Identifying Racial, Ethnic, and Socioeconomic Inequities in the Use of Novel P2Y12 Inhibitors After Percutaneous Coronary Intervention

Background. Novel P2Y₁₂ inhibitors prasugrel and ticagrelor were approved for patients with acute coronary syndrome (ACS) in 2009 and 2011, respectively. We assessed the association of racial, ethnic, and socioeconomic factors with initiation of and adherence to novel P2Y₁₂ inhibitors in a commercially insured population. Methods. We performed a retrospective cohort analysis of adults undergoing percutaneous coronary intervention with placement of a drug-eluting stent, stratified by ACS status, between January 2008 and December 2016 using Clinformatics Data Mart (OptumInsight). We estimated multivariable logistic regression models to identify factors associated with the initiation of clopidogrel vs novel P2Y₁₂ inhibitors as well as subsequent 6-month medication adherence, assessed via pharmacy records. Results. A total of 55,664 patients were included in the analysis. Hispanic ethnicity was independently associated with the initiation of clopidogrel compared with novel P2Y₁₂ inhibitors among ACS patients (odds ratio [OR], 1.19; 95% confidence interval [CI], 1.04-1.36; P<.01). ACS patients with an annual median household income of over $100,000 were less likely to be started on clopidogrel when compared with those who earned less than $40,000 (OR, 0.67; 95% CI, 0.61-0.75; P<.01). Black race, Hispanic ethnicity, and lower household income were each associated with significantly reduced odds of P2Y₁₂ inhibitor adherence. Conclusion. Hispanic ethnicity and lower household income were associated with novel P2Y₁₂ inhibitor initiation, and non-White race and ethnicity were associated with lower P2Y₁₂ inhibitor adherence over 6-month follow-up. These findings highlight continued inequity of care, even in an insured population, and point to a need for new strategies to close these gaps.

J INVASIVE CARDIOL 2022;34(3):E171-E178. Epub 2022 January 16.

Key words: antiplatelet, inequities

In landmark clinical trials, the novel platelet P2Y₁₂ receptor inhibitors prasugrel and ticagrelor reduced atherothrombotic events in patients with acute coronary syndrome (ACS) when compared with clopidogrel, a medication that has been approved by the United States (US) Food and Drug Administration (FDA) since 1997.^1,2 On the basis of these trials, prasugrel was approved by the FDA in July 2009 for use in patients with ACS when percutaneous coronary intervention (PCI) is anticipated, and ticagrelor was approved by the FDA for use in all patients with ACS in July 2011. Guidelines currently recommend 12 months of dual-antiplatelet therapy with aspirin and a P2Y₁₂ inhibitor following ACS, and a shorter duration can be considered following non-ACS PCI.³

Recent studies have demonstrated increases in the use of prasugrel and ticagrelor since initial FDA approval for both on-label and off-label indications.^4,5 However, there are concerns that the high out-of-pocket costs associated with prasugrel and ticagrelor may limit initiation and reduce adherence to these medications due to financial constraints. Non-adherence to P2Y₁₂ inhibitors can result in stent thrombosis, a significant complication following PCI.

Structural racism in healthcare, defined as differential access to quality care, services, and opportunity to achieve one’s full health potential based on race, is pervasive in the US.^6-9 Unequal access to care and disparate outcomes are particularly prominent in the care of cardiovascular patients.^6,10 Racial and ethnic minoritized and poorer patients are less likely to receive novel pharmacotherapies for cardiovascular diseases.^11,12 Furthermore, even if such underserved patients receive a prescription for novel therapies, structural barriers and patient factors, such as insurance or socioeconomic status, health literacy, and provider mistrust, may limit patient adherence to such therapies.^13,14

Prior investigations have not studied racial and socioeconomic inequities in novel P2Y₁₂ inhibitor initiation or subsequent medication adherence. In this study, we sought to evaluate the presence of racial and socioeconomic inequities in the patterns of initiation of clopidogrel vs the novel P2Y₁₂ inhibitors among a large, commercially insured population and characterize medication adherence to these therapies.

The data for this study were obtained from the Clinformatics Data Mart database (OptumInsight), a large private payer administrative claims database consisting of inpatient, outpatient, laboratory, and pharmacy claims for a geographically diverse cohort of 13 million patients annually. Race and ethnicity were determined in this database through a combination of public records, self-report, and proprietary ethnicity code tables, as described previously.¹² Demographic and socioeconomic data, including median household income, were available through ZIP-code linked enrollment data from the US Census Bureau. This study was classified as exempt by the institutional review board at the University of Pennsylvania.

Study population. We identified all patients over age 18 years who underwent PCI with placement of a drug-eluting stent for either ACS or non-ACS indications between January 1, 2008 and December 31, 2016. Patients were identified as having undergone PCI using International Classification of Disease, Ninth Revision (ICD-9-CM) codes 36.01, 36.02, 36.05, 36.06, 36.07, 00.66 or International Classification of Disease, Tenth Revision (ICD-10-CM) codes 02703xx, 02713xx, 02723xx, and 02733xx (except 027x3Tx and 027x3Zx). Diagnosis-related group (DRG) codes were used to identify placement of a drug-eluting stent (DRG codes 246 and 247). Patients were identified as having an ACS indication for PCI if they also had the following diagnosis codes (ICD-9-CMC codes 410.X0 and 410.X1, ICD-10-CM codes I20.0, I21, and I24, DRG codes 121, 122, and 123). Patients with a non-ACS indication for PCI were defined as those undergoing PCI without a concurrent diagnosis code for ACS. These methods have been previously used to identify patients undergoing PCI with a drug-eluting stent for ACS or non-ACS indications.^4,5 The primary outcome was the choice of antiplatelet agent initiated, which was defined as the first filled pharmacy claim for an antiplatelet agent within 30 days of discharge.

We excluded patients without continuous insurance enrollment in the year preceding PCI to ensure complete capture of clinical comorbidities and recent clinical events. We further restricted our cohort to patients who were insured for at least 30 days after PCI and who had previously filled at least 1 pharmacy claim during the year preceding PCI to ensure that we adequately captured P2Y₁₂ inhibitor initiation after PCI. We also excluded any patients who underwent major cardiac surgery during the hospitalization for PCI or had major bleeding, as this may have influenced the choice of antiplatelet agents. Finally, we excluded patients who had been prescribed clopidogrel, prasugrel, or ticagrelor in the 12 months prior to the index PCI encounter as well as those patients who had been prescribed an anticoagulant such as warfarin, dabigatran, rivaroxaban, edoxaban or apixaban, as these may have influenced the choice of P2Y₁₂ inhibitor after PCI.

Statistical analysis. Summary statistics for patient characteristics, stratified by P2Y₁₂ inhibitor initiated, are presented as means with standard deviations for continuous data and total number and percentages for categorical data. Student’s t-test was used to compare continuous data and the Chi-square test was used to compare categorical data.

In order to account for confounding in the relationship between race and socioeconomic factors and the initiation of novel P2Y₁₂ inhibitors after PCI, we estimated multivariable logistic regression models among those patients that filled a P2Y₁₂ inhibitor with the initiation of prasugrel/ticagrelor vs clopidogrel as the dependent variable, stratified by presence of ACS, to identify patterns in on-label vs off-label use. Independent variables included age, sex, US Census region of residence, previous diagnosis of stroke, ZIP-code linked household income, medication copayment for the first fill of a 30-day supply of the P2Y₁₂ inhibitor, Elixhauser comorbidity index (which was assessed using a 1-year look-back period), and yearly fixed effects.¹⁵ Candidate independent variables used in the model were selected a priori based on pathophysiological plausibility and all were used in the multivariable model. Estimated adjusted odds ratios (ORs) are reported with 95% confidence intervals (CIs).

Next, we assessed the association between race and medication adherence. We defined medication adherence using the proportion of days covered (PDC), a commonly used surrogate measure for adherence.¹⁶Adherent patients were defined as having a PDC ratio of >0.80 over 6 months after the index PCI encounter. We then estimated multivariable logistic regression models with adherence as the dependent variable among patients with ACS. Consensus guidelines recommend a shorter duration of P2Y₁₂ inhibitor therapy for patients undergoing stenting for a non-ACS indication, so these patients were excluded from this analysis.³ Similar to the prior model, independent variables included age, sex, region of residence, previous diagnosis of stroke, ZIP-code linked household income, medication copay, and Elixhauser comorbidity index based on a priori pathophysiological plausibility. We estimated secondary models for adherence including interaction effects between Black race or Hispanic ethnicity with copayment or household income.

There were fewer than 15% missing data for race/ethnicity and household income, and the effect of missing data was addressed via multiple imputation analysis.¹⁷ Statistical analyses were performed using SAS, version 9.4 (SAS Institute). All statistical testing was 2 tailed, with P-values <.05 designated as statistically significant.

A total of 55,664 patients who underwent PCI with a drug-eluting stent between 2008 and 2016 met the inclusion criteria for the analysis (Figure 1). Among these patients, a total of 37,748 (67.8%) underwent PCI for an ACS indication and 17,916 (32.2%) underwent PCI for a non-ACS indication.

Baseline demographic, socioeconomic, and clinical information for the entire cohort are presented in Table 1 and are stratified by ACS in Supplemental Table S1 and Supplemental Table S2. Among the 55,664 patients included in the analysis, a total of 36,746 patients (66.0%) filled a prescription for clopidogrel, 8055 patients (14.4%) filled a prescription for prasugrel, 3727 patients (6.7%) filled a prescription for ticagrelor, and 7136 patients (12.8%) did not fill any P2Y₁₂ inhibitor prescription. The mean age of the cohort was 63.2 ± 11.7 years and 70.1% were male. Among included patients, a total of 44,705 patients (80.3%) were White, 5823 patients (10.5%) were Black, 3855 patients (6.9%) were Hispanic, and 1281 patients (2.3%) were Asian. There was wide variation in household income, with 28.7% of patients’ households earning more than $100,000 annually, and 25.5% of patients’ households earning less than $40,000 annually. The copay of prescription per 30-day supply of P2Y₁₂ inhibitor was $21.70 ± 29.70 for patients receiving clopidogrel, $49.10 ± 37.40 for prasugrel, and $56.80 ± 58.60 for ticagrelor.

Results of the multivariable analysis among patients with ACS are presented in Table 2. Patients who were >75 years old as compared with those <54 years old (OR, 3.70; 95% CI, 3.30-4.17; P<.01) or who had a previous history of stroke (OR, 1.25; 95% CI, 1.13-1.31; P<.01) were more likely to receive clopidogrel. Hispanic ethnicity (OR, 1.19; 95% CI, 1.04-1.36; P<.01) was independently associated with the receipt of clopidogrel as compared with the newer P2Y₁₂ inhibitors; however, there was no significant association between Black race (OR, 0.99; 95% CI, 0.88-1.11; P=.83) or Asian race (OR, 0.84; 95% CI, 0.71-1.00; P=.05) and the use of clopidogrel over the newer P2Y₁₂ inhibitors. Patients with an annual household income of >$100,000 were less likely to receive clopidogrel when compared with those who earned <$40,000 (OR, 0.62; 95% CI, 0.57-0.69; P<.001), as were those who earned between $40,000-$100,000 (OR, 0.80; 95% CI, 0.73-0.87; P<.001).

Among non-ACS patients, where prasugrel and ticagrelor initiation was off-label (Table 2), Black race (OR, 0.99; 95% CI, 0.83-1.18; P=.89), Asian race (OR, 0.73; 95% CI, 0.52-1.01; P=.05) and Hispanic ethnicity (OR, 0.93; 95% CI, 0.76-1.15; P=.50) were not independently associated with the receipt of clopidogrel over prasugrel or ticagrelor. Patients with an annual household income of >$100,000 were less likely to receive clopidogrel when compared with those who earned <$40,000 (OR, 0.74; 95% CI, 0.64-0.87; P<.001). There was no significant difference in the initiation of clopidogrel as compared with prasugrel or ticagrelor among patients who earned between $40,000-$100,000 (OR, 0.95; 95% CI, 0.82-1.10; P=.48)

Overall, a total of 15,334 (46.4%) of the ACS patients were classified as adherent, with 12,186 (51.4%) of clopidogrel patients, 2289 (37.3%) of prasugrel patients, and 859 (27.2%) of ticagrelor patients having a PDC ratio of >0.80 (P<.001). Clopidogrel use was associated with increased P2Y₁₂ adherence (OR, 2.03; 95% CI, 1.91-2.16; P<.01). Black race (OR, 0.81; 95% CI, 0.75-0.88; P<.01), Asian race (OR, 0.79; 95% CI, 0.69-0.92; P<.01) and Hispanic ethnicity (OR, 0.77; 95% CI, 0.69-0.86; P<.01) were independently associated with significantly reduced odds of medication adherence when compared with White patients. In addition, patients earning >$100,000 annually had greater levels of adherence compared with patients earning <$40,000 annually (OR, 1.09; 95% CI, 1.01-1.17; P=.03) (Table 3). There was no significant interaction between Black race and household income of <$40,000 (OR, 0.98; 95% CI, 0.83-1.17; P=.83) and there was a statistically significant interaction between Black race and copayment, although the extent of effect modification was clinically small (OR, 1.03; 95% CI, 1.00-1.01; P=.03). There were no significant interactions between Hispanic ethnicity and household income (OR, 1.00; 95% CI, 1.00-1.00; P=.50) or copayment (OR, 0.96; 95% CI, 0.76-1.21; P=.49) on adherence.

Among commercially insured patients who underwent PCI with placement of a drug-eluting stent, we found that Hispanic ethnicity was independently associated with the preferential initiation of clopidogrel as compared with the novel, more potent P2Y₁₂ inhibitors prasugrel and ticagrelor. Moreover, we found that Black race and Hispanic ethnicity were associated with significantly lower overall P2Y₁₂ inhibitor adherence over 6 months following PCI for ACS. Patients living in ZIP codes with lower household incomes were less likely to receive the novel agents, and in addition, had lower medication adherence.

Inequitable care delivery based on race is pervasive and decreased adoption of novel therapies among Black and Hispanic patients has been demonstrated on numerous occasions.^6,11,12,18 We found that Hispanic patients are less likely to receive novel P2Y₁₂ inhibitors, although this was not seen among Black patients. Prior studies have demonstrated that inequities among Hispanic patients may in part be due to language barriers, but also include such factors as implicit biases against patients of color, less patient-centered relationships due to inadequate patient input on treatment decisions, and perceptions of ability to afford medication.^19-21

Lower median household income was significantly associated with clopidogrel initiation over prasugrel or ticagrelor. For patients with an ACS presentation who had a household income of >$100,000 per year, there was over 30% reduction in clopidogrel initiation when compared with a patient with a household income of <$40,000 per year. Notably, there was a graded response in patients in an intermediate income bracket, with roughly a 15% reduction in clopidogrel use for patients earning between $40,000-$100,000 per year. We found similar results in patients undergoing PCI for non-ACS indications; however, the magnitude of association between household income and novel P2Y₁₂ inhibitor initiation was reduced.

There are several plausible explanations for the findings in the current study. First, many patients who earn <$40,000 per year may be less able to afford higher copayments and out-of-pocket expenses associated with prasugrel or ticagrelor, and as a result they may preferentially request clopidogrel from their providers.²² Second, poorer patients may receive care at hospitals where physicians are less likely to use prasugrel or ticagrelor, either because of less knowledge of the treatment benefits or an institutional proclivity toward using lower-cost treatments.²³ Third, a prescribing physician's perception of a patient’s anticipated adherence to dual-antiplatelet therapy may be influenced by the patient’s socioeconomic status. Physician perception of a poorer patient’s ability to adhere to therapy may lead to differential care.²⁴ Non-adherence to dual-antiplatelet therapy after PCI is a significant risk factor for stent thrombosis, a devastating complication of PCI. Providers may believe that they are acting in a patient’s best interest by prescribing a less effective but more affordable option, reasoning that affordability may drive adherence. It should be noted that the current study was performed in a fully insured population and provider notions about adherence and ability to afford medications would likely be magnified in an uninsured population.

We did find that race, ethnicity, and socioeconomic status were significant factors associated with overall P2Y₁₂ inhibitor adherence, with lower adherence rates for prasugrel and ticagrelor as compared with clopidogrel. Socioeconomic status may affect adherence through limited affordability of these medications, as prior studies have demonstrated increased adherence of P2Y₁₂ inhibitors when copayments are reduced.²⁵ Elimination of copayments has been shown to reduce inequities in care.²⁶ However, Black and Hispanic patients face many structural barriers not characterized by this study that may limit ability to adhere to this therapy, including higher rates of distrust of providers and the healthcare system due to a long history of racism in healthcare, poorer health literacy, social determinants of health, and socioeconomic barriers not captured by median income.⁷ Notably, our secondary models including interaction terms confirmed that income and copayment do not meaningfully mediate the relationship between Black race and adherence. Lower rates of patient activation — ie, having the knowledge, skills, and confidence to fully understand and subsequently effect treatment options — has been shown to be lower in Black patients and mediated by lower health literacy.²⁷ Distrust in the healthcare system and poor communication between patients and providers, a lasting byproduct of the long history of racism in US medicine, can lead to lower adherence rates and adoption of recommended care.^28-31

The results of this study highlight the importance of efforts to increase access and affordability of important, life-saving medications and address structural barriers to improve medication adherence among all patients. Numerous previous studies have demonstrated inequitable distribution of health across the spectrum of socioeconomic status.^32,33 Poor patients and patients of color have been repeatedly shown to have worse health outcomes for a variety of medical conditions.

Providers’ assumptions that minoritized patients are less adherent to therapy can occasionally lead to inequitable care delivery.^24,34,35 The current report of lower P2Y₁₂ inhibitor adherence among minority patients ideally will not reinforce those assumptions, but rather encourage clinicians and health systems to focus efforts on addressing structural barriers in order to improve adherence among minoritized patients. Patients of color experience repeated discrimination and less patient-centered patient-provider interactions, further propagating mistrust.²¹ Small-scale interventions, such as phone-based motivational interviewing, have been demonstrated to improve medication adherence among racial minorities.³⁵ On a larger scale, interventions that target patients’ attitudes/beliefs, acknowledge multiple levels of perceived discrimination, and focus on mistrust and/or suspicion of healthcare providers and systems are likely to be of benefit.³⁶ Additionally, improved physician workforce diversity has been shown to increase uptake of important medical therapy among Black men.³⁶

Study limitations. Our study has several limitations. As this is an observational cohort analysis, although we attempted to adjust for covariates, there may be residual confounding present. We used an administrative database to quantitatively study differences in drug prescriptions among a large sample of patients; however, these data lack individual patient-level information that may have affected the decisions to initiate clopidogrel as compared with prasugrel or ticagrelor. The use of billing codes to document antiplatelet use may not indicate all valid reasons for choosing one drug or another, especially reasons that are hard to capture in administrative databases, such as patient preference, detailed patient-level social determinants of health (ie, education level, housing stability, primary language, and English proficiency), provider factors (ie, specialty or subspecialty), hospital factors (ie, performance status, urban/rural status, teaching status), or medication side effects. In addition, our assessment of racial ethnic status was limited to the definitions available in the administrative database, and we may not have complete granularity of patients’ backgrounds. We used ZIP-code linked household income, which may not fully capture an individual patient’s financial status, as there may be variability within the ZIP code. Similarly, it is possible that certain populations were more likely to switch antiplatelet agents due to side effects or clinical events, although this would be difficult to capture in an administrative database. As a result, we did not assess switching between agents. Given the use of an administrative database, we were not able to assess patients who were lost to follow-up. In addition, we did not include patients who never filled any P2Y₁₂ inhibitor, limiting generalizability to the entire population of patients who undergo PCI. Some patients may have filled a P2Y₁₂ inhibitor using a voucher, which would not have been captured in our data source. Finally, barriers to adherence were not characterized, which is a needed next step to improve equitable medication uptake and utilization among all races, ethnicities, and socioeconomic groups.

In a fully insured population, patients with Hispanic ethnicity and lower household incomes were more likely to be initiated on clopidogrel as compared with novel P2Y₁₂ inhibitors after PCI.Black race, Hispanic ethnicity, and lower household income were each associated with reduced medication adherence at 6-month follow-up. Our findings demonstrate inequities of care, even among commercially insured patients, and highlight the need to mitigate these disparities.

• Evaluation of a Real-World In-hospital Antiplatelet-Switching Strategy Following Coronary Interventions: The SWITCH Study
• Acute Stent Thrombosis: Severe In-Stent Malapposition and Insufficient Platelet Inhibition as Partners in Crime!