Impact of donepezil hydrochloride on the care burden of family caregivers of patients with Alzheimer's disease


Professor Manabu Ikeda MD, PhD, Department of Psychiatry and Neuropathobiology, Faculty of Medical and Pharmaceutical Sciences, Kumamoto University, 1-1-1, Honjo, Kumamoto 860-8556, Japan. Email:


Background:  To evaluate the impact of donepezil hydrochloride on the care burden on family members of patients with Alzheimer's disease (AD). At present, donepezil is the only drug approved for the treatment of AD in Japan. Although the care burden on primary caregivers of AD patients comprises both physical and psychological burdens and donepezil is recognized to improve cognitive dysfunction and associated symptoms, there are few data on the effects of the drug on the care burden.

Methods:  Of the uninstitutionalized AD patients who visited a dementia clinic between June 2008 and May 2009 with their primary family caregivers, 416 subjects who satisfied the enrollment criteria were registered for the study. All participants provided informed consent. Assessment included changes in scores on the Japanese version of the Zarit Caregiver Burden Interview (J-ZBI) and the Mini-Mental State Examination (MMSE), as well as the presence of behavioral and psychological symptoms of dementia (BPSD). Caregivers answered the questionnaires at baseline and after 12 weeks treatment with donepezil (starting dose 3 mg, p.o., once daily, followed by 5 mg after 1 or 2 weeks).

Results:  There were significant changes in mean scores on the J-ZBI (−1.9 ± 9.5; < 0.01) and MMSE (+0.9 ± 2.9; < 0.01) from baseline to Week 12, without significant correlation between these two scores. In patients with BPSD, there was a significant decrease in J-ZBI scores over the 12 weeks (P = 0.013); in contrast, in patients without BPSD, the decrease in the J-ZBI score did not reach statistical significance (P = 0.418).

Conclusions:  The results indicate that donepezil improves cognitive function and some of the BPSD. As a possible consequence of improvements in BPSD, donepezil may also reduce caregivers' burden.


Alzheimer's disease (AD) is a progressive dementia characterized by cognitive dysfunction and associated symptoms including behavioral and psychological symptoms of dementia (BPSD). Even though the pathogenesis of BPSD is not yet fully understood,1 clinicians are often expected to reduce the frequency and severity of BPSD by whatever means available; however, this is sometimes quite difficult.2–5 Therefore, BPSD may increase the burden on patients' family and caregivers, and may be the main reason for visits to the hospital and/or clinic by patients and their families.6

Alzheimer's disease not only affects the quality of life (QOL) of patients, but is also considered a considerable burden by caregivers. Zarit et al.7 reported that caregivers of AD patients are called hidden victims of AD and that the impact of AD on the caregivers socially is a significant issue.

Donepezil, an acetylcholinesterase inhibitor, was approved for manufactur in Japan in October 1999 as the first AD drug with an indication for the suppression of progressive dementia symptoms in mild to moderate AD. In August 2007, the indication for donepezil was expanded to the treatment of symptoms in severe AD. It has been recently reported that 10 mg/day donepezil is effective for the treatment of advanced Alzheimer's disease after prolonged treatment at 5 mg/day.8 There are several papers reporting the clinical efficacy of denepezil in Japanese patients.9,10

Nonetheless, because premarketing clinical studies conducted in Japan were focused only on the cognitive effects of donepezil, they did not address the QOL of patients and their caregivers. To our knowledge, only one post-marketing trial has been conducted to investigate the effects of donepezil on BPSD and caregivers' burden.11 However, that study focused on only a few BPSD and did not use a formal assessment scale to establish the caregivers' burden. Thus, we conducted the present post-marketing survey in a general clinical practice setting to investigate the impact of donepezil on the care burden with a focus on the QOL of caregivers.



Outpatients diagnosed with AD and their primary family caregivers who could answer questionnaires throughout the assessment period were evaluated in the present study. The severity of AD was based on the Functional Assessment Staging of Alzheimer's disease (FAST) and the onset of AD (length of illness) was determined. Patients who had received donepezil before or were hypersensitive to any of its ingredients or piperidine derivatives were excluded from the study.

This prospective study was conducted by central registration and investigators filled out registration forms to enroll eligible patients prior to the initiation of donepezil treatment. Fifty-five medical institutions participated in the study over the period 30 June 2008–31 May 2009 and 44 institutions (12 departments of psychiatry, 17 departments of neurology, five departments of neurosurgery, five departments of internal medicine and five other departments) enrolled a total of 416 patients.


The treatment period was 12 weeks per patient. As listed in the dosage and administration information, 3 mg donepezil was administered orally once daily, followed by 5 mg after 1 or 2 weeks. Investigators were asked to record any changes in the donepezil treatment regimen made during the study period, including upward and downward titrations and discontinuations, start/stop date, dose, and reasons for the change.

Patients were excluded from analysis if the patient's treatment had been suspended for 3 weeks or longer, patients were hospitalized or institutionalized, primary caregivers changed, or questionnaires were not answered. When registering patients for the study, no restrictions were placed on concomitant medication, treatment and home help because the present study was a post-marketing survey in the setting of usual clinical practice.


In the present study, we used the Japanese version of the Zarit Caregiver Burden Interview (J-ZBI)12,13 to measure the care burden of primary family caregivers. The original ZBI, designed by Zarit et al. in 1980,7 is a good tool with which to compare care burdens objectively in different countries. The J-ZBI was arranged by Arai et al. in 1997 for use in Japan.12,13 There are some reports on the impact of donepezil on care burden based on data collected in studies conducted in Japan and abroad that were not clinical trials,11,14–18 yet there are no reports on the effects of donepezil in AD patients as determined with the ZBI, except for one regarding dementia with Lewy Bodies (DLB).18

The J-ZBI scores were determined by asking caregivers to answer questionnaires. The J-ZBI consists of 22 questions and the maximum score is 88 points. The questionnaire is structured to ask about a caregiver's mental and health status, economic burden, social restrictions and relationship to the patient in Questions 1–21 and to quantify the overall care burden or ‘single global burden’ in Question 22. Caregivers choose answers that most closely match their feelings from options of ‘never’, ‘rarely’, ‘sometimes’, ‘quite frequently’ and ‘nearly always’, which correspond to scores of 0, 1, 2, 3 and 4 points, respectively. The points for each question are added together to obtain a final score. Scores were also calculated on the subscales of personal strain, which indicate a burden that arises purely from the caring task, and role strain, which represents the burden that occurred when the caring task disrupts established daily activities.19 The severity of the burden was defined on the basis of the total score20 as follows: major burden = 61–88; moderate burden = 41–60; mild burden = 21–40; and minor burden ≦ 20. Assessments were conducted for each level of burden.

In addition, supervision time (time spent providing care for and keeping an eye on AD patients per day) and free time (time spent away from and the length of time it was possible to stay away from AD patients) were determined. The association between care burden and the time spent supervising AD patients is rarely reported and there is no consensus on the relationship between the two. However, referring to the report of Arai et al.,12,21 who suggested that there was significant relationship between care burden and free time, we did determine the amount of free time caregivers had in the present survey.

Cognitive function was assessed using the Mini-Mental State Examination (MMSE). To evaluate changes in BPSD that may have considerable impact on the care burden, symptoms were examined at baseline with respect to abulia/apathy, depression, delusion, hallucination, anxiety, dependency, wandering, aggressive behavior, resistance to care and irritability. These behavioral disturbances were derived from previous studies11,14 and are mostly included in the Neuropsychiatric Inventory (NPI).22 These symptoms were examined again at Week 12 and the findings were compared with those at baseline and classified as: 1, improved; 2 ,unchanged; or 3, aggravated. Any symptoms that could not be evaluated were noted as such.

In addition, concomitant drug therapy and other treatment, as well as the use of home care services, were investigated as part of routine clinical practice to evaluate factors other than donepezil that may impact on the evaluation of efficacy and care burden. The data collected included demographics, utilization of home health services, medical and family history of dementia, and comobidities. Adverse reactions were assessed along with efficacy.


Data from patients receiving donepezil treatment but who violated the predetermined rules were excluded from safety analysis. Then, data without efficacy parameters was further excluded from the safety analysis set for efficacy analysis.

Efficacy analysis

Changes in care burden (overall J-ZBI scores, personal strain, role strain, supervision time and free time) and MMSE from baseline were assessed at Week 12 using paired t-test. Changes in BPSD were assessed based on the ratio between improvement, no change, and aggravation using 95% confidence intervals. Changes in MMSE and care burden were then applied to calculate Pearson's correlation coefficient to evaluate the direct effect of donepezil on care burden. In addition, changes in care burden were analyzed for each stratum of BPSD (improvement, no change, and aggravation). The significance level of the paired t-test was two-sided 5%, whereas that of stepwise selection was 20%.

Safety analysis

Adverse reactions were calculated according to the Japanese version of the Medical Dictionary for Regulatory Activities (MedDRA/J, v12.0; The incidence, subject, and frequency were calculated for each System Organ Class (SOC) and preferred term (PT).



As shown in Figure 1, safety analysis was performed on data from 398 patients after excluding three patients who were later found to be ineligible and 15 patients who did not return to the site after their first visit. Efficacy analysis was performed on data from 169 patients after removing more data from the safety analysis. The main reasons for the elimination of patients from analysis were discontinuation of treatment in 69 patients, unapproved dose and administration in 43 patients, missing or duplicate data on the J-ZBI for 40 patients and signing up of home help after enrollment for 30 patients. Other patients were excluded from analysis because the MMSE was not performed during the designated time frame or was completely missed: 11 patients were excluded because their baseline J-ZBI was not completed in the specified period or lost and two patients were excluded because their J-ZBI was not collected during the specified period at Week 12. One patient was excluded from analysis because the diagnosis was revised as mild cognitive impairment based on FAST.

Figure 1.

Distribution of subjects throughout the study. MMSE, Mini-Mental State Examination; J-ZBI, Japanese version of the Zarit Caregiver Burden Interview.

Patient and caregiver characteristics

Of the 169 patients, 115 were women (68.0%). Mean patient age was 77.7 ± 6.8 years and the mean disease duration was 2.0 ± 1.7 years. The severity of dementia was minor (FAST 4) in 67.5% of patients, moderate (FAST 5) in 20.1% and severe (FAST 6) in 12.4%. The mean MMSE score at baseline was 18.6 ± 4.6 and 8.9% of patients had received medication for BPSD and had taken antidementia drugs other than donepezil (agents with possible antidementia properties, such as non-steroidal anti-inflammatory drugs, vitamin E, and ginkgo biloba) within the 3 months prior to the study.

Most caregivers were women (60.9%) and the mean age of caregivers was 63.8 ± 14.3 years. The relationship of the caregiver to the patient was spouse in 53.3% of cases, child in 32.0% of cases, daughter-in-law in 14.2% of cases, and another relative in 0.6% of cases. Of the caregivers, 38.5% were employed; 38.5% of carers had been looking after the patient before participating in the study for a mean period of 1.0 ± 1.5 years. Of the carers, 17.8% reported having looked after other patients, whereas 79.9% had not.

Change in care burden

Overall, J-ZBI scores decreased significantly by 1.9 ± 9.5 from a mean (± SD) score of 24.0 ± 15.0 at baseline to a score of 22.1 ± 15.4 at Week 12 (P = 0.009). In particular, there was a significant decrease in the personal strain score of 1.5 ± 5.9 from 14.4 ± 8.7 at baseline to 12.9 ± 9.7 at Week 12 (P = 0.002). Although no significant improvement was observed in role strain, the scores for role strain changed by 0.1 ± 3.0 from 4.6 ± 4.8 to 4.7 ± 4.7 (Table 1).

Table 1.  Change in scores on the Japanese version of the Zarit Caregiver Burden Interview
 No. patientsJ-ZBI scoreChange from baselineP (paired t-test)
BaselineWeek 12
  1. Data show the mean ± SD.

  2. J-ZBI, Japanese version of the Zarit Caregiver Burden Interview.

J-ZBI (sum)16924.0 ± 15.022.1 ± 15.4−1.9 ± 9.50.009
J-ZBI (personal strain)16914.4 ± 8.712.9 ± 9.0−1.5 ± 5.90.002
J-ZBI (role strain)1694.6 ± 4.84.7 ± 4.70.1 ± 3.00.796

There was no significant decrease in the time spent supervising AD patients. Supervision time decreased by 17.2 ± 211.9 min from 307.8 ± 297.4 min at baseline to 291.9 ± 301.9 min at Week 12. In addition, although free time (i.e. the time caregivers spent away from patients) tended to increase by 4.8 ± 155.8 min from 343.0 ± 330.1 min at baseline to 345.3 ± 323.6 min at Week 12, the improvement did not reach statistical significance.

Overall MMSE scores were significantly increased by 0.9 ± 2.9 from a score of 18.6 ± 4.6 at baseline to a score of 19.5 ± 5.0 at Week 12 (< 0.001). There was no significant correlation between J-ZBI and MMSE scores (Fig. 2).

Figure 2.

Relationship between the Japanese version of the Zarit Caregiver Burden Interview (J-ZBI) and the Mini-Mental State Examination (MMSE). There was no significant correlation found between the two (r = −0.015; P < 0.844).

The rates of improvement according to BPSD are given in Table 2. At baseline, 134 of the 169 patients evaluated in the efficacy analysis (79.2%) reported having BPSD, whereas 35 patients (20.7%) did not. Baseline J-ZBI scores were significantly higher in the group of patients with BPSD (P = 0.019) than in those without. In those who had BPSD at baseline, J-ZBI scores were significantly decreased by 2.1 ± 8.8 from a score of 25.4 ± 15.2 at baseline to a score of 23.3 ± 15.6 at Week 12 (P = 0.013). In patients without BPSD, J-ZBI scores tended to decreased by 1.1 ± 8.2 from a score of 18.7 ± 13.5 at baseline to a score of 17.6 ± 13.6 at Week 12, but this differnce did not reach statistical significance (P = 0.418). As indicated in Table 3, for the BPSD that improved by 50% or more after treatment, improvement rates were 63.0% for depression (17/27 patients), 60.0% for delusion (18/30 patients), 83.3% for hallucination (10/12 patients), and 50.0% for anxiety (25/50 patients). The J-ZBI scores tended to decrease more with improvements in each of the BPSD symptoms, except wandering and aggressive behavior (Table 4). In particular, a significant reduction in J-ZBI scores was observed in the group of patients whose delusion improved by 5.6 ± 9.4 (P = 0.022) and in the group whose anxiety improved by 4.8 ± 8.5 (P = 0.009). The J-ZBI scores were also significantly decreased by 4.7 ± 8.1 in patients who reported an improvement in dependency (P = 0.014), but were increased significantly by 15.0 ± 6.9 in patients whose dependency worsened (P = 0.022; Table 4).

Table 2.  Scores of Japanese version of Zarit Caregiver Burden Interview depending on the presence of behavioral and psychological symptoms of dementia
 No. patientsJ-ZBI (total score)Change from baselineP (paired t-test)
BaselineWeek 12
  1. Data show the mean ± SD.

  2. J-ZBI, Japanese version of the Zarit Caregiver Burden Interview; BPSD, behavioral and psychological symptoms of dementia.

Baseline BPSD     
 Yes13425.4 ± 15.223.3 ± 15.6−2.1 ± 9.80.013
 No3518.7 ± 13.517.6 ± 13.6−1.1 ± 8.20.418
Table 3.  Improvements in behavioral and psychological symptoms of dementia
n%95% CIn%95% CIn%95% CI
  1. BPSD, behavioral and psychological symptoms of dementia; CI, confidence interval.

Hallucination1083.3(51.6–97.9)216.7(2.1–48.4)00.0 12
Table 4.  Change in scores on the Japanese version of the Zarit Caregiver Burden Interview depending on the presence of behavioral and psychological symptoms of dementia
BPSDNo. patientsChange from baselineP (paired t-test)
  1. BPSD, behavioral and psychological symptoms of dementia.

Overall169−1.9 ± 9.50.009
 Overall71−0.4 ± 10.30.732
 Improved22−4.1 ± 12.40.135
 Unchanged460.4 ± 8.30.738
 Worsened313.7 ± 8.30.738
 Overall27−2.0 ± 10.00.302
 Improved17−4.8 ± 9.50.053
 Unchanged70.3 ± 6.20.907
 Worsened38.3 ± 15.00.437
 Overall30−2.4 ± 12.40.306
 Improved18−5.6 ± 9.40.022
 Unchanged112.8 ± 15.80.567
 Overall12−1.8 ± 14.20.678
 Improved10−5.5 ± 9.80.110
 Unchanged217.0 ± 22.60.481
 Overall50−2.4 ± 9.10.074
 Improved25−4.8 ± 8.50.009
 Unchanged21−1.7 ± 7.50.321
 Worsened49.3 ± 13.30.258
 Overall68−1.1 ± 10.00.349
 Improved22−4.7 ± 8.10.014
 Unchanged42−0.8 ± 9.70.581
 Worsened415.0 ± 6.90.022
 Overall12−0.9 ± 16.10.848
 Improved5−7.8 ± 8.40.108
 Unchanged414.8 ± 14.30.132
 Worsened3−10.3 ± 15.50.368
 Overall43−1.3 ± 9.30.361
 Improved16−2.1 ± 8.40.328
 Unchanged230.0 ± 10.40.984
 Worsened4−5.8 ± 1.70.007
 Overall31−0.3 ± 10.90.896
 Improved3−8.3 ± 12.30.363
 Unchanged250.3 ± 10.80.884
 Worsened33.0 ± 10.10.660
 Overall46−2.3 ± 8.70.073
 Improved20−2.9 ± 9.00.175
 Unchanged22−2.2 ± 8.50.240
 Worsened4−0.8 ± 10.00.890

Adverse reactions

Table 5 indicates that of the 398 patients included in the safety analysis, 46 adverse reactions were observed in 36 patients. Adverse reactions occurred at a rate of 9.0% (36/398 patients). This rate is comparable to that reported in a previous post-marketing survey conducted in patients with mild to moderate AD conducted by Eisai co., Ltd. in 2005 (10.7%; 346/3240 patients). These data were provided us from the company in 2009. In the present study, the frequently observed adverse reactions included agitation (n = 10 patients; 2.5%), nausea (n = 8 patients; 2.0%), diarrhea (n = 5 patients; 1.3%) and two patients each (0.5%) reporting poor appetite, loss of appetite, aggression, headache and vomiting. Other adverse reactions also observed were anger, coprolalia, depression, insomnia, uneasiness, dizziness, parkinsonism, atrial fibrillation, erosive gastritis, pruritus, back pain and irritation (n = 1 patient each; Table 3). Severe adverse reactions included agitation and atrial fibrillation (n = 1 patient each; 0.3%).

Table 5.  Number of patients reported adverse reactions
Adverse reactionNo. patients%
Safety analysis population (398 subjects)369.0
Metabolic and nutritional disorder41.0
Loss of appetite20.5
Psychological disorder123.0
Neurological disorder30.8
Cardiac disorder10.3
Atrial fibrillation10.3
Gastric and intestinal disorder153.8
Erosive gastritis10.3
Skin and subcutaneous disorder10.3
Musculoskeletal and connective tissue disorder10.3
Back pain10.3
Systemic and local disorder10.3


In the present study, J-ZBI scores were reduced after donepezil treatment for 12 weeks. This reduction can be attributed to the effects of donepezil because patients who changed nursing services, which may have had potential effects on care burden, were excluded from analysis. Thus, the present results suggest that donepezil is effective in reducing caregiver burden in AD.

The present study addressed changes in care burden from baseline to Week 12 using J-ZBI and showed a significant difference in the mean J-ZBI score (−1.9 ± 9.5). This result is in line with that of a study conducted in mild to moderate AD patients with a similar drug, rivastigmine,23 which showed that the mean ZBI score decreased by 1.7 ± 8.6 after 6 months treatment. Although the reduction in J-ZBI score by 2 points seems to be an relatively unimportant improvement, Arai et al. defined any reduction in J-ZBI in subjects classified as having mild AD as a successful.24 Accordingly, because the average care burden was classified as mild in the present study, the change in J-ZBI score observed is considered meaningful for caregivers.

There was no significant correlation between changes in ZBI and MMSE scores. This suggests that a decreased burden is not the direct result of improvements in cognitive function. Conversely, in patients with BPSD, J-ZBI scores were significantly decreased, whereas in patients without BPSD the decrease in J-ZBI scores was not significant. Gort et al. reported that BPSD was associated with care burden after assessing care burden and care collapse in relation to various risk factors using the ZBI.25 In addition, there are many other reports that indicate a correlation between improvements in BPSD and care burden.11,26–28 Although the correlation between improvements in BPSD and decreased ZBI scores was not analyzed directly in the present study, improvements in BPSD may consequently lead to a reduction in a caregiver's burden.

In the subscales of J-ZBI, there was a significant decrease in personal strain (< 0.01), although a similar decrease was not observed in role strain. Role strain is considered to represent burden that occurs when the caring task restricts established daily activities. This finding shows that donepezil could not reduce everyday life restrictions felt by caregivers. Conversely, it is considered that personal strain was improved along with improvements in BPSD because personal strain reflects subjective difficulty.

It is important to acknowledge that there are some limitations to the present post-marketing observational study. First, there were many drop outs compared with randomized control trials (RCT) and there were only 169 patients for whom valid data were available for analysis. Similarly, in a 6 month post-marketing observational study of rivastigmine conducted in Belgium with an original 434 patients,23 valid data for analysis was only available for 175 patients, indicated a comparable drop out rate to that seen in the present study. Therefore, it seems inevitable that there will be considerable drop outs in post-marketing observational studies. Second, many patients and caregivers may have dropped out because the care burden worsened. In other words, the patients may have required hospitalization, institutionalization, or the addition of care services because their condition worsened. In any case, these patients were removed from the final analysis. However, this should not have had a major impact on the results because the number of patients who required hospitalization or institusionalization in the present study was small (five patients). Third, because the J-ZBI is a subjective evaluation, the impact of placebo effects cannot be denied. This issue can only be resolved by performing an RCT. However, unlike the RCT, which is conducted under unusual condition with selected patients and families, the present study was performed to elucidate the impact of donepezil under normal clinical settings. For these reasons, the present results must be interpreted with caution.

In conclusion, the results of the present post-marketing survey suggest that donepezil may improve cognitive function, BPSD, and hence care burden and, thus, can be expected to contribute to a better and lasting QOL of family caregivers.


The authors thank the investigators across the 55 institutions who participated in the present study. The authors also thank Mr Takatsuki for his statistical advice.