Neuropsychiatric symptoms predict change in quality of life of Alzheimer disease patients: A two-year follow-up study

Authors

  • Hiroshi Tatsumi ms ,

    1. Department of Psychiatry and Cognitive–Behavioral Medicine, Nagoya City University Graduate School of Medical Sciences,
    2. Nagoya Daini Red Cross Hospital, Nagoya and
    Search for more papers by this author
  • Shutaro Nakaaki md, phd,

    Corresponding author
    1. Department of Psychiatry and Cognitive–Behavioral Medicine, Nagoya City University Graduate School of Medical Sciences,
    • *Shutaro Nakaaki, MD, PhD, Department of Psychiatry and Cognitive–Behavioral Medicine, Nagoya City University Graduate School of Medical Sciences, Mizuho-cho, Mizuho-ku, Nagoya 467-8601, Japan. Email: hzi05510@nifty.com

    Search for more papers by this author
  • Katsuyoshi Torii md ,

    1. Department of Psychiatry and Cognitive–Behavioral Medicine, Nagoya City University Graduate School of Medical Sciences,
    Search for more papers by this author
  • Yoshihiro Shinagawa md, phd,

    1. Department of Psychiatry and Cognitive–Behavioral Medicine, Nagoya City University Graduate School of Medical Sciences,
    Search for more papers by this author
  • Norio Watanabe md, phd,

    1. Department of Psychiatry and Cognitive–Behavioral Medicine, Nagoya City University Graduate School of Medical Sciences,
    Search for more papers by this author
  • Yoshie Murata ms ,

    1. Department of Psychiatry and Cognitive–Behavioral Medicine, Nagoya City University Graduate School of Medical Sciences,
    Search for more papers by this author
  • Junko Sato ms, phd,

    1. Yagoto Hospital,
    Search for more papers by this author
  • Masaru Mimura md, phd,

    1. Department of Neuropsychiatry, Showa University School of Medicine, Tokyo, Japan
    Search for more papers by this author
  • Toshiaki A. Furukawa md, phd

    1. Department of Psychiatry and Cognitive–Behavioral Medicine, Nagoya City University Graduate School of Medical Sciences,
    Search for more papers by this author

  • Field: general topics in psychiatry and related fields.

Abstract

Aim:  To examine the effect of neuropsychiatric symptoms on longitudinal changes in the quality of life (QOL) of patients with Alzheimer disease (AD).

Methods:  First, we investigated whether neuropsychiatric symptoms at baseline predict changes in the QOL of AD patients over time. Then we examined the associations between changes in neuropsychiatric symptoms and changes in QOL. QOL was assessed using the Japanese version of the Quality of Life–Alzheimer Disease (QOL–AD) scale and other clinical instruments [the Mini-Mental State Examination, The Neuropsychiatry Inventory (NPI)] at baseline and again two years later in 96 AD patients among 140 AD patients at baseline. We performed a multiple regression analysis of the baseline QOL–AD score, NPI score (mood, psychosis, and euphoria factor), Mini-Mental State Examination score, and other clinical instrument variables (e.g. Activities-of-Daily-Living scores) to determine their contribution to the change in QOL–AD score.

Results:  While the total QOL–AD score based on the patients' responses did not change significantly, the total QOL–AD score derived from the caregivers' responses declined. Both the Activities-of-Daily-Living score and the mood factor of the NPI score predicted the change in the QOL–AD score as assessed by the caregivers' responses. In addition, there was a significant correlation between the changes in two factors of the NPI, i.e. the mood and psychosis factor, and the changes in the QOL–AD score based on the caregivers' responses.

Conclusions:  The presence of specific neuropsychiatric symptoms (mood and psychosis symptoms) was associated with changes in the QOL of AD patients during the follow-up period.

NEUROPSYCHIATRIC SYMPTOMS OFTEN accompany the cognitive decline that occurs during the course of Alzheimer disease (AD), and the presence of certain neuropsychiatric symptoms, e.g. delusions, is associated with both functional decline and institutionalization.1 More importantly, there have been cross-sectional studies demonstrating that the presence of neuropsychiatric symptoms influences the quality of life (QOL) of AD patients based on either the patients' responses or the caregivers' responses.2,3

Thus, there is good reason to examine the effects of neuropsychiatric symptoms on the QOL of AD patients over time in a longitudinal study, and yet, as far as we have been able to determine in a thorough search of the published reports, there have been few such studies in dementia patients. Although several studies have reported predictors of QOL outcome in longitudinal studies,4,5 none of them used valid methods of assessing QOL or monitored changes in QOL over time. Three recent longitudinal studies have been conducted using valid methods of assessing QOL developed specifically for patients with dementia,6–8 but they yielded inconsistent results. Lyketsos et al.6 found that the total QOL scores of dementia patients based on proxy reports of Alzheimer's Disease Related Quality of Life (ADRQL)9 slightly but significantly declined over two years, whereas Missotten et al.8 did not observe any significant change in proxy reports of ADRQL in regard to dementia patients over two years. The third study assessed QOL changes using patients' reports of either the Dementia Quality of Life Scale (DQoL)10 or the QOL–AD11 but the investigators did not find any significant changes during a one-year follow-up period.7

Despite growing attention to longitudinal changes in QOL in patients with dementia, several important issues remain unresolved. First, previous longitudinal studies have assessed QOL in dementia patients by using proxy reports or patients' reports. Proxy reports may be affected by the care-burden and the expectations and belief systems of caregivers,2,12 and although knowing patients' subjective perceptions is important when assessing QOL, both cognitive impairment and loss of insight may cause biases in the self-reports of dementia patients.13 Thus, information derived from both patients' reports and proxy reports is necessary for optimal assessment of the QOL of dementia patients. Second, while some studies have included patients with more than one subtype of dementia,6 as others have not stated the etiology of the subjects' dementia,7,8 whether disease-specific measures of QOL deteriorate over time in AD patients has remained unclear. Third, although the results of several cross-sectional studies have suggested that neuropsychiatric symptoms affect the QOL of AD patients,2,3,12 few of them have examined correlations between the progression of a broad spectrum of neuropsychiatric symptoms and QOL changes in AD patients, even though it is important to know which sub-symptoms of the various neuropsychiatric symptoms that sometimes occur in AD affect QOL changes over the long-term course of AD.

To resolve these issues, we examined neuropsychiatric symptoms, cognitive function, and QOL in a relatively large sample of AD patients over the course of two years. Our previous study was a cross-sectional study of QOL in AD patients,2 and the present study is a follow-up study to that cross-sectional study. The aims of the present study were: (i) to examine changes in QOL–AD score over a two-year follow-up period; (ii) to compare the patients' responses and the caregivers' responses at the time of the follow-up examination; (iii) to identify the independent clinical variables at baseline that were the best predictors of QOL change over the two-year follow-up; and (iv) to examine the correlation between the changes in the total QOL–AD score over time and both cognitive changes and changes in neuropsychiatric symptoms.

METHODS

Subjects

The baseline sample consisted of 150 Japanese patients with mild to moderate AD and the caregiver of each of them who attended the outpatient clinic of Nagoya City University Hospital, Nagoya Second Red Cross Hospital, or Yagoto Hospital. The patients and their caregivers were originally assessed between September 2003 and August 2004. The inclusion criteria for the patients were: (i) a diagnosis of probable AD according to the NINCDS/ADRDA criteria;14 and (ii) residence with their caregiver in a community dwelling. The exclusion criteria were: (i) presence of a neurological disease other than AD; (ii) a history of severe mental illness or substance abuse before the onset of dementia; (iii) a magnetic resonance imaging examination that revealed a focal brain lesion; (iv) a Mini-Mental State Examination (MMSE) score15 below 10; and (v) inability to obtain reliable informed consent from the patient and/or a relative. ‘Caregiver’ means the principal family member providing day-to-day care to the patient. Ten AD patients were excluded from the analysis because they withdrew their consent or because they could not understand the questionnaire in spite of having an MMSE score of 10 or more.

A total of 140 AD patients (84 women, 56 men) were recruited into the study, and their mean age was 72.4 years [standard deviation (SD) = 7.3 years]. Their mean number of years of education was 9.63 (SD = 0.82), and their mean MMSE score was 20.3 (SD = 4.2). Based on previously reported criteria,2 108 of the 140 AD patients who were enrolled at baseline were evaluated as having either a high or intermediate cognitive level (mean MMSE score, 21.44 ± 2.1; range, 17–23), and the other 32 AD patients (22.9%) were evaluated as having a low cognitive level (mean MMSE score, 13.9 ± 1.8; range, 11–16). Among the 140 patients who were recruited, 80 AD patients were being treated with donepezil (5 mg/day) and the other 60 had never been treated with donepezil. The 140 caregivers consisted of 86 women and 54 men, and their mean age at the start of the study was 61.7 years (SD = 11.01). The mean number of years of education of the caregivers was 12.4 (SD = 1.2), and 58.6% of them were spouses of the AD patients, 19.3% were daughters-in-law, 17.2% were adult children, and 4.9% were other close relatives.

The study protocol was approved by the Ethics Committee of Nagoya City University Graduate School of Medical Sciences. All of the subjects were informed of the purpose and procedures of the study before providing consent.

Baseline assessment

All of the AD patients were assessed at baseline using the following tests by well-trained psychiatrists and neuropsychologists at our outpatient clinic. The neuropsychologists administered the MMSE and the QOL–AD.

1) Japanese version of the Quality of Life–Alzheimer Disease (QOL–AD) Scale

The QOL–AD scale consists of 13 items, and they cover: aspects of physical health, energy level, moods, living situation, memory, family, marriage, friends, self as a whole, ability to do chores around the house, ability to do things for fun, money, and life as a whole.11 There are two versions of the QOL–AD scale: a patient version and a caregiver version; the caregiver version is filled out by the caregiver who is unaware of the patient's answers. Each item is graded using a 4-grade Likert-style scale: poor, fair, good, excellent. A total score is calculated separately for the patient version and the caregiver version, with possible scores on each ranging from 13 to 52. The reliability and validity of the Japanese version of this scale was confirmed by Matsui et al.2 In accordance with the original procedure, the AD patients and their caregivers completed their versions of the questionnaire independently in different rooms in the outpatient clinic. The examiners gave no special suggestions to either the patients or caregivers, except for giving them instructions for completing the QOL–AD questionnaire.

2) Neuropsychiatric inventory

The neuropsychiatric inventory (NPI) is a semi-quantitative assessment based on information provided by the caregiver.16 Hirono et al.17 confirmed the reliability and validity of the Japanese version of this questionnaire.

3) Hyogo Activities-of-Daily-Living Scale (HADL)18

The Hyogo Activities-of-Daily-Living Scale (HADL) is a detailed, comprehensive assessment of the functional abilities of AD patients.18 The reliability and validity of this questionnaire were confirmed by Hirono et al.19

4) Short Memory Questionnaire (SMQ)

The Short Memory Questionnaire (SMQ) is widely used to assess memory disturbances in AD patients.20 It was originally developed as an objective tool to assess memory deficits of AD patients based on information given by the caregiver. It was developed by Koss et al.20 as a reliable, valid, and convenient index of memory disturbances in AD patients. Maki et al.21 confirmed the reliability and validity of the Japanese version of this questionnaire, which consists of 14 questions relating to everyday problems of the patient as assessed by the caregiver.

5) Mini-Mental State Examination15

Follow-up study

The follow-up assessment was conducted between 2005 and 2006, approximately two years after the baseline assessment, and one of the authors (a neuropsychologist), who was blinded to the data obtained during the baseline assessment, administered the follow-up QOL–AD scale to all the patients and caregivers. In addition, we administered the MMSE and NPI again at the time of the two-year follow-up assessment.

Analysis

We used SPSS 11.0J software for Windows to perform the statistical analyses. First, we used the unpaired t-test to identify significant differences in clinical factors between the follow-up group and the lost-to-follow-up group. Second, we compared the total QOL–AD scores and the domain scores between baseline and follow up on the patients' version and the caregivers' version. Third, we assessed correlations between the patients' responses and caregivers' responses in terms of total QOL–AD scores and domain scores at follow up. Lastly, we calculated the Pearson correlation coefficients for change in total score of the QOL–AD over the 2-year period based on the patients' responses and the caregivers' responses and other variables (patient gender, patient age, patient education, MMSE score, SMQ score, HADL score, and scores for the three factors, i.e. psychosis factor, mood factor, and euphoria factor, of the NPI) at baseline. In our previous study,2 we demonstrated the presence of three behavior sub-syndromes (psychosis factor, mood factor, and euphoria factor) of the NPI. The change in total QOL–AD scale was calculated by subtracting the total score at baseline from the total score at follow up. We then performed a multiple linear regression analysis to identify the variables at baseline that were good predictors of the change in the total QOL–AD score. The change in total QOL–AD scores based on the patients' responses and the caregivers' responses were used as the dependent variables in the regressions. When statistically significant associations regarding the independent variables at baseline were noted at a P-value of 0.05, they were entered into a multiple linear regression analysis as predictor variables. Multi-collinearity was assessed by means of this multiple regression analysis. Lastly, we calculated Pearson correlation coefficients for the relationships between the changes in total QOL–AD score over the two-year period based on the patients' responses and the caregivers' responses and the change in the MMSE score and the changes in three factors (psychosis factor, mood factor, and euphoria factor) of the NPI between baseline and follow up. The alpha level was set at 0.05.

RESULTS

Clinical characteristics of the lost-to-follow-up group

At the two-year follow up, only 96 (68.6%) of the 140 patients at baseline were assessed by means of a second QOL–AD, because the other 44 AD patients (31.4%) had been lost to follow up: 24 because of transfer to another hospital, eight because they had been institutionalized, two because they had died, and ten because contact had been lost during the follow-up period.

Table 1 shows the clinical differences at baseline between the follow-up group and the lost-to-follow-up group. The follow-up group had a higher baseline SMQ score, baseline MMSE score, and baseline total QOL–AD score based on the patients' responses than the lost-to-follow-up group. There were no significant differences in demographic data, neuropsychiatric symptoms, or Activities-of-Daily-Living score between the two groups.

Table 1. Clinical comparisons between the follow-up group and the lost-to-follow-up group
BaselineFollow-up group (n = 96)Lost-to-follow-up group (n = 44)Comparison
  1. HADLS, Hyogo Activities-of-Daily-Living Scale; MMSE, Mini-Mental State Examination; NPI, Neuropsychiatric Inventory; QOL–AD, Quality of Life–Alzheimer Disease scale; SMQ, The Short Memory Questionnaire (SMQ).

Age (years, mean)71.8 ± 8.071.5 ± 6.2 P = 0.82
Gender (% female)62.554.5 P = 0.37
Education (years, mean)9.8 ± 0.79.4 ± 0.8 P = 0.16
MMSE (mean)21.7 ± 2.717.4 ± 5.2 P < 0.001
SMQ (mean)23.1 ± 6.018.5 ± 5.2 P < 0.001
NPI (total, mean)18.9 ± 13.919.4 ± 13.6 P = 0.85
HADLS (mean)27.3 ± 7.029.6 ± 7.75 P = 0.08
QOL–AD (total, mean)   
 Patients' ratings27.3 ± 4.425.6 ± 4.2 P = 0.036
 Caregivers' rating24.5 ± 4.724.8 ± 4.8 P = 6.90

Changes in the total QOL–AD score during the two-year follow-up period

As shown in Table 2, the total QOL–AD score based on the patients' responses did not change significantly between baseline and follow up. While the scores on several subscales of the QOL–AD (i.e. living situation, memory, family) decreased significantly, the scores on other QOL–AD subscales did not change significantly. By contrast, both the total QOL–AD score and the QOL–AD subscale scores based on the caregivers' responses decreased significantly.

Table 2. Comparisons between the baseline and follow-up Quality of Life–Alzheimer Disease scale scores (total score, subscale scores) based on the patients' responses
 Patients' responseCaregivers' responses
Baseline scoreFollow-up scoreComparisonBaseline scoreFollow-up scoreComparison
1 Physical health2.24 ± 0.622.18 ± 0.64 P = 0.4562.17 ± 0.621.99 ± 0.58 P < 0.001
2 Energy1.94 ± 0.672.02 ± 0.69 P = 0.4071.89 ± 0.731.32 ± 0.49 P < 0.001
3 Mood1.93 ± 0.661.88 ± 0.54 P = 0.5381.89 ± 0.671.22 ± 0.46 P < 0.001
4 Living situation2.21 ± 0.592.00 ± 0.54 P = 0.0032.2 ± 0.591.64 ± 0.60 P < 0.001
5 Memory2.14 ± 0.691.90 ± 0.67 P = 0.0021.9 ± 0.551.21 ± 0.43 P < 0.001
6 Family2.41 ± 0.822.16 ± 0.68 P = 0.0072.2 ± 0.771.76 ± 0.53 P < 0.001
7 Marriage2.25 ± 0.692.02 ± 0.35 P = 0.0032.25 ± 0.531.96 ± 0.71 P < 0.001
8 Friends2.23 ± 0.681.91 ± 0.46 P < 0.0012.23 ± 0.751.92 ± 0.57 P < 0.001
9 Self2.16 ± 0.541.97 ± 0.33 P = 0.0022.07 ± 0.581.69 ± 0.52 P < 0.001
10 Ability to do chores2.16 ± 0.622.25 ± 0.66 P = 0.2952.03 ± 0.681.47 ± 0.58 P < 0.001
11 Ability to do things for fun1.82 ± 0.631.91 ± 0.05 P = 0.2081.73 ± 0.681.39 ± 0.56 P < 0.001
12 Money2.25 ± 0.591.95 ± 0.30 P < 0.0012.02 ± 0.641.92 ± 0.55 P = 0.06
13 Life as a whole2.04 ± 0.561.91 ± 0.48 P = 0.062.05 ± 0.631.33 ± 0.49 P < 0.001
Total score25.63 ± 4.2926.36 ± 2.47 P = 0.1524.08 ± 4.8820.68 ± 4.06 P < 0.001

Correlation between the patients' responses and the caregivers' responses

At baseline, there was a significant agreement between the total scores of the QOL–AD based on the patients' and caregivers' responses (r = 0.647, P < 0.001). However, the correlation between the QOL–AD scale scores based on the patients' and caregivers' responses in regard to the 32 AD patients (22.9%) who were considered to have a low cognitive level (MMSE: range 11–16) at baseline was not significant (r = 0.20, P = 0.24). With regard to the subscales, there was moderate agreement between the patients' and caregivers' responses.2 However, two years after the baseline assessment there were no significant correlations between total QOL–AD scores based on the patients' and caregivers' responses (r = 0.159, P = 0.121) or the scores for any of the subscales of the QOL–AD (Table 3). At follow up, significant differences were found between the total scores on the QOL–AD according to the patients' and caregivers' responses (P < 0.001, t-test). The total QOL–AD score, based on the patients' responses (26.36 ± 2.47), was significantly higher than the total score based on the caregivers' responses (20.68 ± 4.06) (P < 0.001).

Table 3. Correlations between patients' and caregivers' responses using the Quality of Life–Alzheimer Disease scale at follow up
ItemFollow-up score (Patients' responses)Follow-up score (Caregivers' responses)Correlation
  1. Pearson correlations are given for all items and the total score.

  2. None of the correlations are significant.

1 Physical health2.18 ± 0.641.99 ± 0.580.032
2 Energy2.02 ± 0.691.32 ± 0.490.134
3 Mood1.88 ± 0.541.22 ± 0.460.15
4 Living situation2.00 ± 0.541.64 ± 0.600.161
5 Memory1.90 ± 0.671.21 ± 0.430.148
6 Family2.16 ± 0.681.76 ± 0.53−0.069
7 Marriage2.02 ± 0.351.96 ± 0.710.06
8 Friends1.91 ± 0.461.92 ± 0.570.179
9 Self1.97 ± 0.331.69 ± 0.52−0.055
10 Ability to do chores2.25 ± 0.661.47 ± 0.580.102
11  Ability to do things for fun1.91 ± 0.051.39 ± 0.560.158
12 Money1.95 ± 0.301.92 ± 0.550.099
13 Life as a whole1.91 ± 0.481.33 ± 0.490.132
Total score26.36 ± 2.4720.68 ± 4.060.159

Multiple linear regression analyses for prediction of changes in QOL–AD score

We regarded the change in total QOL–AD score over time based on the caregivers' responses to be a more accurate indicator of this longitudinal study of the QOL changes in the AD patients, because of the disagreement we observed between the QOL scores based on the patients' and caregivers' responses at follow up. In addition, the total QOL–AD score based on the caregivers' responses decreased significantly during the follow-up period, whereas the total QOL–AD score based on the patients' responses did not change significantly.

Table 4 shows significant correlations between the total change score of the QOL–AD (caregivers' responses) and at baseline, patient age, the HADL score, and both the psychosis factor and the mood factor of the NPI. Also, significant correlations between the total change score of the QOL–AD (patients' responses) and three factors of the NPI at baseline were observed. A multiple linear regression analysis was performed using the changes in total QOL–AD scores based on the caregivers' responses and patients' responses as dependent variables, and when statistically significant associations at a P-value of 0.05 were identified, they were entered into a multiple linear regression analysis. We therefore entered patient age, HADL score, and psychosis factor, mood factor of the NPI as independent variables at baseline into a multiple linear regression analysis of the change in total score on the QOL–AD based on the caregivers' responses. We also entered three factors of the NPI, i.e. the psychosis factor, the mood factor, and the euphoria factor, at baseline as independent variables into the multiple linear regression analysis of the changes in total score on the QOL–AD based on the patients' responses. As shown in Table 5, both the HADL score and the mood factor of the NPI at baseline were significant predictors of the change in total score on the QOL–AD (caregivers' responses), and the psychosis factor of the NPI at baseline was a significant predictor of the change in total score on the QOL–AD based on the patients' responses. An assessment of collinearity in these multiple regression analyses revealed that no components with two or more variables had a large variance (0.50 or greater), corresponding to large condition indices for the patients' and caregivers' responses. Thus, there was no evidence of multi-collinearity in these analyses.

Table 4. Correlations between the changes in the QOL–AD scores (patients'/caregivers' responses) and other variables at baseline
 QOL–AD total change score (Patients' responses)QOL–AD total change score (Caregivers' responses)
  • *

    P < 0.05,

  • **

    P < 0.01.

  • Patient age, patient education, MMSE, SMQ, HADL, and NPI were assessed at baseline. HADLS, Hyogo Activities-of-Daily-Living Scale; MMSE, Mini-Mental State Examination; NPI, Neuropsychiatric Inventory; QOL–AD, Quality of Life–Alzheimer Disease scale; SMQ, The Short Memory Questionnaire (SMQ).

Patient age−0.120−0.207*
Patient education−0.106−0.021
MMSE−0.1370.187
SMQ−0.0270.035
HADLS0.061−0.319**
NPI  
 Mood factor−0.266**−0.325**
 Psychosis factor−0.341**−0.205*
 Euphoria factor−0.221*−0.101
Table 5. Variables at baseline predicting the overall change in the total QOL–AD score as assessed using the AD patients' responses and the caregivers' responses
Variablesβt P
(a) Variables at baseline predicting the overall change in the total QOL–AD score as assessed using the AD patients' responses
NPI   
 Mood factor−0.180−1.8050.074
 Psychosis factor−0.257−2.5420.013*
 Euphoria factor−0.178−1.8530.067
Variablesβt P
(b) Variables at baseline predicting the overall change in the total QOL–AD score as assessed using the caregivers' responses
  • *

    P < 0.05,

  • **

    P < 0.01.

  • Patient age, MMSE, SMQ, HADL, and NPI were assessed as independent variables at baseline.

  • (a) adjusted R2 = 0.50 (P < 0.000); this explained 50% of the total variance of the dependent variable.

  • (b) adjusted R2 = 0.52 (P < 0.000); this explained 52% of the total variance of the dependent variable.

  • Therefore, the adjusted R2 values in both (a) and (b) are considered high.

  • β, standardized regression coefficient; HADLS, Hyogo Activities-of-Daily-Living Scale; MMSE, Mini-Mental State Examination; NPI, Neuropsychiatric Inventory; QOL–AD, Quality of Life–Alzheimer Disease scale; SMQ, The Short Memory Questionnaire (SMQ).

Patient age−0.079−0.7910.431
HADLS−0.268−2.7070.008**
NPI   
 Mood factor−0.256−2.5870.011**
 Psychosis factor−0.110−1.0960.276

Correlations between changes in total QOL–AD scores over time and the changes in MMSE scores and changes in NPI scores

The total MMSE score declined significantly during the two-year follow-up period [baseline MMSE score, 21.71 ± 2.7; follow-up MMSE score, 17.86 ± 3.3 (P < 0.001)] and the scores for three factors of the NPI (psychosis factor, mood factor, and euphoria factor) increased significantly. No significant correlations were observed between the changes in total score on the QOL–AD over time based on the patients' responses or the changes in total MMSE score and NPI score, or between the change in total score on the QOL–AD over time based on the caregivers' responses and the change in total MMSE score. However, significant correlations were found between the change in total score on the QOL–AD over time based on the caregivers' responses and the changes in and the scores for the psychosis factor (r = −0.333, P = 0.001) and mood factor (r = −0.250, P = 0.014) of the NPI. Assessment of the effect of donepezil on the change in QOL–AD score based on the patients' responses and the caregivers' responses showed no significant differences between the changes in MMSE score or changes in NPI score between the AD patients treated with donepezil (n = 62) and without donepezil (n = 34).

DISCUSSION

The total QOL–AD scores based on the patients' responses did not change significantly during the follow-up period of this study, but the total QOL–AD scores based on the caregivers' responses decreased significantly, and there was disagreement between the patients' and the caregivers' responses at follow up. We therefore regarded the changes in QOL–AD score based on the caregivers' responses to be a more accurate index of QOL–AD than the changes based on the patients' responses. The main findings of the present study were that both the HADL score and the mood factor score of the NPI predicted the change in the QOL–AD score based on the caregivers' responses. In addition, significant correlations were found between the changes in the mood factor score and psychosis factor score of the NPI and the change in total QOL–AD score derived from the caregivers' responses. However, the MMSE score change was not correlated with the QOL–AD score changes derived from the caregivers' responses.

Consistent with the results of a previous study,7 we did not observe a clinically significant change in the QOL–AD score based on the patients' responses. In contrast to the results of our study at baseline,2 no significant correlations were observed between the patients' responses and the caregivers' responses at follow up and in either the total QOL–AD score or the scores for any of the items (baseline: r = 0.60, P < 0.01 vs follow up: r = 0.15, P = 0.12). In addition, at follow up the patients' self-estimates of their QOL scores were significantly higher than the caregivers' estimates of the patients' QOL. Logsdon et al.11,12 stated that cognitive function did not seem to be the most important factor affecting the level of agreement between patients' and their caregivers' responses. Although the QOL–AD was described as a brief and easily administered instrument for reliably assessing self-reported QOL in dementia patients, despite the presence of cognitive impairment, in the original study,11,12 the patients' and the caregivers' responses at the follow-up assessment did not agree well in our study. Moreover, despite the significant correlation between the patients' and caregivers' responses in regard to the total QOL–AD scores at baseline, the correlation between the total QOL–AD scores based on the patients' and caregivers' responses for the 32 AD patients (22.9%) who were considered to have a low cognitive level (MMSE range, 11–16) at baseline was not significant (r = 0.20, P = 0.24). One possible explanation for the lack of a significant correlation is that insight is an important factor in determining the validity of QOL self-assessments by AD patients, and the AD patients may have overestimated their ability about QOL because of impaired insight. Vogel et al.22 suggested that the impaired insight of AD patients had a significant negative impact on the correlation between patients' and caregivers' ratings of QOL in AD patients, and several studies have provided evidence that the impairment becomes more pronounced, independently of changes in other domains, such as memory, attention, or mood, as the disease progresses.23,24 Thus, the loss of insight may contribute to inappropriate perceptions of QOL by AD patients with advanced dementia because the QOL–AD questionnaire includes a wide range of domains (interpersonal, environment, functional, physical and psychological status). Another possible explanation is that, as Logsdon et al. suggested, the discrepancy between the patients' and the caregivers' responses may reflect a real difference in the way the caregivers perceive the patients' QOL rather than a difference in the severity of the cognitive impairment of the patient. Caregivers may underestimate the patients' QOL if the psychological status of the patients, such as depression, leads to a negative bias. Unfortunately, as we did not assess either the patients' insight or the caregivers' psychological status, we cannot provide a clear explanation for the discrepancy between the patients' and the caregivers' responses in the QOL–AD scale either in relation to the AD patients with a low cognitive level at baseline or at follow up. It is important to bear in mind that having a low cognitive level may cause patients' responses to the QOL–AD to differ from their caregivers' responses.

There were significant decreases in the scores for almost all of the domains of the QOL–AD based on the caregivers' responses during the two-year follow-up period. Although Lyketsos et al.6 reported that caregiver-reported QOL assessed using the ADROL declined, the change was small, and the ADROL scores of about half of the dementia patients either did not change or improved. Therefore, this finding (the decreases in QOL–AD scores based on assessments by caregivers) may provide important evidence regarding longitudinal changes in the QOL of AD patients as assessed by their caregivers. Two explanations may help to interpret this finding. One explanation is that the decreases in QOL–AD scores based on assessments by caregivers may reflect a true deterioration in the QOL of their AD patients, and the other is that the biases of proxy ratings, such as the caregivers' burden or depression, may lead to decreases in QOL–AD scores because the biases are associated with caregivers' perceptions of the global QOL of their AD patients.12,25 Studies that include assessments of caregivers' burden or depression are needed to resolve the issue of the influence of these biases on the changes in the QOL of AD patients as assessed by caregivers.

The main finding of this study was that scores for both the psychosis factor and mood factor of the NPI at baseline were associated with changes in QOL–AD scores based on the assessments by the caregivers. The results of multiple linear regression analyses showed that both the mood factor of the NPI and the HADL predicted the changes in QOL–AD score based on the caregivers' assessments. These results were confirmed by the correlation between the changes in NPI score and changes in the QOL–AD score based on the caregivers' assessments. Activities-of-Daily-Living ability is known to be a requirement for not only memory but for complex cognitive processes, such as organization and planning. Both apathy and depression may play an important role in daily complex cognitive processes in AD patients,26–28 and psychotic symptoms may further reduce their ability to do chores and to do things for fun.2 Taken together, while the presence of neuropsychiatric symptoms was found to be strongly associated with changes in AD patients' QOL as assessed by their caregivers in this study, the ability of decreases in activities of daily living to serve as a significant predictor of changes in QOL may be attributable to neuropsychiatric symptoms, particularly, mood factor. As previous studies have indicated that the presence of executive dysfunction affects both the neuropsychiatric symptoms and instrumental activities of daily living of AD patients,26–28 executive dysfunction is likely to contribute to instrumental activities-of-daily-living dysfunction requiring complex cognitive processes. However, because we were unable to assess executive function in this study, it is unclear whether instrumental activities-of-daily-living dysfunction was associated with executive dysfunction. Appropriate measures of executive function may clarify the correlations between instrumental activities-of-daily-living dysfunction and executive dysfunction. Whether mood factor is the most important predictor among the various neuropsychiatric symptoms typically observed in AD patients remains unclear. However, in our previous study,2 the score for mood factor predicted the QOL–AD score based on the caregivers' responses in regard to patients with both mild and moderate AD, whereas psychosis factor was a predictor of the QOL–AD score based on the caregivers' responses in relation to only patients with moderate AD. Therefore, mood factor may impact the QOL–AD scores of patients with a wide range of AD severity. Although psychosis factor was a significant predictor of changes in total QOL–AD scores based on the patients' responses, a significant association between changes in total QOL–AD scores based on the patients' responses and changes in NPI scores was not observed.

The MMSE score at baseline was not a significant predictor of changes in total QOL–AD scores in our study, and the MMSE changes were not correlated with the changes in total QOL–AD scores. These findings are consistent with those of previous follow-up studies.6,7 As QOL consists of multiple dimensions that may in turn consist of various domains (psychological wellbeing, behavioral competence, objective environment, perceived quality of life) in the QOL model,29 these findings suggest that neuropsychiatric symptoms have the strongest impact on total QOL–AD scores. However, our data must be interpreted with caution. Although we did not observe any significant differences in neuropsychiatric symptoms between the lost-to-follow-up group and the follow-up group, the lost-to-follow-up group had lower baseline MMSE and SMQ scores than the follow-up group, and the lost-to-follow-up group represented a relatively high proportion of the patients (n = 44, 31.4%). The lost-to-follow-up group is likely to have experienced progressive deterioration at the end of the two-year follow-up period, implying that the effect of impaired cognition on changes to total QOL–AD scores may have been underestimated in this study. The results of previous cross-sectional studies2,11,12 have indicated that cognitive status was not a significant predictor of QOL in patients with dementia, whereas the results of longitudinal studies6–8 on QOL in dementia patients have not been consistent. Missotten et al.8 suggested that the role of the MMSE was important to the QOL score. Taken together, the above findings suggest that the lost-to-follow-up group may be a source of an exclusion bias, in other words, that if it had been possible to trace the lost-to-follow-up group at the end of the two-year follow-up period, cognitive status as well as the presence of neuropsychiatric symptoms may have been found to have affected the change in the QOL of the AD patients. Thus, it was impossible to determine the effect of cognition on the change in the QOL–AD score based on the results of this study. A large-scale longitudinal study with regular serial interval assessments is required to overcome these limitations.

A number of studies have suggested that donepezil may produce neuropsychiatric changes as well as cognitive changes.30,31 However, we did not observe a significant difference between AD patients treated with and without donepezil in the changes in the QOL–AD scores associated with neuropsychiatric changes based on the patients' responses or the caregivers' responses. Although several RCT studies of donepezil therapy have been conducted using QOL as a clinical measure in AD patients, they did not show a significant difference in QOL between the donepezil group and a placebo group.32,33 However, we do not mean to imply that the influence of donepezil should be ignored. The most plausible explanation is that a two-year follow-up period may not be appropriate for examining the clinical effects of donepezil, because most clinical trials of donepezil therapy have reported a maximum clinical benefit period of within 24 weeks.34 Shorter follow-up studies are needed to address the effects of donepezil on the QOL of AD patients.

Finally, we must point out several limitations of this study. First, we could not control the effects of either care services or social networks on the QOL of the AD patients, because this study was performed in a naturalistic setting. Nevertheless, QOL–AD assessments provide clinically important information to caregivers, allowing the effectiveness of social intervention to be improved. Second, we assessed QOL only twice: at baseline and two years later, and thus we were unable to observe the annual variations in QOL score described in a previous study.8 In addition, whether the presence of neuropsychiatric symptoms had a short-term impact on QOL determined on the basis of both the patients' and caregivers' responses remains unclear. Third, the two-year follow-up rate was not very high (68.5%). Although we did not find any significant differences in demographic data, neuropsychiatric symptoms, or HADL scores between the follow-up group and the lost-to-follow-up group, we cannot rule out the possibility that the lost-to-follow-up group may have developed serious neuropsychiatric symptoms, as suggested by Aalten et al.35,36 Furthermore, as described above, the cognitive status of the lost-to-follow-up group may have progressively deteriorated. Despite these limitations, the strength of this study lies in the fact that changes in the QOL–AD scores of a relatively large sample of patients with mild to moderate AD living at home based on the patients' and caregivers' responses were analyzed by means of detailed assessments of neuropsychiatric symptoms as well as cognition during a two-year follow-up period. This is the first longitudinal study to clarify the association between a broad spectrum of neuropsychiatric symptoms and QOL changes in AD patients. This study provides important additional information in relation to effective treatment of neuropsychiatric symptoms in order to improve the QOL of AD patients.

Ancillary