Hypertension is a major health problem worldwide and requires long-term management and continuous therapy. The prevalence among aged people is substantial; approximately 80% of very elderly patients (defined as those aged over 80 years) have been reported to have hypertension [1]. Similarly, according to a report from the National Health Service in England, the prevalence of hypertension was 66.6% in the population aged over 75 years, as compared with 29.6% in the overall adult populations in 2014 [2]. As the elderly population increases (especially the very elderly population) [3], the number of patients with chronic diseases requiring long-term care is also increasing. As a result, the healthcare expenditures are increasing worldwide, representing an important social and public health issue [4].

Effective management of hypertension in the elderly is vital for preventing disease progression and related comorbidities and for reducing health care costs [5, 6]. Although hypertension is a principal cause of cardiovascular and cerebrovascular events and is associated with increased mortality [6, 7], before the publication of the Hypertension in the Very Elderly Trial, the benefits of treatment in very elderly patients with hypertension were controversial. This trial confirmed that treatment of hypertension reduced mortality even in the very elderly, indicating its importance [8, 9].

Age has been considered a potential factor that can influence treatment persistence and adherence [10, 11], and previous studies have mainly focused on elderly populations [12, 13]. In 2008, van Wijk et al. measured persistence with antihypertensive treatment in elderly populations in the US, Canada, and the Netherlands, and found similar patterns of persistence among the three countries [12].

The patterns of medication persistence in Asian population may differ from those in Western countries because of differences in race, healthcare systems, and socioeconomic environment [14, 15]. Different racial and ethnic groups show diverse perceptions and attitudes towards health and medicines, which may affect adherence to therapy [16]. Various reimbursement coverages, drug prices, and financial barriers depending on country-specific healthcare system and socioeconomic environment have been reported to affect medication persistence [14, 15, 17].

However, very few studies have evaluated hypertensive treatment persistence and adherence in very elderly vs. elderly hypertensive patients in Asia, including Korea. In Korea, the National Health Insurance, a public insurance program operated by the Ministry of Health and Welfare, covers most of the Korean population claims data, including patient data such as the diagnosis, prescription medication, and medical costs, collected by each medical institution. Using these nationwide healthcare claims data in Korea, this study aimed to compare the antihypertensive treatment persistence and adherence patterns in very elderly patients with uncomplicated hypertension to those of matched elderly and adult patients. In addition, we also aimed to identify the factors associated with non-persistence in the very elderly population.

To our knowledge, this is the first nationwide population-based study focused on antihypertensive medication persistence and adherence in a very elderly population at the time of initiation of antihypertensive monotherapy in Korea. We found that very elderly patients were more likely to discontinue antihypertensive treatment than elderly or adult patients, and that the predictors of treatment non-persistence in the very elderly differed from those in the other age groups. The one-year treatment persistence rate in very elderly patients was significantly lower than that in elderly patients (59.5% vs. 65.2%, p < 0.001); after adjustment for potential confounding variables, very elderly patients remained at elevated risk of treatment non-persistence compared with elderly patients (aHR, 1.20; 95% CI, 1.13–1.27). Despite the statistically significant differences in treatment persistence and adherence among the three age cohorts, the one-year treatment adherence was more than 60% in all three groups. These results corroborate the findings from other studies conducted in Western countries [12, 13]. In 2008, van Wijk et al. reported that approximately 25% of elderly patients studied were non-persistent, defined as a treatment gap of at least 180 days during the first year of treatment, with little difference among the three examined countries (the US, Canada, and the Netherlands) [12]. In 2014, Tu et al. found that 64.6% of elderly patients in Canada were therapy-persistent for over one year [13].

The present study found that elderly patients aged 65–79 years were more persistent with antihypertensive treatment than adults, while the very elderly showed poor treatment persistence. This result is consistent with previously published results from Canada [13] and Sweden [21], which showed reduced odds of medication persistence in patients older than 75 years compared to those aged 66–70 years. However, other previous studies that have investigated the impact of aging on treatment persistence have reached conflicting results [22, 23]. These discrepancies may be partially explained by differences in age composition, reference groups, and methodology used in measuring persistence, such as a prescription gap, between the studies.

Unlike in previous studies, the present study included only uncomplicated hypertensive patients. Therefore, the Charlson Comorbidity Index score was low, and two-thirds of the study patients had no comorbid conditions. The findings from previous studies investigating comorbidities as a risk factor for treatment non-persistence have been inconsistent [13, 23]. In the present investigation, we found that patients with one or more comorbidities were significantly more likely to remain treatment persistent than those without a comorbidity in the very elderly population, whereas the presence of comorbidities did not predict poor persistence in adults or in the elderly. These findings may be explained by the fact that very elderly patients with comorbidities might be more likely to continue hypertensive treatment because they require regular visits to healthcare facilities for diseases other than hypertension. Furthermore, in the very elderly population, patients with comorbidities might be more likely to continue hypertensive treatment because they require regular visits to healthcare facilities for diseases other than hypertension.

Beneficiaries of medical aid receive prescription drugs free of charge in Korea. Accordingly, in our study, being a beneficiary of medical aid was positively associated with treatment persistence in the very elderly cohort (aHR, 0.83; 95% CI, 0.71–0.98). This finding is consistent with those of several studies that reported sufficient health insurance or prescription drug coverage as the principal contributing factor to treatment persistence [10, 14, 24]. On the other hand, in our study, the insurance type was negatively associated with treatment persistence in the adult cohort (aHR: 1.21; 95% CI: 1.06–1.39). This finding may be explained by the difference in eligibility for medical aid between the adult and very elderly populations in Korea. Adults who receive medical aid in Korea tend to have a very low economic status, which is a recognized risk factor for lower treatment persistence [14, 25].

Moreover, very elderly patients diagnosed with dementia or dyslipidemia were significantly more likely to remain persistent with antihypertensive treatment. The finding of the association between dyslipidemia and treatment persistence corroborates a number of published studies [12, 23, 25, 26]. However, our finding of higher persistence in very elderly patients with dementia contrasts with the results from several previous studies reporting a higher discontinuation rate for antihypertensive treatment among patients with dementia or cognitive impairment [27, 28]. These contradictory findings are somewhat puzzling, and may be partially explained by the lower medication complexity in older adults with cognitive impairment, as observed in one previous study [29]. The care management by caregivers may also increase the treatment persistence in the very elderly with dementia [30], which is supported by the findings of recent studies that dementia was associated with the likelihoods of continuing statin therapy (aHR, 0.84; 95% CI, 0.73–0.98) and antiplatelet therapy in elderly patients after ischemic stroke [31, 32].

Ds and BBs as initial antihypertensives were negatively associated with treatment persistence, as compared with ARBs, in the very elderly cohort, whereas the rate of non-persistence in the adult cohort was slightly but significantly higher for CCB, AB, and ACEI initiators, as well as BB and D initiators, as compared with ARB initiators. The Eighth Joint National Committee guidelines do not recommend BBs for initial hypertensive therapy due to the relatively high risks of cardiovascular death, myocardial infarction, and stroke with these agents. In addition, the Canadian Institute for Health Information reported that Ds associate with the lowest persistence when newly prescribed in patients over 65 years of age [33, 34]. Marcum et al. analyzed the rate of hospitalization from adverse drug reactions in older veterans and found that BBs and Ds were associated with significantly elevated rates [35]. Thus, the low persistence commonly observed in elderly and very elderly patients when beginning a BB or D may stem from the higher risk of adverse drug reactions from these antihypertensive classes.

The increased risk of adverse drug events and inconsistency in the efficacy associated with changes in pharmacokinetic and pharmacodynamic profiles with aging also may negatively affect drug persistence particularly in the very elderly compared with the elderly patients [36, 37].

Some limitations should be considered when interpreting our findings. First, due to the nature of retrospective claims data, we could not confirm whether the patients actually took the medications that were prescribed. Therefore, our study likely overestimated the actual treatment adherence and persistence. Second, the study only included patients who started monotherapy for uncomplicated hypertension. We excluded patients who began combination therapy in order to minimize confounding by disease severity, which could not be reliably measured in our cohort. Therefore, our results may not be generalizable to patients administered combination therapy. Finally, while we measured the health insurance type as a marker of socioeconomic position, we could not consider other important socioeconomic variables due to limitations of the claims data.

This study showed that very elderly patients were more likely to discontinue antihypertensive treatment when compared with elderly patients; approximately 40% of very elderly uncomplicated hypertensive patients were treatment non-persistent during the first year. In addition, this study suggests that a low comorbidity index and lack of medical aid support negatively affect the treatment persistence uniquely in this population.