Reliability and validity of the Japanese version of the Agitated Behaviour in Dementia Scale in Alzheimer's disease: three dimensions of agitated behaviour in dementia

Authors


Dr 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

Abstract

Background:  Agitation in dementia seriously affects not only patients' quality of life (QOL), but also caregivers' QOL. Thus, an appropriate assessment of agitated behaviour in dementia is needed for clinical management. We developed the Japanese version of the Agitated Behaviour in Dementia scale (ABID), examined its reliability and validity, and carried out its factor analysis to elucidate its factor structure.

Methods:  The Japanese version of the ABID was given caregivers of 149 Japanese patients with Alzheimer's disease (AD). The internal-consistency, test–retest reliability and concurrent validity of the Japanese version of the ABID were then examined. A factor analysis was used to examine the agitated behavioural dimensions underlying ABID.

Results:  The Japanese version of the ABID showed an excellent internal reliability for both frequency ratings (Cronbach's α= 0.89) and reaction ratings (Cronbach's α= 0.92), and an excellent test–retest reliability for both frequency ratings and reaction ratings. The total score for the frequency ratings of the ABID was significantly associated with the Cohen-Mansfield Agitation Inventory (CMAI), and the total score for the reaction ratings of the ABID was significantly associated with the Zarit Burden Interview. The factor analysis showed three subtypes: physically agitated behaviour, verbally agitated behaviour and psychosis symptoms.

Conclusions:  The Japanese version of the ABID promises to be useful for assessing agitated behaviour in patients with AD. Importantly, understanding these subtypes of agitated behaviour might have implications for individualized treatment plans.

INTRODUCTION

Among the various neuropsychiatric symptoms of dementia, agitation is the most distressing manifestation. Agitated behaviours include a wide variety of behaviours, such as making noises, screaming and hurting others. Cohen-Mansfield1 defined agitation as ‘inappropriate verbal, vocal, or motor activity that is not judged by an outside observer to result directly from the needs or confusion of the individual.’ Agitated behaviours are common among patients with Alzheimer's disease (AD), dementia with Lewy bodies and frontotemporal dementia.2 The prevalence of agitation increases with the severity of dementia,3 resulting in serious daily functional impairment, as agitated behaviours are a complex phenomenon affected by interactions among cognitive impairment, pain, mental discomfort and environmental factors, such as the need for social contact and overstimulation.1 Thus, the presence of agitated behaviour might adversely affect not only the patients' quality of life (QOL) but also the caregivers' QOL.

However, the reported prevalence of agitation in patients with dementia is inconsistent. Some studies2 have suggested a low prevalence of agitation among AD patients with dementia (20–30%); these studies used a multisymptom rating scale, such as the Neuropsychiatric Inventory (NPI).4 Other studies5,6 have reported a high prevalence of agitation among AD patients with dementia (60–90%); these studies used an instrument specific for agitated behaviours in dementia, such as the Cohen-Mansfield Agitation Inventory (CMAI).7 These differences in estimates might arise from not only the lack of a uniform definition for agitation, but also the use of different measurements for assessing agitation and the variety of patient settings (e.g. community residents, nursing homes and outpatient clinics). Cohen-Mansfield suggested that agitation in dementia is manifested in a wide variety of verbal and physical behaviours.1 More importantly, agitated behaviours in dementia might occur without psychosis (delusions, hallucinations).

Agitated behaviours are problematic in that they can cause severe distress to the caregivers of dementia patients. For the best management of agitation, both a careful and detailed assessment of agitated behaviours and the distress experienced by caregivers in response to the patient's agitated behaviour symptoms are clinically important. Thus, a need exists for the appropriate assessment of agitated behaviour in dementia. The CMAI is the most widely known instrument for measuring agitation in dementia and has been regarded as a useful tool for assessments in clinical trials.8 The original 29 items of the CMAI were devised to assess agitation among nursing home residents.7 Later, a revised and expanded version of the CMAI that included 36 items was developed as a community form to assess agitation among community-residing persons.9 Rabinowitz et al. supported the construct validity of the CMAI in large samples with dementia by showing the robustness of the factor structure that emerged on the CMAI across countries.10 Cohen-Mansfield proposed that agitated behaviours can be classified into several subtypes: verbally non-aggressive, verbally aggressive, physically non-aggressive and physically aggressive.1

Another assessment with a specific focus on agitation has also been developed. The Agitated Behaviour in Dementia scale (ABID), originally devised by Logsdon et al.,11 is a useful assessment that utilizes a caregiver-based rating scale. The items in the ABID were derived from clinical experience and other behaviour assessments, such as the CMAI. Thus, a strong correlation between the ABID and the CMAI has been reported.11 The ABID is believed to have several advantages for detecting and quantifying agitation using agitation-specific rating scales. First, the ABID uses a 16-item scale to assess a wide range of behaviours associated with agitation during the 2 weeks preceding the interview in community-residing patients with mild to moderate levels of dementia. Second, the ABID has the advantage of being able to assess not only the frequency of agitated behaviours, but also the caregivers' reactions to each behavioral problem in patients with mild to moderate dementia. Thus, the ABID provides the possibility of assessing two important outcomes – the frequency of agitation and the caregiver's level of distress in response to the dementia of outpatients. Third, the internal consistency reliability of the ABID has been reported to be excellent (alpha coefficient 0.70).11 Also, the test–retest reliability of the ABID frequency and the reaction ratings have been reported to have a good external reliability, with correlation coefficients (ICC) of 0.73 and 0.60, respectively.11 The validity of the ABID has been confirmed.11 Furthermore, Teri et al. suggested that the ABID is an appropriate assessment for evaluating the efficacy of clinical non-pharmacological interventions for agitation in dementia patients.12 Last, this scale is simple and easy to administer, requiring less than 20 min to complete.

In Japan, although the reliability and the validity of the CMAI have already been established, only the original 29 items of the CMAI for nursing home residents have been published.13

The ABID has several clinical advantages over the CMAI for the assessment of dementia in patients with agitated behaviours. First, the ABID might be an appropriate assessment for community-dwelling subjects with mild to moderate levels of dementia.11 In contrast, the CMAI might be more appropriate for more severely disturbed nursing home residents. Even the community version of the CMAI, which contains 36 items, could not detect any significant differences in terms of the agitation level among AD patients with mild to moderate levels.8 Second, the CMAI measures only the frequency of agitated behaviours over the preceding 2 weeks. However, the ABID has caregiver reaction scales in addition to frequency scores. In addition, while the frequencies of observable agitated behaviours over the preceding 2 weeks are rated on the CMAI, the ABID assesses the observable agitated behaviours for each of the preceding 2 weeks and sums the results for a total frequency score.8,9,11 Thus, the ABID can provide weekly changes in agitated behaviours during the previous 2 weeks in response to clinical interventions, such as clinical non-pharmacological interventions for agitation in dementia patients. Third, psychosis symptoms (delusions, hallucinations) are related to agitated behaviours.2,14 However, the CMAI did not include any items with psychotic symptoms. In contrast, the ABID includes items related to psychosis (delusions, hallucinations). Thus, for the appropriate assessment of agitated behaviour in dementia, we decided to develop a Japanese version of the ABID for community-residing patients with mild to moderate levels of dementia. Furthermore, we are unaware of any published data regarding the factor structure of the ABID in AD patients. Thus, first, we developed a Japanese version of the ABID using back-translation and ascertained both its reliability and validity. Next, we examined the factor structure of the Japanese version of the ABID among a large sample of AD patients. We hypothesized that different behaviour subtypes (e.g. verbally agitated behaviour and physically agitated behaviour) might underlie the agitation assessed using the ABID, similar to the classification of agitation into several subtypes proposed by Cohen-Mansfield.1,15

METHODS

Participants

A total of 169 consecutive Japanese patients with AD who attended Nagoya City Universal Hospital and Yagoto Hospital in Nagoya, Japan, as outpatients between September 2003 and August 2004 were recruited for the present study.

The diagnostic evaluation included a complete history and physical examination, routine blood tests, either a magnetic resonance imaging (MRI) or a computed tomography (CT) scan of the brain, and neuropsychological testing. The study inclusion criteria consisted of (i) a diagnosis of probable AD according to the National Institute of Neurological and Communication Disorders and Stroke/Alzheimer's Disease and Related Disorders Association (NINCDS/ADRDA) criteria;16 (ii) very mild to moderate functional severity (grade 0.5, 1 or 2 on the clinical dementia rating [CDR17]); (iii) no history of antipsychotic or antidepressant medication, as these medications can affect the neuropsychiatric symptoms of AD patients; and (iv) residence with a caregiver in a community dwelling. Patients were excluded if (i) other neurological diseases were present; (ii) the patient had a previous history of mental illness or substance abuse before the onset of dementia; (iii) either an MRI or a CT scan showed focal brain lesions; or (iv) reliable informed consent could not be obtained from the patient and/or his/her relatives.

In the present study, AD patients with severe cognitive impairment (grade 3 on the CDR) were not enrolled, because the ABID is most appropriate for assessing observed agitated behaviours in patients with mild to moderate levels of dementia.11

Of the 169 patients originally screened, 20 AD patients were excluded from the analysis, because these patients did not meet the inclusion criteria.

The study protocol was approved by the Ethics Review Committee of Nagoya City University Graduate School of Medical Sciences. After the purpose of the study was explained, written informed consent was obtained from each AD patient and, when necessary, from his/her caregivers.

Translation of ABID

The Agitated Behaviour in Dementia Scale (ABID) was developed by Logsdon et al.11 It consists of a 16-item questionnaire seeking responses from caregivers about the common agitated behaviours in dementia patients. With the original authors' permission (R.G. Logsdon and L. Teri), we translated the original English version into Japanese. We followed the standard back-translation procedure to ascertain the semantic equivalence of the Japanese version with the original English version. The back-translated version was examined by Logsdon to point out possible discrepancies. We repeated this process until Logsdon agreed that the original and back-translated versions matched closely.

Procedure

The Japanese version of the ABID was given to all the caregivers by a well-trained psychiatrist. In accordance with the procedure recommended for the original ABID, described by Logsdon et al.,11 the caregivers completed the questionnaire while visiting the outpatient clinic and received minimal assistance from the interviewer.

The ABID includes items that have been identified by Logsdon et al.11 as most problematic in individuals with dementia, and that can be observed and described objectively. The caregivers first rated each behaviour according to the frequency of occurrence during each of the 2 weeks immediately before the assessment on a scale of 0–3 (0: did not occur during the week; 1: occurred once to twice during the week; 2: occurred three to six times during the week; 3: occurred daily or more often). The two weekly scores for each item were then added together, and the resulting item scores ranged from 0 to 6. The item scores were summed to obtain the total score, with possible scores ranging from 0 to 96. Then, the caregivers rated their own reactions to each problem behaviour on a scale of 0–4 (0: not upsetting; 4: extremely upsetting). The caregiver's reactions were rated once for each item and then summed. The total reaction scores had a possible range of 0–64.

At the time of the administration of the ABID, the following tests were also carried out.

  • 1The Cohen-Mansfield Agitation Inventory (CMAI):15 This test was originally developed to measure agitation in nursing home residents It consists of 29 observable agitated behaviours rated using a 7-point Likert-type scale according to the frequency of occurrence during the prior 2 weeks (0: never occurred; 7: occurred several times per hour). The reliability and validity of the CMAI have been established by Finkel et al.7 The revised and expanded version of the CMAI, including 36 items, has been developed as a community form to assess agitation among community-residing persons and has been used for senior day center participants.9 Homma et al. confirmed the reliability and validity of the original 29 items of the CMAI in nursing home residents.13
  • 2The Zarit Burden Interview (ZBI):18 This test consists of 22 items rated using a 5-point Likert scale (never = 0, nearly always = 4) aimed to assess caregiver burden. The total burden was obtained by adding the scores for all the items with a range of 0–88, with higher scores showing a greater burden. The reliability and the validity of the Japanese version of the test battery have been confirmed.3
  • 3Mini-Mental State Examination (MMSE).19

Data analysis

We used spss 11.0J software for Windows for the statistical analysis.

Reliability

The reliability of this scale was assessed in two ways. First, the test–retest reliability was assessed in 70 caregivers of AD patients at an interval of 1 month after the initial evaluation. The 70 AD patients had not taken any antipsychotic medications not only before the first assessment, but also during the 1-month interval. The test–retest reliability was estimated using analysis of variance intraclass correlation coefficients (anova icc). In general, an anova icc above 0.70 shows a good reliability. Second, the internal consistency of the scale was estimated using Cronbach's alpha coefficients (n= 149). A Cronbach's alpha coefficient above 0.70 is indicative of a good internal consistency.

Validity

The concurrent validity of the Japanese version of the ABID was verified by examining the correlation between the ABID frequency ratings and the CMAI, and the correlation between the ABID reaction ratings and the ZBI. The alpha level was set at 0.01.

Factor analysis

A principal component factor analysis using varimax rotation was carried out on the 16 items for both the frequency rating and the reaction rating of the ABID. The models included factors with an eigenvalue >1. An item was considered to load onto a factor if its factor loading score exceeded 0.30.

RESULTS

Demographic and clinical characteristics

Table 1 shows the mean scores and standard deviations of the clinical and demographic characteristics of both the AD patients and their caregivers. Among the caregivers (n= 149), 73.8% were spouses (n= 110), 19.4% were daughters-in-law (n= 29) and 6.7% were adult children (n= 10). In Japan, the prevalence of AD is reportedly higher in women than in men.2 In the present study, we also observed a predominance of women (n= 90) among the AD patients. Consequently, male spouses were the main caregivers. Thus, in the present study, the percentage of male caregivers was higher than that of female caregivers.

Table 1.  Demographic data of Alzheimer's disease patients and their caregivers
 AD patientsCaregivers
Mean ± SDMean ± SD
(n= 149)(n= 149)
  1. ABID, Agitated Behaviour in Dementia Scale; AD, Alzheimer's disease; MMSE, Mini-Mental State Examination.

Sex (males/females)59/9087/62
Age (years)73.4 ± 8.362.6 ± 10.2
Education (years)9.7 ± 0.810.4 ± 2.8
Dementia history  
Duration of illness (years)3.1 ± 1.1NA
MMSE18.9 ± 3.728.9 ± 1.7
ABID questionnaire  
 Frequency ratings25.1 ± 18.2NA
 Reaction ratings19.9 ± 15.3NA

Table 2 shows the mean of each item according to the ABID frequency and reaction ratings.

Table 2.  Item means of both the Agitated Behaviour in Dementia Scale frequency scores and the Agitated Behaviour in Dementia Scale caregiver reaction scores among 149 Alzheimer's disease patients
Number; Agitated behavioural charactersFrequency ratingsReaction ratings
Mean ± SDMean ± SD
(n= 149)(n= 149)
  1. ABID, Agitated Behaviour in Dementia Scale.

 1 Verbally threatening or aggressive toward others2.29 ± 2.011.77 ± 1.44
 2 Physically threatening or aggressive toward others1.19 ± 1.961.05 ± 1.60
 3 Harmful to self0.67 ± 1.400.59 ± 1.18
 4 Inappropriate screaming or crying out1.46 ± 1.750.66 ± 0.73
 5 Destroying property0.91 ± 1.620.80 ± 1.35
 6 Refusing to accept appropriate help2.21 ± 1.751.82 ± 1.41
 7 Trying to leave (or leaving) home inappropriately1.44 ± 2.061.21 ± 1.60
 8 Arguing, irritability or complaining2.50 ± 1.951.89 ± 1.42
 9 Socially inappropriate behaviour1.07 ± 1.511.02 ± 1.59
10 Inappropriate sexual behaviour0.40 ± 1.010.36 ± 0.89
11 Restlessness, fidgetiness, inability to sit still1.05 ± 1.761.02 ± 1.59
12 Worrying, anxiety and/or fearfulness3.49 ± 2.012.29 ± 1.37
13 Easily agitated or upset2.83 ± 2.212.13 ± 1.47
14 Waking and getting up at night (other than trip to the bathroom)1.19 ± 1.520.89 ± 1.24
15 Incorrect, distressing beliefs1.53 ± 2.111.11 ± 1.50
16 Seeing, hearing or sensing distressing people or things that are not really present0.89 ± 1.680.62 ± 1.20

Reliability

The test–retest reliability (n= 70) of the ABID frequency and reaction ratings after 1 month showed an excellent external reliability, with a correlation coefficient (ICC) of 0.85 (95% CI = 0.75–0.96) and 0.89 (95% CI = 0.82–0.93), respectively. The alpha coefficients for the ABID frequency and reaction ratings (n= 149) were 0.89 (95% CI = 0.87–0.92) and 0.92 (95% CI = 0.90–0.94), respectively. These scores showed an excellent internal consistency for the total scores.

Concurrent validity

The frequency ratings on the ABID were positively correlated with the total CMAI scores (r= 0.86, 95% CI = 0.72–0.91, P < 0.001). Similarly, the caregiver reaction ratings on the ABID were positively correlated with the total ZBI scores (r= 0.696, 95% CI = 0.58–0.78, P < 0.001). Both the frequency ratings and the reaction ratings were significantly correlated with the total MMSE score (r=−0.82, 95% CI =−0.84 to −0.78, P < 0.001, r=−0.83, 95% CI=−0.85 to −0.76, P < 0.001) and the duration of illness (r= 0.573, 95% CI = 0.43–0.64, P < 0.001, r= 0.573, 95% CI = 0.45–0.68, P < 0.001). However, even after controlling for the MMSE score and the duration of illness, the partial correlation between the frequency ratings on the ABID and the total CMAI scores remained significant. Also, after controlling for the MMSE and the duration of illness, the partial correlation between the caregiver reaction ratings on the ABID and the total ZBI scores remained significant.

Factor analysis

An exploratory principal component analysis with varimax rotation using eigenvalues >1 reduced the 16 variables to three factors for both the frequency rating and the reaction rating of the ABID. The three factors explained 69.6% of the variance in the frequency rating data of the ABID. The three factors also explained 73.1% of the variance in the data of the reaction rating data of the ABID. Visual inspection of the scree plot also supported a three-factor solution. Tables 3 and 4 show the rotated component matrix of the three-factor solution. The first factor in the frequency rating data of the ABID had high loadings on such items as ‘physically threatening or aggressive toward others’, ‘destroying property’ and ‘restlessness, fidgetiness, inability to sit still’. According to the subtypes of agitated behaviours,1,5,20 while ‘physically threatening or aggressive toward others’, ‘destroying property’, ‘harmful to self’ and ‘inappropriate sexual behaviour’ are categorized into physically aggressive behaviour, ‘restlessness, fidgetiness, inability to sit still’ and ‘trying to leave home inappropriately’ with high loadings are regarded as physically non-aggressive behaviour. As in the present study, Cohen-Mansfield et al. reported that the latter behaviours were the least disruptive, yet were manifested at a very high frequency among various types of agitated behaviours.5 Therefore, the first factor was termed ‘physically agitated behaviour’. The second factor in the frequency rating data of the ABID included most of the items corresponding to verbally agitated behavior, such as ‘arguing, irritability, or complaining’ and ‘refusing to accept appropriate help’, as suggested by Cohen-Mansfield et al.5,20 Therefore, we named this factor ‘verbally agitated behaviour’. The third factor in the frequency rating data of the ABID mainly contained items representing psychosis symptoms (delusion, hallucination), such as ‘seeing, hearing or sensing distressing people or things that are not really present’ and ‘incorrect, distressing beliefs’. Therefore, the third factor was interpreted as representing the ‘psychosis symptoms’.

Table 3.  Factor analysis of the Agitated Behavior in Dementia Scale frequency scores among 149 Alzheimer's disease patients
 Agitated behavioural characteristicsFactor 1Factor 2Factor 3
  1. ABID, Agitated Behaviour in Dementia Scale, Factor loadings of 0.511 or more are in boldface.

 2Physically threatening or aggressive toward others0.9440.1680.033
 5Destroying property0.9260.1070.005
11Restlessness, fidgetiness, inability to sit still0.8970.1940.042
 3Harmful to self0.860−0.0130.118
 7Trying to leave (or leaving) home inappropriately0.8360.2080.164
10Inappropriate sexual behaviour0.7150.0670.053
 8Arguing, irritability or complaining−0.0920.8820.186
 1Verbally threatening or aggressive toward others0.0290.8400.304
 4Inappropriate screaming or crying out−0.1010.7640.440
13Easily agitated or upset0.3450.7400.142
12Worrying, anxiety and/or fearfulness0.2890.7140.037
 9Socially inappropriate behaviour0.1710.6960.263
 6Refusing to accept appropriate help0.3810.5620.132
16Seeing, hearing or sensing distressing people or things that are not really present0.0020.1700.838
15Incorrect, distressing beliefs0.1260.2880.741
14Waking and getting up at night (other than trip to the bathroom)0.1480.2630.511
Table 4.  Factor analysis of the Agitated Behaviour in Dementia Scale caregiver reaction rating scores among 149 Alzheimer's disease patients
 Agitated Behavioral characteristicsFactor 1Factor 2Factor 3
  1. ABID, Agitated Behaviour in Dementia Scale, Factor loadings of 0.522 or more are in boldface.

 8Arguing, irritability or complaining0.900−0.0430.151
 1Verbally threatening or aggressive toward others0.8870.1220.121
 4Inappropriate screaming or crying out0.842−0.0550.255
13Easily agitated or upset0.7400.4360.180
12Worrying, anxiety and/or fearfulness0.7290.4400.026
 9Socially inappropriate behaviour0.7270.2470.204
 6Refusing to accept appropriate help0.6360.4480.126
14Waking and getting up at night (other than trip to the bathroom)0.5220.1500.340
 2Physically threatening or aggressive toward others0.2720.8960.035
 5Destroying property0.1680.891−0.007
 3Harmful to self0.0030.8840.034
11Restlessness, fidgetiness, inability to sit still0.2700.8830.076
 7Trying to leave (or leaving) home inappropriately0.3000.8130.153
10Inappropriate sexual behaviour−0.0090.7090.121
16Seeing, hearing, or sensing distressing people or things that are not really present0.1990.0320.872
15Incorrect, distressing beliefs0.3860.1660.711

Similarly, a factor analysis of the reaction rating data of the ABID (Table 4) showed that the first factor had a high loading on items associated with ‘verbally agitated behaviour’. Therefore, we named this factor of the reaction rating of the ABID ‘verbally agitated behaviour’. The second factor included items associated with ‘physically agitated behaviour’, and we named this factor of the reaction rating of the ABID ‘physically agitated behaviour’. The third factor contained items associated with ‘psychosis symptoms’. Thus, we named this factor of the reaction rating of the ABID ‘psychosis symptoms’.

DISCUSSION

The present study shows that the Japanese version of the ABID scale has an excellent internal consistency reliability for the frequency (alpha coefficient 0.89) and the reaction ratings (alpha coefficient 0.92). The Japanese version of the ABID scale also has an excellent test–retest reliability for the frequency (ICC 0.85) and the reaction ratings (ICC 0.89). As mentioned in the introduction, the internal consistency reliability of the ABID (alpha coefficient 0.70) and the test–retest reliability of the ABID frequency and the reaction ratings, with ICC of 0.73 and 0.60, respectively, have been reported by Logston et al.11 Thus, almost similar levels were obtained for both the internal consistency reliability and the test–retest reliability of the Japanese version of the ABID scale. Furthermore, we showed that each score (the frequency and the reaction ratings) had a satisfactory concurrent validity. Three factors underlying the agitated behaviour evaluated using the Japanese version of the ABID were identified: (i) physically agitated behaviour; (ii) verbally agitated behaviour; and (iii) psychosis symptoms. Therefore, the Japanese version of the ABID scale might be useful for assessing agitated behaviour in community-residing AD patients with mild to moderate dementia.

Cohen-Mansfield proposed that agitated behaviours can be divided into two dimensions: (i) aggressive versus non-aggressive; and (ii) verbal versus physical behaviours.1,15 Thus, agitated behaviours can be classified into four subtypes: (i) aggressive–physical behaviours; (ii) aggressive–verbal behaviours, (iii) non-aggressive–physical behaviours; and (iv) non-aggressive–verbal behaviours. Most previous factor analyses of the CMAI carried out in different countries have supported these dimensions. However, some studies15,21 have shown that the CMAI consisted of three factors: (i) physically aggressive behaviours; (ii) physically non-aggressive behaviours; and (iii) verbally agitated behaviours. Other three- or four-factor solutions (aggressive behaviour, physically non-aggressive behaviour, verbally agitated behaviour, and hiding or hoarding) of the CMAI have been found.5,10 Several items in the CMAI belonged to either physically aggressive behaviour or aggressive behaviour. For example, while several items such as hitting, kicking and pushing were included as physically aggressive behaviour in the former studies,15,21 the latter studies5,10 included such items as aggressive behaviour. Despite this disagreement, these factor analyses of the CMAI (the original 29 items) have similar clinical features, as either physically aggressive behaviour or aggressive behaviour were loaded on the first factor in all the previous studies. The difference between ‘physically aggressive behaviour’ and ‘aggressive behaviour’ might reflect a difference in whether some items, such as screaming, cursing or verbal aggression, within the subtype of verbally aggressive behaviours are loaded onto the first factor.10

In line with previous factor analyses of the CMAI, we named the first factor in the ABID frequency scores as ‘physically agitated behaviour’. Several items (e.g. ‘harmful to self’) in this factor corresponded to the agitated behaviours that Cohen-Mansfield described as being physically aggressive.1 However, two items in the ABID with high loadings, ‘restlessness, fidgetiness, inability to sit still’ and ‘trying to leave home inappropriately’, might correspond to physically non-aggressive behaviour, such as ‘general restlessness’, ‘pacing, aimless wandering’ and ‘trying to get to a different place’ in the CMAI. Previous factor analyses of the CMAI have consistently regarded these items as belonging to the ‘physically non-aggressive behaviour’ factor.5,10 Logsdon et al.22 has suggested that both wandering and restlessness can cause significant severe problems among the various agitated behaviours, because these behaviours are associated with severe caregiver distress. However, our factor analysis of the caregiver's reaction ratings in the ABID showed that both ‘restlessness, fidgetiness, inability to sit still’ and ‘trying to leave home inappropriately’ were included in the same factor that included other items in the ABID frequency scores termed as ‘physically agitated behaviour’. Thus, it seems reasonable to assume that these two items are regarded in a similar manner to the other items included as ‘physically agitated behaviour’. Unlike the CMAI, the ABID might not have the sensitivity to discriminate between psychically aggressive behaviour and physically non- aggressive behaviour.

With regard to the second factor of the ABID frequency scores, we observed that most of the items were represented as ‘verbally agitated behaviour’. All the previous factor analyses of the CMAI have confirmed that constant requests for attention, complaining and negativism are the core concepts of verbally agitated behaviour.5,10,15,21 Interestingly, several items, such as ‘easily agitated or upset’, ‘worrying, anxiety and/or fearfulness’ and ‘socially inappropriate behaviour’, were included in the factor termed ‘verbally agitated behaviour’ in the ABID. Manifestations of verbally agitated behaviours are often associated with anxiety, pain and discomfort.20,23 Cohen-Mansfield et al.20,23 suggested that some verbal behaviours in dementia might arise as a result of social isolation caused by either loneliness or the need for social contact in patients with dementia and that poor communication abilities are the root of such behaviours. Also, negative social interactions between dementia patients and their caregivers can cause ‘easily agitated’, ‘anxiety’ or ‘socially inappropriate behaviour’. The study by Cohen-Mansfield and Libin20 showed that environmental factors, including social isolation and discomfort, play an important role in the cause of verbal behaviours. Taken together with these findings, the factor termed ‘verbally agitated behaviour’ in the present study suggests that verbally agitated behaviour in patients with dementia might reflect a representation of discomfort, anxiety or fear.

A previous study suggested that agitation was strongly associated with both delusions and hallucinations in patients with dementia.24 In our factor analysis of the ABID, both delusions and hallucinations emerged as a third factor (psychosis factor). The factor analyses of the CMAI did not show any psychosis factor because, unlike the ABID, the CMAI does not include any items associated with either delusion or hallucination. However, factor analyses of the NPI in patients with AD have shown that agitation is a distinct syndrome from psychosis.25 In these studies, although agitation was regarded as the subsyndrome, ‘hyperactivity’, both delusions and hallucinations were included in the psychosis syndrome. The findings in the present study support the notion that the psychosis factor in patients with mild to moderate AD can be separated from other physical and verbal agitation factors. Additional studies of patients with more advanced stages of dementia are needed to confirm the presence of this third factor.

Sleep or night-time behavioural disturbances, such as ‘waking and getting up at night’, can be considered as wandering associated with physically agitated behaviour.5,22 However, in the present study, ‘waking and getting up at night’ did not load on the first factor, which was termed as ‘physically agitated behaviour’. Instead, ‘waking and getting up at night’ was included in the factor termed ‘psychosis symptoms’ in the frequency rating data of the ABID. In contrast, in the reaction rating data of the ABID, ‘waking and getting up at night’ was included in the ‘verbally agitated behaviour’ factor. The reason why ‘waking and getting up at night ’ did not load on the ‘physically agitated behaviour’ factor for either the frequency rating or the reaction rating of the ABID is unclear. However, the factor loading of the item ‘waking and getting up at night’ was the smallest among all the factors for both the frequency rating and the reaction rating of the ABID (0.511 and 0.522, respectively). Logsdon et al.22 suggested that ‘waking and getting up at night’ was most common among dementia patients with severe cognitive impairment. Thus, further large-scale studies that include patients with severe AD might be required to determine definitively to which factor the item ‘waking and getting up at night’ belongs.

Finally, we must address several limitations of the present study. First, our sample did not include AD patients with severe cognitive impairment (grade 3 on the CDR). As the stage of dementia advances, agitated behaviours are known to increase from mild to severe stages.11 When examining changes in agitated behaviors among patients with AD using the Japanese version of the ABID in a longitudinal study, it is important to note that even mild or moderate levels of dementia at the time of the first assessment can progress to a more severe stage during the follow-up period as the disease progresses. Thus, a future study might be required to examine the utility of applying the ABID to patients with more advanced AD. Second, agitated behaviours in dementia are known to fluctuate throughout the day.26 This previous study showed that agitation gradually increased from the morning until a peak around 1600 hours, then decreased thereafter. However, we did not examine the temporal pattern of the agitated behaviours in AD patients. A study using the ABID to examine the temporal pattern of agitated behaviours in patients with dementia is needed. Third, the data used in the present study was obtained from two outpatient clinics. Future large-scale studies are needed to validate these important findings for a wide range of dementia patients.

In conclusion, the present study suggests that the Japanese version of the ABID is a reliable and valid instrument, and that agitated behaviours can be divided into three subtypes: (i) ‘physically agitated behaviour’; (ii) ‘verbally agitated behaviour’; and (iii) ‘psychosis’. As suggested by Cohen-Mansfield,1,20 understanding these subtypes of agitation might have important clinical implications for the formulation of individualized treatment plans for agitation in patients with dementia. Clinical interventions for agitation in patients with dementia that are based on these three subtypes of agitated behaviour might be beneficial for distressed caregivers in Japan.

ACKNOWLEDGEMENTS

The authors gratefully acknowledge a grant from a Grant-in-Aid for Scientific Research (c) (22530750, 22591293) from the Ministry of Education, Culture, Sports, Sciences and Technology in Japan.

Ancillary