.3)	0.001
Stroke/TIA	21 (13.2)	51 (4.4)	<0.001
Congestive heart failure	25 (15.8)	60 (5.2)	<0.001
Peripheral arterial disease	24 (15.4)	111 (9.6)	0.03
History of major bleeding	1 (0.6)	12 (1.0)	0.63
Prior PCI	53 (33.5)	409 (35.3)	0.67
Prior CABG	21 (13.2)	198 (17.0)	0.23
Prior myocardial infarction	31 (20.1)	239 (21.0)	0.80
Atrial fibrillation	10 (6.3)	53 (4.6)	0.33
Index PCI indication
STEMI	2 (1.3)	51 (4.4)	0.06
NSTEMI	24 (15.1)	121 (10.4)	0.07
Unstable angina	27 (17.0)	174 (14.9)	0.49
Stable angina	65 (40.9)	531 (45.5)	0.27

BMI indicates body mass index; CABG, coronary artery bypass graft; CFI, Claims‐Based Frailty Indicator; NSTEMI, non–ST‐segment–elevation myocardial infarction; PCI, percutaneous coronary intervention; STEMI, ST‐segment–elevation myocardial infarction; and TIA, transient ischemic attack.

Adverse Outcomes Based on Claims‐Defined Frailty

After 18 months of follow‐up, patients with frailty based on the Hopkins CFI had higher cumulative incidence of NACE (23.1%) compared with patients without frailty (10.7%; log‐rank P<0.001; Figure 1). Patients with frailty also had higher incidence of major adverse cardiovascular and cerebrovascular events, all‐cause death, MI, and major bleeding (Table 2). In adjusted analyses, patients with frailty had higher risk of these outcomes compared with nonfrail patients (Table 3). Consistent with the primary analyses, results from adjusted analyses evaluating Hopkins CFI as a continuous variable showed that increasing CFI was associated with increased risk of most outcomes including NACE and major adverse cardiovascular and cerebrovascular events (Table S2). When minor bleeding was included in the bleeding end point, patients with frailty had higher incidence and adjusted risks of bleeding and NACE (Tables S3 and S4).

Figure 1. Cumulative incidence of net adverse events based on claims‐based frailty in the EXTEND‐DAPT study.

All outcomes were assessed from time of randomization (12 months following PCI) to 18 months post‐randomization (30 months following PCI) in the DAPT Study. NACE was a composite outcome inclusive of all‐cause mortality, MI, stroke, and major bleeding. CFI indicates Claims‐Based Frailty Indicator; DAPT, dual antiplatelet therapy; MI, myocardial infarction; NACE, net adverse clinical events; and PCI, percutaneous coronary intervention.

Table 2.

Cumulative Incidence of Adverse Outcomes Based on the Hopkins CFI

Outcome	Frail (CFI >0.12; N=159)	Non‐Frail (CFI ≤0.12; N=1167)	Log‐rank P value
NACE	23.1	10.7	<0.001
MACCE	14.1	5.7	<0.001
Death	6.5	2.0	<0.001
Cardiovascular	1.9	0.9	0.20
Noncardiovascular	4.6	1.2	0.001
MI	6.6	3.4	0.041
Stroke	3.3	1.2	0.029
Major bleeding	11.1	3.4	<0.001

All outcomes were assessed from time of randomization (12 months following percutaneous coronary intervention) to 18 months after randomization (30 months following percutaneous coronary intervention) in the Dual Antiplatelet Therapy Study. NACE was a composite outcome inclusive of all‐cause mortality, MI, stroke, and major bleeding, and MACCE was a composite of all‐cause mortality, MI, and stroke. CFI indicates Claims‐Based Frailty Indicator; MACCE, major adverse cardiovascular or cerebrovascular events; MI, myocardial infarction; and NACE, net adverse clinical events.

Table 3.

Adjusted Risks of Adverse Outcomes for Patients With Frailty Based on Hopkins CFI >0.12

Outcome	Adjusted HR (95% CI) (reference: CFI ≤0.12)	P value
NACE	2.24 (1.38–3.63)	0.001
MACCE	2.93 (1.54–5.57)	0.001
Death	3.05 (1.10–8.47)	0.032
Cardiovascular	1.73 (0.32–9.49)	0.529
Noncardiovascular	4.39 (1.22–15.75)	0.023
Myocardial infarction	2.60 (1.06–6.37)	0.038
Stroke	3.27 (0.81–13.14)	0.095
Major Bleeding	2.43 (1.11–5.31)	0.026

Outcomes were adjusted for age, sex, total number of lifestyle‐related cardiovascular risk factors (diabetes, hypertension, smoking, or obesity), presence of an additional cardiovascular risk factor (any 1 of the following: prior percutaneous coronary intervention, prior coronary artery bypass graft, prior stroke or transient ischemic attack, prior myocardial infarction, or myocardial infarction at the time of index percutaneous coronary intervention), and Dual Antiplatelet Therapy Study arm (30 or 12 months of dual antiplatelet therapy). CFI indicates Claims‐Based Frailty Indicator; HR, hazard ratio; MACCE, major adverse cardiovascular or cerebrovascular events; and NACE, net adverse clinical events.

Treatment Effects of Extended Duration DAPT Based on Claims‐Defined Frailty

Baseline characteristics of subgroups based on DAPT duration and frailty defined by the Hopkins CFI were similar (Table S5). In comparisons of patients by randomized treatment arm, the cumulative incidence of NACE after PCI was 11.5% with 30 months of DAPT and 9.9% with 12 months of DAPT among nonfrail patients (HR, 1.18 [95% CI, 0.83–1.68]; P=0.36). Among patients with frailty, the cumulative incidence of NACE was 26.7% with 30 months of DAPT and 19.5% with 12 months of DAPT (HR, 1.42 [95% CI, 0.73–2.75]; P=0.30) (interaction P=0.61; Figure 2; Tables 4 and 5). There were no statistically significant differences in incidence of any of the individual end points with 30 versus 12 months of DAPT when stratified by frailty status, though both nonfrail and frail patients had numerically higher incidence of major bleeding with extended DAPT (Tables 4 and 5).

Figure 2. Cumulative incidence of net adverse events based on claims‐based frailty and duration of DAPT in the EXTEND‐DAPT study.

All outcomes were assessed from time of randomization (12 months following PCI) to 18 months post‐randomization (30 months following PCI) in the DAPT Study. NACE was a composite outcome inclusive of all‐cause mortality, MI, stroke, and major bleeding. CFI indicates Claims‐Based Frailty Indicator; DAPT, dual antiplatelet therapy; MI, myocardial infarction; NACE, net adverse clinical events; and PCI, percutaneous coronary intervention.

Table 4.

Incidence of Adverse Outcomes by Dual Antiplatelet Therapy Duration and CFI

Outcome	30 Mo nonfrail CFI ≤0.12 (n=571)	12 Mo nonfrail CFI ≤0.12 (n=596)	Log‐rank P value	30 Mo frail CFI >0.12 (n=81)	12 Mo frail CFI >0.12 (n=78)	Log‐rank P Value
NACE	11.5	9.9	0.36	26.7	19.5	0.30
MACCE	5.1	6.3	0.32	15.2	13.0	0.69
Death	2.5	1.5	0.25	7.6	5.2	0.54
Cardiovascular	1.1	0.7	0.33	1.3	2.6	0.55
Noncardiovascular	1.5	0.9	0.27	6.5	2.7	0.50
Myocardial infarction	2.9	3.8	0.39	7.7	5.3	0.51
Stroke	1.1	1.2	0.84	2.7	3.9	0.63
Major bleeding	4.2	2.6	0.15	14.4	7.8	0.24

CFI indicates Claims‐Based Frailty Indicator; MACCE, major adverse cardiovascular or cerebrovascular events; and NACE, net adverse clinical events.

Table 5.

Risks of Adverse Outcomes for 30 Versus 12 Months of Dual Antiplatelet Therapy Stratified by Claims‐Based Frailty

Outcome	30 vs 12 Mo nonfrail CFI ≤0.12, HR (95% CI)	30 vs 12 Mo frail CFI >0.12, HR (95% CI)	P value for interaction
NACE	1.18 (0.83–1.68)	1.42 (0.73–2.75)	0.61
MACCE	0.78 (0.48–1.27)	1.19 (0.51–2.75)	0.40
Death	1.63 (0.71–3.76)	1.48 (0.42–5.23)	0.90
Cardiovascular	1.57 (0.44–5.56)	0.49 (0.05–5.42)	0.40
Noncardiovascular	1.68 (0.55–5.12)	2.46 (0.48–12.67)	0.70
Myocardial infarction	0.75 (0.40–1.43)	1.52 (0.43–5.39)	0.33
Stroke	0.89 (0.30–2.66)	0.65 (0.11–3.89)	0.76
Major bleeding	1.61 (0.84–3.08)	1.81 (0.67–4.89)	0.83

CFI indicates Claims‐Based Frailty Indicator; HR, hazard ratio; MACCE, major adverse cardiovascular or cerebrovascular events; and NACE, net adverse clinical events.

When minor bleeding was included, patients with extended duration DAPT had a similar relative effect on the incidence of bleeding for patients with CFI <0.12 (8.1% versus 4.0%, P=0.003; HR, 2.09 [95% CI, 1.26–3.45]) and CFI ≥0.12 (20.7% versus 11.8%, P=0.16; HR, 1.77 [95% CI 0.78–4.02]; interaction P=0.75; Tables S6 and S7). Absolute risk differences for all outcomes are shown in Table S8.

Sensitivity Analyses Using Frailty‐Associated Characteristics in Claims Data

In sensitivity analyses based on a cumulative deficit approach for frailty, a total of 761 (57.4%) patients had 0 to 3 FACs, 326 (24.6%) had 4 to 5 FACs, and 239 (18.0%) had ≥6 FACs. The overall distribution of FACs is shown in Table S9, and baseline characteristics are shown in Table S10. Aside from demographic details and traditional comorbidities, the most common FACs identified in claims data were arthropathies/musculoskeletal disorders (32.0%), neurotic/personality/other nonpsychotic mental disorders (15.8%), and other diseases of the urinary system (13.7%; Table S11).

Consistent with the primary analyses, patients with frailty based on having ≥6 FACs had increased incidence of adverse outcomes in unadjusted analyses and after multivariable adjustment (Tables S12 and S13; Figure S2). Evaluating FACs as a continuous variable showed that increasing number of FACs was associated with increased risk of most outcomes including NACE and bleeding (Table S14). When minor bleeding was included in the bleeding end point, patients with greater number of FACs similarly had higher incidence and adjusted risk of bleeding and NACE (Tables S15 and S16).

There were no statistically significant differences in incidence of any of the individual end points with 30 versus 12 months of DAPT when stratified by frailty status based on ≥6 FACs (Tables S17 and S18; Figure S3), and findings for NACE and bleeding did not change when minor bleeding was included (Tables S19 and S20).

DISCUSSION

In this study we showed that frailty defined by a previously validated claims‐based index was associated with NACE in a subgroup of older patients who underwent PCI and were enrolled in the DAPT Study. We found numerical increases in NACE with longer DAPT duration among more frail patients; however, these analyses were underpowered, and differences were not statistically significant. Findings from this study support further inquiry into how frailty impacts ischemic and bleeding risks with antiplatelet therapies for CAD, and also support supplemental use of administrative claims data in the context of cardiovascular clinical trials.

Though administrative claims data have been widely used for observational studies, ³⁴ , ³⁵ , ³⁶ , ³⁷ they have not been frequently used in studies of randomized clinical trials. However, inclusion of claims in data collection for clinical trials may provide several unique benefits, such as more efficient end point ascertainment, lower costs, and longer‐term follow‐up. ²⁶ , ²⁷ , ²⁸ , ³⁸ , ³⁹ , ⁴⁰ , ⁴¹ Claims data may also provide important clinical information not otherwise assessed in clinical trial data. Frailty is rarely measured or included in clinical trials, although it is a strong predictor of adverse outcomes. ¹ , ⁷ Consequently, the generalizability of most clinical trial findings to frail patients is uncertain. However, novel measures of frailty have been developed using claims data ¹⁵ , ¹⁶ , ¹⁷ and may have usefulness when included in clinical trials. ⁴²

In this post hoc analysis of the DAPT Study, we were able to use claims data to identify a subgroup of clinical trial patients with greater frailty and show that patients with CAD and frailty had increased risk of NACE in long‐term follow‐up. However, our cohort was smaller than the overall DAPT Study because only a minority were in the United States and had inpatient claims data available with Medicare fee‐for‐service insurance. Not all US patients had PCI performed at a hospital that participated in the CathPCI Registry, and outpatient and observation Medicare claims data were also not available, which likely contributed to the smaller size of our cohort. As a result, there was low power to detect significant differences in treatment effects between frail and nonfrail patients, although adverse outcomes were generally increased among frail patients. These findings appear to be consistent with prior studies showing that patients with CAD and frailty based on other measures such as the Fried frailty phenotype are at increased risk of adverse events, including death, ⁴³ , ⁴⁴ , ⁴⁵ as well as MI, stroke, and bleeding. ⁵ , ¹⁰ , ¹¹ , ⁴⁶ , ⁴⁷ , ⁴⁸ , ⁴⁹ , ⁵⁰

We add to the evidence available from these prior observational studies by demonstrating that claims data can potentially be used to approximate frailty when it is not otherwise prospectively measured in a clinical trial. Ideally, clinical trials would directly measure frailty at baseline using validated in‐person assessment tools, and investigators should carefully consider this in early stages of trial design. However, due to reasons such as inadequate time, the labor‐intensive nature of such assessments, prohibitive cost, or lack of prioritization by investigators, this is rarely done. Frailty assessments based on previously validated claims‐based indices may have prognostic value similar to other methods of measuring frailty. ⁴² Data from administrative claims have particular advantages for use in clinical trials given that they are ubiquitously used in health care, are often less expensive than other data collection methods, and can be easily analyzed retrospectively. In addition, claims data with long‐term follow‐up, such as in Medicare, allow for longitudinal assessment of outcomes associated with frailty. Claims data are already acknowledged as an important component of national efforts to expand use of real‐world evidence. ⁵¹ Further investigation on the use of claims data within clinical trials to assess frailty and perhaps other unmeasured variables of interest is therefore warranted. Greater use of claims data would require more regular linkage with clinical trial data, as has occurred retrospectively in this study and prospectively within several pragmatic trials. ⁵² , ⁵³

In randomized comparisons in our study, we did not find statistically significant differences in treatment effects of extended duration DAPT after PCI for frail and nonfrail patients. Even so, we did find that frail patients had a numerically 37% higher incidence of NACE and 85% higher incidence of major bleeding with 30 months of DAPT, reinforcing the anticipated low power of our analysis. Importantly, decisions related to antiplatelet therapy following PCI often depend on balancing the protective benefits of therapy for reducing ischemic events with potential bleeding risks. Prior observational evidence has shown that frailty is associated with higher risk of bleeding related to cardiac catheterization and other causes of major bleeding within 30 days of acute coronary syndrome, ¹⁰ , ¹¹ but there is little evidence from clinical trials on how frailty impacts bleeding risks of antiplatelet therapies. In the overall DAPT Study, increasing age was a significant predictor of bleeding but not ischemic events. ¹⁴ Additional data from the Clopidogrel for High Atherothrombotic Risk and Ischemic Stabilization, Management, and Avoidance Trial showed that long‐term clopidogrel use more than doubled the risk of bleeding for patients ≥75 years of age compared with placebo. ⁵⁴

Some of the bleeding risk associated with age may be related to frailty‐associated mechanisms including underlying inflammatory dysregulation, decreased thrombus stability, decreased platelet turnover, variations in pharmacokinetics due to decreased muscle mass, and increased vascular fragility. ⁶ , ¹¹ , ⁵⁵ However, clinical trials of antiplatelets for patients with CAD tend to exclude frail patients and do not specifically measure frailty or assess its effect on bleeding and other outcomes. When considered with a prior EXTEND‐DAPT study that suggested attenuated benefits and greater harms with DAPT in contemporary real‐world populations, ⁵⁶ our study may further bring into question the use of extended duration DAPT in frail patients. Our work also emphasizes the need to assess frailty more routinely in clinical trials of patients with CAD. Importantly, frailty is not necessarily equivalent to advanced age, and additional tools are needed to assess frailty more accurately in clinical practice and trial settings. Examples of these include validated measures such as the Fried phenotype and Clinical Frailty Scale. ²⁹ , ⁵⁷ Development of a frailty score specific to older patients with CAD could also help predict outcomes and guide medication decisions related to DAPT following PCI.

Our study has several limitations. We indirectly assessed frailty based on ICD‐9‐CM diagnosis codes in claims data rather than more widely used measures such as the Fried phenotype. However, the validated index we used was based on the Fried definition and has been consistently associated with worse clinical outcomes in prior studies. Furthermore, a claims‐based definition of frailty using a cumulative deficit approach was also used and demonstrated similar findings. Even so, frailty encompasses important domains that include psychosocial and nutritional factors, which could not be directly assessed in our data. ³ For example, lack of independence at home is a marker of frailty and is associated with increased mortality after PCI, ⁵⁸ but is not fully captured in claims data. Another limitation is that only inpatient ICD‐9‐CM diagnosis codes from Medicare fee‐for‐service beneficiaries were available and used during the time period of the trial. Further evidence is therefore needed to determine how frailty and associated outcomes can be assessed with the currently used inpatient and outpatient ICD‐10 codes, in populations enrolled in Medicare Advantage, and in those <65 years of age. As previously noted, our cohort was a selected subgroup of patients with Medicare in the DAPT Study, and the number of patients defined as frail was relatively small. Therefore, our analyses were likely underpowered to detect statistically significant differences, particularly for outcomes with few events, and all randomized comparisons should be considered exploratory. The size of our cohort also limited the number of variables we could include in adjusted analyses. Lastly, our study was limited to patients who met inclusion and exclusion criteria for the DAPT Study. Therefore, our findings are not necessarily generalizable to the larger population or to frail patients who would have likely been excluded from trial participation.

CONCLUSIONS

In this study we found that frailty based on claims data was associated with higher net adverse clinical events in a subgroup of older patients enrolled in the DAPT Study. We had insufficient power to detect differences in the treatment effect for extended duration DAPT following PCI for frail and nonfrail patients, although our findings suggest that further investigation is warranted. Claims data may be useful for augmenting risk prediction and assessing heterogeneity of treatment effects in cardiovascular clinical trials.

Sources of Funding

This research was funded by the National Heart, Lung, and Blood Institute (grant 1R01HL 136 708‐01, Dr Yeh).

Disclosures

Dr Faridi receives research funding from the National Institutes of Health/National Heart, Lung, and Blood Institute (K23HL161424), outside the scope of the current work. Dr Strom is funded by a grant from the National Institutes of Health/National Heart, Lung, and Blood Institute (1K23HL144907). Dr Strom also reports research funding from Edwards Lifesciences, Ultromics, Anumana, and HeartSciences; consulting for Bracco Diagnostics; and speaker honoraria from Northwest Imaging Forums, outside of the current work. Dr Kundi is funded by a grant from the Scientific and Technological Research Council of Turkey (3501;120S422), outside the scope of the current work. Dr Curtis has a contract with the American College of Cardiology for his role as Senior Medical Officer, NCDR, and receives salary support from the American College of Cardiology, NCDR. He holds equity interest in Medtronic. Dr Shen is an employee of Biogen. Dr Secemsky receives research funding from the National Institutes of Health/National Heart, Lung, and Blood Institute (K23HL150290), Food and Drug Administration, Harvard Medical School's Shore Faculty Development Award, AstraZeneca, BD, Boston Scientific, Cook, CSI, Laminate Medical, Medtronic, and Philips. He also is a consultant and speaker for Abbott, Bayer, BD, Boston Scientific, Cook, CSI, Endovascular Engineering, Inari, Janssen, Medtronic, Philips, and VentureMed. Dr Yeh receives consulting fees and research funding from Abbott Vascular, Boston Scientific, and Medtronic; research funding from BD Bard, Cook, and Philips; and consulting fees from Elixir Medical, Shockwave, and Zoll. The remaining authors have no disclosures to report.

Supporting information

Tables S1–S20

Figures S1–S3