Effect of antiplatelet persistence on long-term mortality and predictors of non-persistence in ischemic stroke patients 75 years and older: a nationwide cohort study

Stroke is one of the major causes of death in very elderly patients, with nearly one-third of all strokes occurring in patients aged 75 years and older [1‐3]. Considering the steady increase in life expectancy, the incidence of stroke is also expected to be more than double over the next 30 years, with a majority of increase in patients aged 75 years and older [3]. Of all strokes, 87% are ischemic [3] and antiplatelet therapy (APT) is the gold standard for secondary prevention [4, 5]. Persistence with APT after ischemic stroke has been proven to result in better clinical outcomes, including decreased vascular death [6, 7]. However, scientific evidence of optimal APT for patients aged 75 years and older have been quite limited since they were substantially excluded from most randomized clinical trials due to age limits or the presence of multiple comorbidities [1, 8, 9]. The benefits of early antiplatelet use after ischemic stroke were similarly observed in patients aged 75 years and older compared to younger patients in a previous meta-analysis [10]. However, it remains unclear whether persistence with APT carries long-term mortality benefits to ischemic stroke survivors in the same old age group. Thus, we analyzed data from a large-scale nationwide claims database in South Korea to gather real-world evidence on the benefits of persistence with APT on long-term mortality in patients aged 75 years and older with ischemic stroke. Furthermore, if persistence with APT were also beneficial to this age group, we aimed to identify the potential predictors of non-persistence to offer a meaningful perspective for improving persistence.

In this nationwide retrospective cohort study, we confirmed that even among those aged 75 years and older, persistent APT after initial ischemic stroke was associated with reduced risks for both CVD and non-CVD death. In fact, the risk of mortality of any cause, CVD-related, and non-CVD deaths for those who discontinued APT within 6 months were more than doubled compared to patients who continued APT for 6 months and beyond. The higher risk of CVD mortality in the non-persistent group is presumably due to a higher incidence of recurrent stroke or coronary artery diseases including acute coronary syndromes in this group than in the persistent group. The risk of recurrence after initial ischemic stroke is known to remain elevated for several years [23] and its rates are even higher in elderly patients. The incidence of first-ever [24] ischemic stroke was also an independent risk factor for major cardiovascular events, including incident coronary artery disease, acute coronary syndrome and myocardial infarction in Canadian patients aged 66 years and older [25]. The risk of these events increases even more when APT is discontinued prematurely [6, 7], and this likely resulted in a higher risk of CVD mortality in patients aged 75 years and older in this study. Furthermore, even if patients survive recurrent stroke and acute coronary syndrome, the consequences of these events may lead to complications that increase the risk of non-CVD death. Cognitive and physical dysfunction due to the sequelae of stroke recurrence or acute coronary syndrome may force patients to be hospitalized longer and be more bed-ridden, which would eventually cause more non-CVD deaths due to events such as falls, pulmonary thromboembolism, aspiration pneumonia, various hospital-acquired infections, and depression [26]. This may also further explain our findings. In addition, our Kaplan-Meier curves for cumulative probability of event between the two groups showed that the degree of difference for all-cause death, CVD, and non-CVD death all remained in a similar range throughout the study period. This suggested that APT for at least 6 months after ischemic stroke continued to bring mortality benefits to patients aged 75 years and older over time. Thus, persistence and adherence to antiplatelets after ischemic stroke should also be emphasized to patients in this age group.

When it comes to the factors influencing the non-persistence with APT after ischemic stroke for patients 75 years and older, older age, less total prescribed drugs (< 4 total drugs), and having more comorbidities (CCI score ≥ 6) were positive predictors of premature discontinuation of APT within 6 months.

In terms of age, both the cognitive and physical functions of every individual undoubtedly decline with increasing age, and impairment in these functions have been well-known risk factors for medication non-adherence in the elderly population [27‐29]. This trend of older age and lesser medication adherence was consistent with a previous Chinese study that was also conducted among ischemic stroke patients [30].

Regarding polypharmacy, the present study confirmed that taking less than 4 total prescribed drugs was predictive of early discontinuation of APT. This result was quite contrary to the general belief that the more drugs patients have to take, the poorer is their medication adherence. This tendency might be explained by the “health belief model.” [31, 32] According to the health belief model, those who believe they are at lower risks of developing certain diseases are more likely to engage in relatively unhealthy behaviors [31, 32]. Since the general medical condition of patients with 1–3 total prescribed drugs is likely to be better than those with 4 or more total medications, the significance of continuing APT after ischemic stroke may have been less important for them. Other studies that reported a positive relationship between a lower total number of drugs and medication adherence of antihypertensives or statins also used health belief model for the interpretation of their results [33, 34].

Nevertheless, our findings revealed that the health belief model becomes ineffective when the individuals’ general health status was too serious, as having more comorbidities (CCI score ≥ 6) was a notable predictor of non-persistence with antiplatelets after ischemic stroke for patients aged 75 years and older. Various studies have reported that more comorbidities for geriatric ischemic stroke patients are significantly associated with more functional dysfunctions, subsequently limiting activities of daily living [35‐38]. We believe that this is the reason for the increased risk of early APT discontinuation in patients with CCI scores of 6 and higher. Having more comorbidities was also confirmed as a risk factor for poorer antihypertensive adherence in a previous study [39].

Interestingly, the conventional socioeconomic risk factors for medication non-adherence, such as household income, type of insurance, and residential area [40‐42] were not predictive of APT non-persistence after ischemic stroke in the patients aged 75 years and older. This may imply that for older patients with ischemic stroke, medication adherence is influenced more by an individual’s health condition or therapy-related factors rather than by socioeconomic factors. A Canadian registry-based stroke study also reported that the socioeconomic status of elderly ischemic stroke patients was not a decisive factor in determining better adherence to secondary preventive drugs [43]. Further studies regarding the specific characteristics of factors affecting APT adherence in very elderly ischemic stroke survivors should be performed to confirm these results.

The present study had some meaningful implications. First, by using nationally representative large-scale claims data, we provided real-world evidence that being persistent with APT also had long-term mortality benefits for ischemic stroke patients aged 75 years and older. Although previous studies have confirmed the benefit of early antiplatelet use for geriatric ischemic stroke patients [4, 10], evidence concerning its persistent use in very old ischemic stroke patients has been relatively limited. Second, we included various potential factors, including age, sex, household income, residence, health insurance type, comorbidities, and polypharmacy, that could influence APT persistence as covariates and discovered the independent predictors of non-persistence with APT after ischemic stroke in very old patients.

However, this study also had some limitations. First, the severity of stroke or neurological status could not be assessed due to the nature of the claims database. Second, the particular reasons for early APT discontinuation, such as the possible side effects of APT (bleeding and gastrointestinal toxicity) and the patients’ social history, including smoking and alcohol consumption status, were not addressed because of the lack of information in the NHIS-NSC. Nevertheless, we believe that these missed factors were likely to be distributed evenly to both case and control groups, considering the large sample size and the well-established representativeness of the NHIS-NSC. Third, since our investigation of medication persistence was only limited to APT, patients who switched from APT to anticoagulation therapy, due to the detection of paroxysmal atrial fibrillation (pAF), may have been misplaced into the non-persistent group. However, both the extent and range of work-up for detection of pAF among cryptogenic stroke cases substantially varies from one institution to another, and the majority of institutions are reported not to perform routine electrocardiogram (ECG) at follow-up outpatient visits after discharge [44]. In addition, detection rates of pAF for cryptogenic stroke patients were low (1.4% at 6 months and 2.0% at 12 months) with conventional ECG monitoring, and they were still relatively lower, even with an insertable cardiac monitor with only 8.9% at 6 months and 12.4% at 12 months [45]. Furthermore, even if we did discover all the patients who were detected with pAF, since anticoagulation was performed with only warfarin during our study period, it is technically impossible to properly evaluate the persistence and adherence of warfarin since its dose and prescriptions are adjusted in accordance with patients’ time in the therapeutic range. In addition, the non-persistent group already had significantly higher mortality risk despite the possibility of including those who switched to anticoagulation therapy in the non-persistent group. Thus, the effect of this limitation on the overall results of the present study is less significant. Fourth, since the follow-up of participants of this study was initiated at the time of the first outpatient antiplatelet prescription after discharge from the hospital, there is a possibility of potential immortal time bias. However, considering the fact that majority of ischemic stroke patients in Korea are prescribed with antiplatelets at discharge [46], the time period between discharge and the first outpatient prescription cannot just be regarded as untreated period. Thus, we believe that the impact of potential immortal time bias on the overall results of our study would be relatively small. Lastly, the over-the-counter (OTC) aspirin intake was not reflected because the NHIS-NSC contains only insurance-covered prescription information. However, we supposed that most of the patients would have taken aspirin through prescription rather than OTC purchase, since patients can receive the former at a discounted price via Korean health insurance coverage [47]. In addition, a previous study proved that prescription claims data can provide valid estimates regardless of missed OTC exposures [48].

In this retrospective population-based study, persistence with APT was markedly associated with a reduced risk for ACM, CVD mortality, and non-CVD mortality in ischemic stroke patients aged 75 years and older. Moreover, older age, having less than 4 total prescribed drugs, and a CCI score of 6 and higher were independent predictors of APT non-persistence. Therefore, improving APT persistence for ischemic stroke patients in this age group is also recommended by understanding factors associated with non-persistence.