Professional Documents
Culture Documents
The Modified CAMDEX
The Modified CAMDEX
611
pp ‒
Blackwell Science, LtdOxford, UKJIRJournal of Intellectual Disability Research-Blackwell Publishing Ltd, Original ArticleCAMDEX informant interview for use with adults with DSS. L. Ball et al.
© Blackwell Publishing Ltd, Journal of Intellectual Disability Research , ‒
Journal of Intellectual Disability Research
613
S. L. Ball et al. • CAMDEX informant interview for use with adults with DS
process, mapping onto standard diagnostic criteria Table 1 Sample attrition and progression in diagnosis over years
and thereby ensuring that a reliable and valid judge-
ment can be made as to whether dementia is present Diagnosis at Diagnosis at
or not. A diagnosis of dementia (specifically AD) Time 1 n at Time 1 Time 2 n at Time 2
should both reflect and, more importantly, predict a
progressive decline in cognitive and functional ability. Not AD 65 Not AD 45
In this paper, using longitudinal diagnostic and AD 7
neuropsychological data from a previous prevalence Deceased 4
Refused 9
and ongoing longitudinal study of older people with
AD 9 Not AD 0
DS (Holland et al. ), we report on the validity AD 4
and reliability of a modified version of the informant Deceased 5
interview of the Cambridge Examination for Mental Refused 0
Disorders of the Elderly (CAMDEX) (Roth et al.
) as a tool for the diagnosis of dementia in people AD, Alzheimer’s disease.
with DS.
The CAMDEX
Methods The CAMDEX was originally developed in as
Approval for the prevalence and longitudinal studies, a standardized instrument for the diagnosis of mental
from which the data used in this paper are taken, was disorder in the elderly general population, with par-
obtained from the Cambridge Local Research Ethics ticular reference to the early detection of dementia
Committee. We have used data from two of the (Roth et al. ). It was subsequently published by
assessment procedures employed as part of these past Cambridge University Press (Roth et al. ) and a
and ongoing studies: the modified CAMDEX infor- revised version approved in (Roth et al. ).
mant interview and the Cambridge Cognitive Exam- The schedule includes an informant interview, an
ination (CAMCOG) neuropsychological test battery interview with the participant, an objective examina-
(Holland et al. ). tion of cognitive function (CAMCOG), a standard-
ized schedule for recording observations, and a
physical examination and information on laboratory
investigations.
Participants
The informant interview provides a means for col-
At Time , participants ( females and males) lecting information in a structured manner about
made up a population sample of all individuals with those areas of function that are likely to change with
DS living in one health district (population ), the onset of dementia or of any other mental disorder.
who were over the age of years on July . It includes several questions about the informant’s
Eligible participants were identified first through observations on each of the following: the person’s
written communication, and then by direct contact, memory, general mental and intellectual functioning,
with local care-providers for people with ID and with judgement, general performance, specific higher cor-
community health teams. Each person with DS and tical functions, and personality, as well as the pres-
his/her main carer were then contacted and the con- ence or not of specific symptoms and relevant
sent of the person with DS and/or the agreement of previous medical and family history. Its validity and
his/her main carer was sought. Fifty-six of this origi- reliability in the general elderly population have been
nal took part in a follow-up study approximately shown to be good with, for example, information on
years later (Time ). At the time of the second memory and mental functioning from the informant
assessment nine participants or their carers did not interview highly correlated with objectively measured
wish to be interviewed again and nine of the partici- decline (Neri et al. ). On measures of inter-rater
pants had died, five of whom had suffered from AD reliability in the general population the correlation
prior to their death (see Table ). between total scores obtained by the two raters has
© Blackwell Publishing Ltd, Journal of Intellectual Disability Research , ‒
Journal of Intellectual Disability Research
614
S. L. Ball et al. • CAMDEX informant interview for use with adults with DS
been found to be high (r = ., P < .) as has the Huppert et al. ), was completed at Time and
level of agreement on individual items (median phi at Time by all participants who were able to do so.
coefficient = ., range = .–.) (Roth et al. This was used to obtain objective evidence of change
). in the areas of cognitive function that characteristi-
cally decline with dementia in the general elderly
population. Hon et al. () found that older par-
Diagnostic assessment (modified CAMDEX
ticipants with DS (> years) performed significantly
informant interview)
worse than younger ones on a slightly modified ver-
This is a two-step procedure: the first requiring the sion of the CAMCOG. Scores on the CAMCOG
collection of information pertinent to the diagnosis subscales for orientation, language, memory, praxis,
of dementia and the second, the use of this informa- attention/calculation, abstract thinking and percep-
tion to determine whether the necessary clinical cri- tion, as well as the total score, have been shown to
teria are met for the diagnosis of dementia. For the differentiate significantly between normal individuals
first of these steps, the CAMDEX informant inter- and those with dementia (Huppert et al. ).
view has been modified for use in our previous stud-
ies. This modification took into account the fact that
Validity
cognitive and functional abilities may already be
impaired, because of the person’s pre-existing ID. If As there is currently no ‘gold standard’ tool for the
a particular problem is said to be present in answer diagnosis of dementia in people with DS, the tradi-
to one of the questions (e.g. ‘Does he or she have tional approach to establishing the validity of diag-
difficulty in remembering recent events?’), the infor- nostic instruments has been to compare findings with
mant is then asked whether this is a change (i.e. a the judgements of clinicians. This is potentially prob-
deterioration), or whether this is something that has lematic, as clinicians tend to make these clinical deci-
always been a problem. There must be evidence of sions using broadly the same assessment methods as
deterioration in that particular function (e.g. mem- the instruments that are developed and thus high
ory), as observed by the informant, if that symptom levels of agreement are likely, whether or not the
is then to be scored as being present. assessments are valid. If, for example, dementia is
For the second step in both these studies, the clin- over-diagnosed by clinicians generally and over-
ical information from the informant interview was diagnosed when using a particular diagnostic tool to
presented to A.J.H. (psychiatrist) and F.A.H. (psy- the same degree, a high level of agreement between
chologist), blind to the previous diagnostic status, to the two would be a false indicator for the validity
neuropsychological test performance, and to the age of the test.
and gender of the person with DS. A judgement was In the case of dementia (specifically AD), one
then made as to whether the person had deteriorated important test of the validity of a diagnostic tool is
in each of the cognitive and functional domains in whether it corresponds with independent evidence of
which change is necessary in order to meet the crite- progressive decline prior to diagnosis (concurrent
ria for the diagnosis of dementia, or of any other validity). The concurrent validity of the CAMDEX
psychiatric disorder. A consensus diagnosis was informant interview was established by examining the
reached by rating these findings against the opera- agreement between informant interview-based diag-
tional criteria for a CAMDEX, ICD- and DSM IV nosis and objective measurement of decline on cog-
diagnosis, also taking into account relevant labora- nitive tasks over the preceding years.
tory information (e.g. thyroid function). All diag- An even stronger test of validity is whether the
noses reported in this paper are based on CAMDEX diagnosis predicts the course of the illness (predictive
criteria. validity). If it does not, then clearly the diagnosis is
incorrect, as AD is a progressive brain disorder. The
predictive validity of the informant interview was
Neuropsychological assessment (CAMCOG)
examined by determining the degree to which diag-
The CAMCOG neuropsychological test battery, nosis at Time predicts both subsequent decline on
which forms part of the CAMDEX (Roth et al. ; cognitive tasks and diagnosis at Time .
© Blackwell Publishing Ltd, Journal of Intellectual Disability Research , ‒
Journal of Intellectual Disability Research
615
S. L. Ball et al. • CAMDEX informant interview for use with adults with DS
Not AD at Time 2 31 1 – – 32
AD at Time 2 2 4 – 3 9
Total 33 5 – 3 41
Four were diagnosed with AD. The informants were since Time is significantly more likely to have
either relatives () or paid carers (), who had occurred in those diagnosed with AD at Time than
known the participants for a mean of years. All but in those without a diagnosis of AD (P < .,
two of these carers had known the participants for Fisher’s Exact). The odds of a participant having
more than years. One had had months of direct shown a decline of >. points are at least times
contact and one only weeks, but both had obtained greater if he/she has been diagnosed with CAMDEX
detailed information from colleagues spanning over AD than if he/she has not (odds ratio = ., %
several years. The responses of the informants CI = .–.). This is a conservative estimate
were rated simultaneously and independently by a based on the lower bound of the confidence interval
psychiatrist (P.T.) who conducted the informant for the odds ratio.
interview, and a psychologist (S.L.B.) who observed. A detailed look at ‘false positive’ and ‘false nega-
For each participant the ratings were compared for tive’ cases provides some further support for the con-
all items in the interview. The level of agreement current validity of the informant measure. The two
between interviewers for each item was calculated by individuals diagnosed as having AD at Time who
means of the Kappa coefficient. had not shown decline in CAMCOG score since
Time had been diagnosed with only mild AD. One
had shown decline on memory and praxis subscales
Results (- and -, respectively) but improvements in other
areas of cognition, and the other showed decline on
Diagnoses
the praxis subscale only (-). This observation would
At Time , nine out of the participants received a suggest that these two people probably do have AD,
diagnosis of AD. At Time , of the remaining but as the diagnosis was made early in the course of
participants received a diagnosis of AD. Of these, four the disease, the level of cognitive decline over the
had been diagnosed with AD at Time and seven previous years was insufficient to meet the ‘decline’
were new cases since baseline. Five of the nine diag- criteria we set for this study. Further follow-up
nosed with AD at Time had died before Time . All assessments are required to confirm whether or not
were still suffering from AD at the time of their this is the case.
deaths. The one participant in the group of considered
not to have AD, who had in fact showed decline on
the CAMCOG between Times and , was found to
Concurrent validity
be suffering from a serious physical illness at the time
Diagnosis at Time was cross-tabulated with decline of the CAMCOG assessment at Time . This explains
on CAMCOG since Time . This revealed that seven her impaired performance on the task at that time.
out of the nine individuals diagnosed with AD at Thus, even though she showed cognitive decline
Time (%) had shown a decline of >. points when the two CAMCOG scores were compared, the
between the assessments at Times and . In com- absence of a diagnosis of AD was correct.
parison, only one of the individuals without AD
at Time (%) had shown a decline of >. points Predictive validity
(see Table ). Informant interview-based diagnosis
Predicting cognitive decline
was shown to have a specificity of . [% confi-
dence interval (CI) = .–. using Wilson’s Only three of the nine participants who were diag-
method, Wilson ] when compared with objec- nosed with AD at Time were subsequently able to
tively measured decline (as described above). participate in cognitive assessment at Time . Five
Although sensitivity was calculated to be ., the had died and one was too ill to be assessed. The
small number of participants showing cognitive findings from this part of the study, although sup-
decline (n = ) is reflected in a wide confidence inter- portive of predictive validity of the diagnostic assess-
val (% CI = .–., Wilson ), making it ment, are limited because of the small numbers of
difficult to draw clear conclusions from this result. people diagnosed as having AD at Time that could
However, a decline of >. points on the CAMCOG be re-assessed at Time .
© Blackwell Publishing Ltd, Journal of Intellectual Disability Research , ‒
Journal of Intellectual Disability Research
617
S. L. Ball et al. • CAMDEX informant interview for use with adults with DS
Not AD at Time 1 45 7 – 52
AD at Time 1 – 4 5 9
Total 45 11 5 61
© Blackwell Publishing Ltd, Journal of Intellectual Disability Research , ‒
Journal of Intellectual Disability Research
618
S. L. Ball et al. • CAMDEX informant interview for use with adults with DS
the difficulty level of the CAMCOG entailed the against which subsequent decline can be measured.
exclusion of participants who did not score suffi- The valid and reliable diagnosis of dementia and the
ciently high above the floor of the test at baseline, exclusion of other possible causes of observed decline
thus reducing numbers further and limiting the are the starting point for the development of appro-
degree to which results can be generalized to the DS priate support for the person concerned and his/her
population as a whole. family or other carers, as well as for the initiation of
In terms of the agreement between independent treatment.
ratings of individual items, the informant interview
was shown to have excellent inter-rater reliability. We
did not examine inter-informant reliability, as in our Acknowlegements
earlier study we found that it was essential that the
informant had known the person for at least The most recent study has been funded by the
months and had direct experience of supporting Down’s Syndrome Association with a grant from the
that person over that time. Any informant without National Lotteries Community Fund. A.J.H. is sup-
this experience was found not to be able to answer ported by the Heath Foundation. We are most grate-
the questions reliably. We thus determined that the ful for their support and for the help we have received
‘gold standard’ for informants must be that stated from many people with DS and their families and
above. other carers.
As described earlier, the diagnostic procedure used
in this and our previous studies has consisted of two
stages, the first being the collection of relevant infor- References
mation (using the CAMDEX informant interview)
American Psychiatric Association () Diagnostic and Sta-
and the second being the rating of this information tistical Manual of Mental Disorder. American Psychiatric
against recognized diagnostic criteria. The CAM- Association, Washington DC.
DEX informant interview standardizes the history- Aylward E., Burt D., Thorpe L., Lai F. & Dalton A. ()
taking procedure, enabling the clinician to extract Diagnosis of dementia in individuals with intellectual dis-
information on changes in cognition and behaviour ability: report of the task force for development of criteria
that can then be rated against the standard sets of for diagnosis of dementia in individuals with mental retar-
dation. Journal of Intellectual Disability Research , –
criteria used for the diagnosis of dementia (and its
.
subtypes) or other mental disorders in the general
Deb S. & Braganza J. () Comparison of rating scales
population (i.e. ICD-, DSM-IV, Criteria for FTD, for the diagnosis of dementia in adults with Down’s
criteria for Lewy body dementia, CAMDEX criteria, syndrome. Journal of Intellectual Disability Research ,
or criteria for affective disorders, etc.). The infor- –.
mant-based diagnostic process is part of a compre- Evenhuis H. M. () Evaluation of a screening instru-
hensive clinical assessment that includes, where ment for dementia in ageing mentally retarded persons.
Journal of Intellectual Disability Research , –.
necessary, investigations that rule out disorders that
might mimic the decline seen in dementia (e.g. Gedye A. () Dementia Scale for Down’s Syndrome. Gedye
Research and Consulting, Vancouver BC.
hypothyroidism).
Holland A. J., Hon J., Huppert F. A. & Stevens F. ()
We recommend that where cognitive or behav-
Incidence and course of dementia in people with Down’s
ioural change affecting an older person with DS has syndrome: findings from a population-based study. Jour-
been observed, a proper assessment is essential. This nal of Intellectual Disability Research , –.
should include a full informant and patient interview Holland A. J., Hon J., Huppert F., Stevens F. & Watson P.
that identifies the nature and course of the change. It () Population-based study of the prevalence and pre-
should identify mental state abnormalities and social sentation of dementia in adults with Down’s Syndrome.
British Journal of Psychiatry , –.
factors that might inform the diagnostic process, and
Hon J., Huppert F. A., Holland A. J. & Watson P. ()
where clinically indicated, investigations to exclude
Neuropsychological assessment of older adults with
other possible causes of decline. A neuropsychologi- Down’s syndrome: an epidemiological study using the
cal assessment can identify the person’s cognitive Cambridge Cognitive Examination (CAMCOG). British
strengths and weaknesses and provide a baseline Journal of Clinical Psychology , –.
© Blackwell Publishing Ltd, Journal of Intellectual Disability Research , ‒
Journal of Intellectual Disability Research
620
S. L. Ball et al. • CAMDEX informant interview for use with adults with DS
Huppert F. A., Jorm A. F., Brayne C., Girling D. M., Roth M., Huppert F., Tym E. & Mountjoy C. ()
Barkley C., Beardsall L. & Paykel E. () Psychometric CAMDEX: The Cambridge Examination for Mental
properties of the CAMCOG and its efficacy in the diag- Disorders of the Elderly. Cambridge University Press,
nosis of dementia. Aging, Neuropsychology and Cognition Cambridge.
, –. Roth M., Tym E., Mountjoy C., Huppert F., Hendrie H.,
Landis J. R. & Koch G. G. () The measurement of Verma S. & Goddard R. () CAMDEX – a standar-
observer agreement for categorical data. Biometrics , dised instrument for the diagnosis of mental disorder
–. in the elderly with special reference to the early detection
Neri M., Roth M., Vreese L. P. D., Rubichi S., Finelli C., of dementia. British Journal of Psychiatry , –.
Bolzani R. & Cipolli C. () The validity of informant Wilson E. B. () Probable inference, the law of succes-
reports in assessing the severity of dementia: evidence sion, and statistical inference. Journal of the American
from the CAMDEX interview. Dementia and Geriatric Statistical Association , –.
Cognitive Disorders , –. World Health Organization () ICD-: International
Roth M., Huppert F., Mountjoy C. & Tym E. () Statistical Classification of Diseases and Related Health Prob-
CAMDEX-R: The Cambridge Examination for Mental lems. WHO, Geneva.
Disorder of the Elderly (Revised Edition). Cambridge
University Press, Cambridge. Accepted February
© Blackwell Publishing Ltd, Journal of Intellectual Disability Research , ‒