Anti-Dementia Drug Persistence Following Donepezil Initiation Among Alzheimer’s Disease Patients in Japan: LIFE Study

INTRODUCTION

As the world’s population continues to grow and age, the prevalence of dementia is projected to surge from 57 million cases in 2019 to 153 million cases by 2050 [1]. Alzheimer’s disease (AD) is the most common form of dementia, and accounts for over 60% of dementia cases [2]. Treatments for AD include pharmacologic and non-pharmacologic therapies, but none of the available options are effective in stopping or reversing the underlying neurodegenerative processes [2]. Aducanumab is a monoclonal antibody aimed at reducing the formation of cerebral amyloid plaques, and received accelerated approval in the US as the first disease-modifying treatment for AD [3]. However, this medication has not been approved in the European Union or Japan. Due to the lack of pharmacologic treatments for AD that target disease progression, there is a strong interest in aducanumab and other disease-modifying therapies.

Donepezil, a cholinesterase inhibitor (ChE-I), is frequently used as a first-line therapy to treat AD symptoms. However, numerous studies from various countries have reported high discontinuation rates, with low tolerability to adverse events cited as a major reason for discontinuation [4–17]. Furthermore, cognitive function can decline quickly after donepezil discontinuation [18], and there are few alternatives that physicians can choose to replace donepezil. Accurately determining the anti-dementia drug persistence rates following donepezil initiation in AD patients and identifying the medications prescribed after discontinuation may inform the development and improvement of treatment strategies, including those for disease-modifying therapies that are currently being developed.

Despite Japan’s unprecedented population aging and rising prevalence of AD, there is little information on its use of anti-dementia drugs. Therefore, this study was conducted to determine anti-dementia drug persistence following donepezil initiation and subsequent drug switching patterns among AD patients in Japan using insurance claims data.

METHODS

RESULTS

Figure 1 shows a flowchart of patient selection for this study. We first identified 57,288 patients who had been prescribed donepezil in the LIFE Study database. Among these, 52,814 patients had a new prescription of donepezil after receiving a diagnosis of AD. After applying the exclusion criteria, we obtained a sample of 20,474 patients for the final analysis.

Fig. 1

Flowchart of patient selection.

The patient characteristics are summarized in Table 1. The mean age (standard deviation) was 82.2 (6.3) years, and 65.7% were women. Cerebrovascular disease was the most prevalent (14.6%) comorbidity, and 3.6% of patients were institutionalized in an LTC facility. In addition, the majority of patients had other drug prescriptions. Among the LTC needs categories, 62.5% of the patients were independent, 10.7% had support needs, 19.3% had low LTC needs, and 7.6% had moderate-to-high LTC needs. The patients were followed-up for a mean duration of 483 days.

Figure 2A shows the Kaplan– Meier curve of anti-dementia drug persistence in all 20,474 study patients. The persistence rates were 89.1% at 30 days, 79.4% at 90 days, 72.6% at 180 days, 64.5% at 360 days, and 58.3% at 540 days after the index prescription date. Figure 2 also shows the Kaplan– Meier curves of anti-dementia drug persistence according to sex, LTC needs categories, LTC facility institutionalization status, number of Charlson comorbidities, and number of other drug prescriptions. From the Cox proportional hazards analysis (Table 2), women had a hazard ratio (95% confidence interval) for discontinuation of 0.99 (0.95–1.03) relative to men after adjusting for age. Using “independent” as the reference category, the hazard ratio (95% confidence interval) for discontinuation was 1.01 (0.94–1.09) for patients with support needs, 1.13 (1.07–1.19) for patients with low LTC needs, and 1.30 (1.20–1.41) for patients with moderate-to-high LTC needs. In addition, LTC facility institutionalization was significantly associated with higher donepezil discontinuation (hazard ratio: 1.12; p = 0.049), whereas other drug prescriptions were significantly associated with lower donepezil discontinuation (p < 0.05). For the subgroup analysis, we excluded patients with moderate-to-high LTC needs; this left a sample of 18,924 patients (mean age: 82.0 years; women: 65.4%). The persistence rates in this subgroup were 89.4% at 30 days, 79.9% at 90 days, 73.2% at 180 days, 65.2% at 360 days, and 59.0% at 540 days after the index prescription date.

Fig. 2

Kaplan– Meier curves of anti-dementia drug persistence following donepezil initiation according to patient characteristics.

Among the 20,474 study patients, we identified 987 patients (4.8%) who had also used galantamine, 1089 patients (5.3%) who had also used rivastigmine, and 3545 patients (17.3%) who had also used memantine. Figure 3 shows the patient distributions by duration from the start date of the other anti-dementia drugs relative to the donepezil discontinuation date. Only 43 patients (0.2%) were prescribed galantamine and 53 patients (0.3%) were prescribed rivastigmine 61 days or earlier before donepezil discontinuation. Next, 590 patients (2.9%) were prescribed galantamine and 627 patients (3.1%) were prescribed rivastigmine within 60 days before or after donepezil discontinuation. Finally, 354 (1.7%) patients were prescribed galantamine and 409 patients (2.0%) were prescribed rivastigmine 61 days or later after donepezil discontinuation. In contrast, memantine was used more frequently: 1267 patients (6.2%) were prescribed memantine 61 days or earlier before donepezil discontinuation, 1325 patients (6.5%) were prescribed memantine within 60 days before or after donepezil discontinuation, and 953 patients (4.7%) were prescribed memantine 61 days or later after donepezil discontinuation.

Fig. 3

Patient distributions by duration from the start of other anti-dementia drugs relative to donepezil discontinuation.

In the sensitivity analysis, we analyzed donepezil persistence as a gap of ≤60 days between donepezil prescription episodes. Here, we calculated the persistence rates to be 86.8% at 30 days, 75.0% at 90 days, 67.4% at 180 days, 58.4% at 360 days, and 51.9% at 540 days after the index prescription date.

DISCUSSION

This large-scale medium-term retrospective cohort study examined anti-dementia drug persistence following donepezil initiation using insurance claims data from over 20,000 AD patients in Japan. This is the first study from Japan that describes persistence among AD patients who received new prescriptions of donepezil. Furthermore, this is, to our knowledge, the first study to characterize anti-dementia drug persistence according to care needs levels. Our results showed that Japanese AD patients have low anti-dementia drug persistence accompanied by very low switching rates to galantamine and rivastigmine.

Several studies from various countries have previously reported low persistence with donepezil and other anti-dementia drugs. Fisher et al. followed 33,793 new ChE-I users in British Columbia, Canada, and found that 49% had discontinued ChE-I use within a year after initiation [8]. Next, Herrmann et al. conducted a population-based cohort study on 28,961 new ChE-I users in Ontario, Canada, and reported that 55.9% had discontinued ChE-I use by the end of follow-up (mean duration of 823 days for community-dwelling patients) [9]. Brewer et al. analyzed 20,729 anti-dementia drug users in Ireland and reported that 30.1% and 43.8% of patients had discontinued their medications after 6 months and 12 months, respectively [10]. Similarly, Kostev et al. estimated that 42.2% of donepezil users in Poland had discontinued its use after 12 months [6]. In Japan, Kitamura et al. analyzed 8,025 donepezil users in a specific-cohort post-marketing observational study and found the persistence rate to be 61.2% at 12 months [5]. Our study estimated the 360-day persistence rate to be 64.5%, which was generally similar to those reported in the studies above. The small variations in the reported persistence rates may be influenced by study-based or country-based differences, such as follow-up durations, anti-dementia drug prices, and out-of-pocket paymentrates.

In this study, the patients demonstrated a rapid discontinuation of anti-dementia drugs by 6 months, with a more gradual discontinuation thereafter. This could be indicative of a subset of patients who discontinued their medications due to the occurrence of adverse events (e.g., nausea and diarrhea) soon after initiation. This is supported by the findings of 2 open-label multicenter studies, which reported that the majority of adverse events occur within 24 weeks (approximately 6 months) of donepezil initiation [20, 21]. The steeper curve observed in the initial 90 days may be due to a higher proportion of adverse events occurring during that period. The subsequent discontinuation rate was more gradual, which may have been the result of a reduction in adverse events after 6 months. This tapering off could also have been influenced by a variety of factors, including patients not feeling satisfied with the drug’s effectiveness after several months, progressive onset of behavioral and psychological symptoms of dementia, increased difficulties in taking medications due to worsening dysphagia, and the emergence of physical problems (e.g., fractures or pneumonia). The decision to discontinue donepezil may have been made by the patients, physicians, or caregivers.

Anti-dementia drug persistence rates have been associated with sex [6, 10–17], region [11, 14], and physicians’ specialties [6, 15]. Our study did not identify any sex-related differences in persistence, and the results of previous studies are conflicting. Several studies have found no sex-related differences [10, 11, 13, 17], whereas others have reported higher discontinuation rates in women [6, 12, 14, 15]. Studies have also produced varied findings on the relationship between anti-dementia drug persistence and dementia severity, with studies reporting contradictory results on the association between discontinuation and Mini-Mental State Examination scores [14, 17]. Another study found an association between donepezil discontinuation and Clinical Dementia Rating scores [16]. Gardette et al. identified Mini-Mental State Examination scores to be a predictor of anti-dementia drug switching [17]. Although our study did not have access to information on the degree of cognitive impairment, we were able to analyze LTC needs levels, which incorporate disease severity as a fundamental component. Our results showed that higher LTC needs levels were positively associated with anti-dementia drug discontinuation, which could be explained by 2 different scenarios. In the first scenario, a patient may have been initially certified with LTC needs due to physical impairment, and subsequently developed AD at a later date. These patients may have problems (e.g., polypharmacy, dysphagia, and worsening of general physical status) in persisting with anti-dementia treatment due to their high LTC needs. In the second scenario, a patient may have developed progressive dementia that increased his/her LTC needs levels after donepezil initiation. Although donepezil is thought to have beneficial effects for mild-to-moderate and severe AD for some time after initiation, these effects eventually decline [22, 23]. This could explain the increase in discontinuation as patients and their families begin to feel that the medication is no longer effective. Furthermore, patients may have a reduced understanding of their circumstances as AD progresses and refuse to take their medications.

A comprehensive assessment of drug persistence requires access to all of a patient’s prescription-related data from health facilities, which can be acquired from insurance claims data. Therefore, we conducted our study using a LIFE Study database comprising claims data. However, the use of claims data gives rise to several limitations. First, the analysis was based on prescription data, and we could not ascertain the actual intake of anti-dementia drugs by the patients. However, this limitation is shared by the vast majority of pharmacoepidemiological studies on community-dwelling patients. Second, we could not directly assess AD severity. Access to the scores from dementia severity measures (e.g., Clinical Dementia Rating and Mini-Mental State Examination) may shed light on the clinical indications for pharmaceutical therapy and contribute to improvements in AD treatment. For example, Umegaki et al. reported that Clinical Dementia Rating scores of ≥3 were associated with higher donepezil discontinuation rates [16]. Although our analysis showed that LTC needs levels of ≥3 were associated with higher discontinuation rates, the lack of dementia severity measures in our data precluded a more in-depth exploration of this relationship. Nevertheless, our study population comprised patients with new prescriptions of donepezil, and the majority are likely to have similar levels of dementia severity at baseline. Third, we were unable to account for variations in drug doses. In Japan, donepezil prescriptions generally begin at a dose of 3 mg/day for 1–2 weeks, which is then increased to 5 mg/day if no adverse effects are observed. For patients with advanced dementia, the dose may be further increased to 10 mg/day after a month. High doses of donepezil may be associated with higher discontinuation rates [22, 24]. As these prescription patterns in Japan can change at the weekly level based on physicians’ assessments of their patients, it would be difficult to analyze and interpret the variations in doses. Finally, we could not account for variations in the patients’ education levels, which may be associated with drug adherence.

Despite these limitations, our study provides important insight into anti-dementia drug persistence among AD patients in Japan, which is struggling with rapid population aging and increasing dementia prevalence. Our study also showed that the vast majority of AD patients who had discontinued donepezil were not switched to another anti-dementia drug, highlighting the current lack of pharmaceutical options for AD in Japan.