Incidence of Major Adverse Cardiovascular and/or Limb Events among Patients Using Aspirin for Secondary Prevention

Background: Low-dose aspirin (ASA) is often used for secondary prevention of major adverse cardiovascular events (MACE) and/or major adverse limb events (MALE) in chronic coronary artery disease (CAD) or peripheral artery disease (PAD). The COMPASS trial demonstrated that adding rivaroxaban to low-dose ASA significantly decreased relative risk of these events. However, its absolute impact will depend on real-world effectiveness of ASA monotherapy, as clinical trials differ from clinical practice. Objectives: To examine real-world rates of MACE and MALE among patients receiving ASA as secondary prevention. Methods: We used a large US claims database linked to electronic medical records to identify patients with chronic CAD/PAD who were using ASA for secondary prevention. We then assessed occurrence of MACE and MALE from date chronic CAD/PAD was established to earliest of death, disenrollment, or end of study. Results: A total of 1,285 patients met selection criteria (mean age: 69.6 years; 60.7% male). Over a mean follow-up of 2.3 years, 16.0% experienced MACE or MALE (7.8 events per 100 person-years). Patients were more likely to experience events than the “ASA-only” COMPASS arm, including MACE (11.7% vs. 5.6% in patients with CAD; 14.0% vs. 6.9% in patients with PAD) and MALE (e.g., critical limb ischemia: 12.9% vs. 1.0% in PAD). Conclusions: Real-world risk of MACE/MALE among patients using ASA as secondary prevention are substantially higher than that reported in the COMPASS trial. Further study is needed to determine whether emerging secondary prevention strategies have greater absolute impact on real-world clinical outcomes than those observed in clinical trials.


Introduction
Patients with coronary artery disease (CAD) and peripheral artery disease (PAD) are at elevated risk for build-up of atherosclerotic plaques. When ruptured, the sub-endothelial matrix is exposed to circulating blood, which in turn promotes the coagulation cascade and platelet aggregation, leading to arterial thrombi. Via this mechanism, patients with CAD or PAD are at high risk of experiencing major adverse cardiovascular events (MACE) (Bhatt et al., [1]) (e.g., stroke, myocardial infarction [MI], cardiovascular [CV] death) or major adverse limb events (MALE) (e.g., limb ischemia, amputation, arterial revascularization) (Gerhard-Herman et al., [2]). Low-dose (75-162 mg/d) aspirin (ASA) inhibits platelet aggregation and thus the risk of arterial thrombi (Antithrombotic Trialists Collaboration et al., [3]). When used as secondary prevention, it has been demonstrated to reduce risk of MACE by 20%, including a 9% reduction in risk of CV death (each versus placebo) (Antithrombotic Trialists Collaboration et al., [3]).
This risk reduction typically corresponds to an absolute reduction in annual incidence of fatal events of between 10-20 per 1,000, and a lesser, albeit still significant, reduction in vascular death and other serious events (Antithrombotic Trialists Collaboration [4]). The effectiveness of ASA in this context of secondary prevention is thus well established, and the American Heart Association currently assign it a level A, class I recommendation as secondary prevention for MACE and MALE among patients with known coronary heart disease or atherosclerotic vascular disease (Smith et al., [5]).
The COMPASS study was a large (n=27,395), double-blind, double-dummy, randomized trial that compared rivaroxaban (a selective direct factor Xa inhibitor) in combination with low-dose ASA versus low-dose ASA alone for the prevention of secondary CV-related events among patients with chronic CAD and/or PAD (Eikelboom et al., [6]). Patients randomized to rivaroxaban (plus low-dose ASA) were less likely than those randomized to low-dose ASA alone to experience the composite outcome of MI, stroke and CV death (hazard ratio: 0.76; 95% confidence interval 0.66-0.86; p<0.001); the trial was stopped early as pre-specified criteria for superiority were attained earlier than anticipated. In part based on evidence from COMPASS, the United States Food and Drug Administration (FDA) has approved rivaroxaban (in combination with ASA) to reduce the risk of CV death, MI, and stroke among patients with chronic CAD or PAD (it is also indicated to reduce While rivaroxaban is a welcome addition to the armamentarium of secondary prevention among patients with existing CAD and PAD, the magnitude of its clinical benefit is likely at least in part dependent on the effectiveness of ASA as secondary prevention. For example, if most patients with prevalent disease currently receive secondary prevention with ASA-and if such therapy sufficiently reduces risk of MACE and/or MALE to an acceptably low level-clinician familiarity with ASA and its low acquisition cost (typically purchased as over-the-counter [OTC] medication in the US) will likely dissuade practitioners from prescribing additional therapies as prophylaxis. Conversely, should rates of MACE and/or MALE remain relatively high among patients treated with ASA, the importance of additional secondary preventionincluding add-on therapy with an agent such as rivaroxabanincreases. Consequently, to better understand the potential impact to clinical practice of this recent expansion of available secondary prevention options, we used a large US healthcare claims database linked to patients' electronic medical records (EMR) to examine the real-world incidence of MACE and/or MALE among patients with chronic CAD and PAD, who used ASA as secondary prevention. Our objective was to determine if the risk of MACE and MALE amongst real-world patients currently using an "ASA-only" approach justifies more aggressive secondary prevention measures.

Data Source
We used the Optum Integrated Database, which includes administrative healthcare data (i.e., claims, eligibility for medical and pharmacy benefits, demographic data) linked to EMR for days subsequent to the initial outpatient visit, without evidence of emergent care for CAD (or PAD) between visits (the second outpatient visit was designated the "index date") ( Figure 1). We focused attention on the period spanning July 1, 2009 to September 29, 2016 so that all selected patients had ≥6 months of pre-index information and a minimum of one day of follow-up. For patients with multiple index dates, we selected the earliest identified.
Patients with <6 months enrolment prior to index date were dropped from the analyses, as were those: 1) with evidence of MACE or MALE within the 30-day period prior to index date; 2) who were aged <18 years at index date; 3) with evidence of atrial fibrillation during the 6-month pre-index period (these patients were dropped as anticoagulant therapy was already indicated for this condition at the time the work was conducted); 4) without evidence of use of ASA during pre-index period (evidence established through either ≥ 1 relevant notations in patients' EMR or ≥1 claims for such therapy [in the US, most ASA use is OTC, but physicians may write prescriptions for such therapy]); or 5) without claims and EMR data available during pre-index. Follow-up for each patient began on the index date and extended until the earliest of death, disenrollment from health plan, or end of study period. All available information from patients' claims and EMR were compiled for the pre-index period and all of follow-up.

Measures
Demographic (e.g., age, gender, geographic region, race, insurance type), clinical characteristics (e.g., body mass index Mortality was assessed using specific information within the database (e.g., hospital discharge disposition, evidence of death recorded in the EMR, diagnosis of death [e.g., ICD-9-CM 348.82, ICD-10 G93.82]), augmented by methods described by Paramore and colleagues (Paramore, Thomas, Knopf, Cragin, & Fraeman, [8]). Specifically, evidence during follow-up of a serious event (e.g., metastatic cancer, trauma, aortic dissection, pulmonary embolism) resulting in ambulance service or emergent care (i.e., hospital admission or ED visit) without subsequent claims for medical care or pharmacotherapy, and disenrollment the next month, were assumed to be fatal events. For all patients with evidence of death, those with CV-related diagnoses on the last date for which data were available (or during the admission where applicable) were assumed to have experienced CV-related death.

Statistical analyses
Pre-index measures were summarized using descriptive statistics (e.g., means, standard deviations [SDs] for continuous variables; numbers and percentages for categorical variables).
The number and percentage of patients experiencing MACE and/ or MALE was estimated, as were the corresponding rates, which were expressed per 100 person-years (PYs) and were estimated by dividing the number who experienced the event by the total time at risk for the event (in years) and multiplying the result by

Discussion
In our real-world cohort of patients with prevalent and chronic CAD or PAD who were using ASA as secondary prophylaxis,  [3]). While we found substantially higher rates of MACE and MALE associated with ASA use than these studies, we note that the latter are from randomized trials, and therefore These differences also raise questions into whether ASA only is sufficient as secondary prevention. Given that in COMPASS rivaroxaban plus low-dose ASA reduced the risk of MACE (the primary outcome of COMPASS) by 24% (vs. low-dose ASA alone), a comparable outcome (assuming comparable real-world performance of this combination regimen to that reported in COMPASS) would be a reduction of MACE from 11.8% (n=151) of patients in clinical practice to 9.0% (N=115). However, we note that such a risk reduction would also be in tandem with an increase in risk of major bleeding (fatal bleeding, bleeding into a critical organ, bleeding into a surgical site requiring reoperation and bleeding that led to hospitalization) (subjects randomized to receive rivaroxaban plus low-dose ASA had a 1.7-fold increased risk of major bleeding vs. those randomized to ASA-only) (Eikelboom et al., [6]). Despite Other retrospective observational studies investigating the performance of ASA as secondary prevention among patients with established CAD or PAD appear to be lacking. Information from the Reduction of Atherothrombosis for Continued Health (REACH) registry has been used to estimate rates of MACE or MALE among persons with existing CAD or PAD. In one study based on data from the REACH registry that focused on 65,531 "COMPASS-like" patients who were using ASA as secondary prevention, the investigators reported 4.2 MACE per 100 PYs (Darmon et al., [11]). While this reported rate is nominally lower than the 5.5 per 100 PYs observed among patients receiving ASA in our study, it is consistent with our findings in that it is numerically higher than the 2.9 per 100 PYs reported in COMPASS (Darmon et al., [11]). While conjectural, it is possible that application of COMPASS selection criteria may have resulted in exclusion of some patients at relatively high risk for MACE or MALE (i.e., the rate of these events among all REACH participants may be greater than that estimated among the  [13][14][15]). In the aforementioned REACH study, 79% of "COMPASS-eligible" patients had evidence of ASA use (Darmon et al., [13]); however, self-reported ASA use has declined between calendar-years 2012 and 2015from 22.8% to 20.4% (Stuntz & Bernstein, [15]). Another examination, this one of patients with PAD that was based on the National Health and Nutrition Examination Survey (NHANES) found that 64.2% reported that they did not use ASA for secondary prevention; overall, between 61% and 75% reported not using any pharmacotherapies for secondary prevention (Pande, Perlstein, Beckman, & Creager, [16]). This is consistent with well-known risks of non-adherence-the American Hospital Association has estimated that 12% of patients that are prescribed therapies do not fill their prescriptions (American Heart Association). While unknowable from our database, these and related challenges for secondary prevention may help explain to at least some degree the relatively high rates of MACE and MALE we observed in a real-world population of chronic with CAD and PAD. Regardless, these challenges-and our findings-serve as a stark reminder to clinicians of the importance of effective secondary prevention.
Like any observational study, ours has several limitations that should be noted. First is the issue of missing data among our initial CAD and PAD patient sample. While we initially identified Other observational studies have confirmed that ASA use may be under recorded (Berger & Ladapo, [17]); accordingly, it is likely that ASA use was more widespread among patients in our sample than available data would indicate. Many patients who use ASA for secondary prevention likely use over-the-counter (OTC) (as opposed to prescription) preparations, and our study focused on those with prevalent CAD/PAD, which means that the relevant information on OTC products may not have been recorded during the study period. However, the patients who used ASA in our sample were likely to be "true positives", as they all had evidence of use of such therapy-predominantly from information available in their EMR. Given the goals of our study, we thought it important to maximize specificity (i.e., risk not including all "true positives") in order to limit the possibility of including patients who did not use ASA for secondary prevention (i.e., "false positives"). While reasons for omission of this information from the EMR are unknowable, we note that rates of MACE and MALE were comparable between those with and without evidence of pre-index ASA use (data not shown). Even if all patients in our sample did in fact use ASA for secondary prevention, the relatively high rates of MACE and MALE across our sample (as compared with either patients randomized to ASA in COMPASS and other trials or real-world experience as described by REACH) indicate that the effectiveness of such therapy in preventing CV events in clinical practice is suboptimal.
Second, both claims and EMRs are subject to errors of omission and/or commission, with unknowable impact to our study. Third, and as noted above, our decision to include almost all patients with evidence of chronic CAD and/or PAD resulted in a sample who have lived with CAD or PAD for varying lengths of time, with resulting impacts on disease severity/risks of CV events (although we note that all patients were required to have not experienced either ED visits or hospitalizations resulting in diagnoses of CAD or PAD for at least 6 months prior to index date). Accordingly, while our sample likely reflects the mix of patients in typical clinical practice, resulting estimates of MACE and MALE reflect a somewhat "average" rate across varying patient-types. Fourth, we focused our attention on the first occurrence of each event during the follow-up period, and thus are likely underestimating "true" rates of MACE and MALE (i.e., our findings are likely conservative in terms of the current risk of MACE and/or MALE among patients using ASA for secondary prevention). Finally, while large and geographically diverse, ours was a convenience-not a random-sample. Consequently, the generalizability of our findings to all patients with chronic CAD and/or PAD is unknown [18,19].

Conclusion
In conclusion, rates of MACE and MALE among patients in clinical practice with chronic CAD and/or PAD who use ASA for secondary prevention are substantially greater than those reported in the COMPASS trial. Our findings reinforce that low-dose ASA monotherapy may be a suboptimal secondary prevention strategy in clinical practice. Further study is needed to understand the degree to which alternative management strategies-including but not limited to the addition of rivaroxaban to low-dose ASAmay reduce the incidence and burden of MACE and MALE among patients with chronic CAD and/or PAD in real-world settings.