Functional Abilities and Cognitive Decline in Adult and

bs_bs_banner
Journal of Intellectual Disability Research

370
doi: 10.1111/jir.12113
volume 59 part 4 pp 370384 april 2015
Functional abilities and cognitive decline in adult and

aging intellectual disabilities. Psychometric validation
of an Italian version of the Alzheimers Functional
Assessment Tool (AFAST): analysis of its clinical
significance with linear statistics and artificial
neural networks
L. P. De Vreese,1 T. Gomiero,2 M. Uberti,3 E. De Bastiani,2 E. Weger,2 U. Mantesso2 &
A. Marangoni2
1 Local Health Agency, Dementia Project, Modena, Italy
2 ANFFAS Trentino Onlus, Trento, Italy
3 Sospiro Foundation, Cremona, Italy
Abstract
Purpose (a) A psychometric validation of an
Italian version of the Alzheimers Functional
Assessment Tool scale (AFAST-I), designed for
informant-based assessment of the degree of
impairment and of assistance required in seven
basic daily activities in adult/elderly people with
intellectual disabilities (ID) and (suspected)
dementia; (b) a pilot analysis of its clinical significance with traditional statistical procedures and
with an artificial neural network.
Methods AFAST-I was administered to the professional caregivers of 61 adults/seniors with ID with a
mean age ( SD) of 53.4 ( 7.7) years (36% with
Down syndrome). Internal consistency (Cronbachs
Correspondence: Dr Tiziano Gomiero, ANFFAS Trentino Onlus,
Project DAD, Trento, Italy (e-mail: tiziano.gomiero@anffas.tn.it).
coefficient), inter/intra-rater reliabilities (intraclass coefficients, ICC) and concurrent, convergent

and discriminant validity (Pearsons r coefficients)
were computed. Clinical significance was probed by
analysing the relationships among AFAST-I scores
and the Sum of Cognitive Scores (SCS) and the
Sum of Social Scores (SOS) of the Dementia Questionnaire for Persons with Intellectual Disabilities
(DMR-I) after standardisation of their raw scores in
equivalent scores (ES). An adaptive artificial system
(AutoContractive Maps, AutoCM) was applied to
all the variables recorded in the study sample,
aimed at uncovering which variable occupies a
central position and supports the entire network
made up of the remaining variables interconnected
among themselves with different weights.
Results AFAST-I shows a high level of internal
homogeneity with a Cronbachs coefficient of
2014 MENCAP and International Association of the Scientific Study of Intellectual and Developmental Disabilities and
John Wiley & Sons Ltd
volume 59 part 4 april 2015

371
L. P. De Vreese et al. Clinical significance and artificial neural networks
0.92. Inter-rater and intra-rater reliabilities were

also excellent with ICC correlations of 0.96 and
0.93, respectively. The results of the analyses of the
different AFAST-I validities all go in the expected
direction: concurrent validity (r = 0.87 with ADL);
convergent validity (r = 0.63 with SCS; r = 0.61 with
SOS); discriminant validity (r = 0.21 with the frequency of occurrence of dementia-related
Behavioral Excesses of the Assessment for Adults
with Developmental Disabilities, AADS-I). In our
sample age and gender do not correlate with the
scale and comparing the distribution of the
AFAST-I and DMR-SCS and DMR-SOS
expressed as ES, it appears that memory disorders
and temporal and spatial disorientation (SCS)
precede the loss of functional abilities, whereas
changes in social behaviour (SOS) are less specific
in detecting cognitive deterioration sufficient to
provoke functional disability and vice versa. The
results of AutoCM analysis reveal that the hub
(core) of the entire network is represented by the
functional domain personal/oral hygiene in the
entire study sample and use of toilet in a subgroup
of subjects who obtained an ES equal to 0 at
DMR-SCS.
Conclusions These results confirm the reliability
and validity of AFAST-I and emphasise the complexity of the relationship among functional status,
cognitive functioning and behaviour also in adults/
seniors with ID.
Keywords artificial neural networks, dementia in
Alzheimers disease, functional abilities, intellectual
disabilities
Introduction
In recent years the life expectancy of people with
intellectual disabilities (ID) has rapidly increased
and currently settles around an average of 66 years
(Croce 2007; Coppus et al. 2008). With advancing
age, together with an increasingly frequent
comorbidity (Sullivan et al. 2011) and the resultant
impact on the already precarious mental balance
deriving from life-span neuro-psycho-pathological
alterations, three types of problems in the diagnosis,
treatment and care of aging persons with ID can be
easily traced and predicted (De Vreese et al. 2009):
(a) decreased levels of personal autonomy and
social integration; (b) a further, progressive deterioration of cognitive functioning with repercussions in
almost all areas of adaptive behaviour; (c) the onset
or exacerbation of problem behaviours arising from
the combination of variables related to unstable or
deteriorating daily efficiency, to inadequate environmental support and to inter-current organic or
mental health problems.
On the axis III of psychiatric disorders which
includes mental retardation, the ICD-10 criteria
(World Health Organization 1992) for a clinical
diagnosis of primary dementia in persons with ID
are listed as follows: (a) a disturbance lasting more
than six months, of recent verbal and visual
memory together with a decline in other cognitive
domains with respect to a premorbid level, and
interfering significantly with personal activities of
daily living; (b) no medical history or other possible
objectifiable causes of cognitive impairment (e.g.
depression) or clouding of consciousness (delirium);
(c) a decline in emotional control or motivation that
can manifest itself through lability, irritability,
apathy and rudeness of social behaviour.
However, because (older) adults with ID decline
faster in cognitive, functional and adaptive skills
than the general population (Burt et al. 2005), the
clinical diagnosis of a dementing illness is not
always straightforward, especially in individuals with
Down syndrome (DS) (Prasher 2005).
Therefore, it would be very useful to assess a possible decrease in the level of personal efficiency in
the course of time, through the determination of a
baseline that delineates the highest level of lifespan
functioning reached by the person with ID in order
to more easily capture which cognitive deficits are
to be considered a primary outcome of agingdependent ID and which, instead, are signs of a
further impairment secondary to an (incipient)
neurodegenerative dementia that is associated by
definition with a slowly progressive loss of previously possessed skills.
Aims of the study

An accurate assessment of functional status, both in
the diagnostic work-up of dementia and in postdiagnosis (Glinas 2007) (e.g. monitoring of the
diseases progression, verifying the efficacy of pharmacological and non-pharmacological interventions,

372
measures to guide the organisation of services or

medical-legal aspects) are of fundamental importance. Therefore, measurement tools are warranted
that are designed not only for a specific disease such
as dementia (Glinas 2007; De Vreese et al. 2008)
but also for selected populations such as persons
with ID with that specific disease (Gomiero et al.
2006). None of the functional scales recommended
for dementia screening in ID by Aylward et al.
(1997) have been validated for the Italian population with ID, with the exception of the Vineland
Adaptive Behavior Scale (Sparrow et al. 1984,
Italian edition by Balboni and Pedrabissi, 2003)
which however predominantly probes behaviour and
has not been designed on purpose to document
functional decline over time in the ID population
with (suspected) dementia. A more recent and
monothematic scale that assesses specifically the
degree of functional decline in basic daily activities
in people with ID and (suspected) Dementia in
Alzheimers disease (DAD) is the Alzheimers
Functional Assessment Tool (AFAST, http://www
.emedicinehealth.com/alzheimers_disease_in__down
_syndrome/page6_em.htm#alzheimers_functional
_assessment_tool_toileting_dining_and_walking).
The primary purpose of this study is the psychometric validation of an Italian version of AFAST
(henceforth called AFAST-I) in terms of its internal
consistency, inter-rater and intra-rater (test-retest)
reliabilities, criterion-related and construct validity.
Secondly, we analysed the complex relationship
between cognition and functional abilities (Glinas
2007) after standardisation of the raw scores of
AFAST-I and the Italian version of the Dementia
Questionnaire for Persons with Intellectual Disabilities (DMR-I) (De Vreese et al. 2008), being aware
that the statistical and clinical significance, although
related, are two different constructs (Thompson
2003; Agnoli & Furlan 2008; Lancioni et al. 2008).
Furthermore, due to our limited ability to analyse
complex conditions, such as those relating to
human behaviour where a multiplicity of factors are
involved and where the relationships are mostly
non-linear, we submitted the data collected in this
sample to a particular artificial neural network,
called AutoCM, (see the Method section). This
method of analysis is able to highlight the interconnections between socio-demographic and clinical
variables otherwise undetectable with traditional
linear statistical methods (Gomiero et al. 2011) with

the aim of deepening our knowledge of the clinical
significance of AFAST-I.
Materials and methods

Study sample
Four organisations in the field of ID collaborated in
this study: the non-profit Italian parental association
of intellectual and relational disabilities of Trento
(ANFFAS Trentino Onlus), which directs community housing and day care centres around the province of Trento, the cooperative Social Laboratory,
which manages sheltered employment centres, the
Sospiro Foundation of Cremona and Villa Maria di
Rovereto, which directs residential services. All sites
were involved in a multi-centre multi-disciplinary
prospective project examining the effects of aging
and development of dementia on ID, called the
DAD project (http://www.validazione.eu/dad/). This
longitudinal project (Gomiero et al. 2006; De
Vreese et al. 2012) was approved by the Directory
Board of ANFFAS Trentino Onlus (Protocol no.
005932 of 15 September 2005).
Sixty-one adults/seniors with ID with a mean age
( SD) of 53.4 ( 7.71) years (with a range of 39 on
up to 64 years) were included in the study. Typological diagnosis of ID was principally done by
medical record review and examination of clinical
phenotype. All participants were fully or partly able
to speak and functionally sighted and hearing,
without current clinically relevant chronic psychiatric or organic comorbidity as documented by
medical records, and without a recent history of
acute organic (e.g. hospitalisation with or without
(sub-syndromal) delirium) (Marcantonio et al. 2003)
or socio-psychological (e.g. bereavement, relocation,
change in work activity) stressors. Explicit verbal,
but not written, informed consent was obtained
from those subjects who were able to consent.
Where such consent was not possible, family
members or guardians provided the subjects willingness to participate in the study. Table 1 summarises the socio-demographic characteristics and
some clinical conditions of the study participants.
The degree of severity of ID was defined according
to the DSM-IV-TR criteria (American Psychiatric
Association 2000) and diagnosis of dementia was

373
100
61
* Only for these cases we could collect clinical diagnosis.
DS, Down syndrome; non-DS, non-Down syndrome. Home, persons living with their family who attend day care or sheltered employment centers.
61
Total
100
39
Non-DS
64
done following the criteria proposed by Silverman

et al. (2004). As may be seen from Table 1 there
were no subjects who satisfied the diagnostic
criteria of a questionable dementia (i.e. substantial
uncertainty regarding dementia status, with some
indications of mild functional and cognitive declines
present). A clinical diagnosis of (suspected) dementia could not be made for the 13 residents of the
Villa Maria Nursing Home for organisational
problems.
Total
100
61
67
47
Female
Total
54
21
25
33
13
15
Nursing home
Community housing
Home
23
14
Male
86
4
8
2
100
41
2
4
1
48*
No dementia
Possible
Probable
Possible dementia with complications
Total
8
56
15
21
100
5
34
9
13
61
Profound (25)
Severe (2635)
Moderate (3649)
Mild (50)
Total
22
DS
36
Diagnosis of dementia
%
n
Level of ID
%
n
Diagnosis
Table 1 Study sample composition by diagnosis, level of ID gender and care setting
Gender
Care setting
Study procedure
AFAST-I was administered by raters with extensive
first-hand knowledge about aging ID subjects (physicians, professional educators, psychologists and
education experts). All raters attended a three-hour
briefing aimed at achieving a general consensus on
scale administration by practical examples before
the interviews. In order to be selected as informants (interviewees), professional caregivers needed
to be well acquainted with the subjects general
behaviour, and to have spent a sufficient amount
of time (night shift included) with the subject
during the two weeks prior to the AFAST-I assessment. The scale was administered in a structured
interview format in a quiet room and the average
administration time was about 10 min. The
respondent was given a copy of the scale so that
he/she could see the items that were read aloud by
the rater.
Inter-rater reliability was examined by assessing
the subjects level of efficiency in activities of daily
living twice on the same day by two independent
informants. To assess intra-rater reliability, the
interviewer submitted one of the two informants
again to the AFAST-I scale after two weeks from
the initial interview. A history of negative life events
since the initial interview was obtained at the time
of the second interview so as to identify possible
factors contributing to any significant change that
might have been observed in the AFAST-I scale.
To reduce transcription errors and counting,
interviewees responses were entered directly in
electronic format via a protected gateway to the
website (http://www.validazione.eu/dad), created
at purpose for data gathering from all the sites
involved in the DAD project and only accessible to
the authors of this study.

374
Instruments
Alzheimers Functional Assessment Tool
AFAST assesses some basic tasks deemed most
important for the quality of life of dementing
persons and outlines their typical decline in the
course of dementia due to Probable or Possible
Alzheimers disease both in the general population
(Glinas 2007) and in persons with ID (Burt et al.
2005). Each domain explored is graded on a Likert
scale of 67 unipolar points ranging from 0 (no particular difficulties present) to a maximum of 56
points (total dependence). The total score of the
scale is equal to 39 and is inversely proportional to
the degree of autonomy of the person seen in the
following basic functions: (a) use of toilet; (b)
dining; (c) walking/motor; (d) bathing; (e) dressing
(i.e. skills and appropriate dress); (f) personal/oral
hygiene (e.g. brushing hair and teeth, use of sanitary pads, shaving); (g) environmental awareness.
With the exception of the last two domains, all
others are similar to those included in the Katzs
Basic Activities of Daily Living scale (ADL) (Katz
et al. 1970), but unlike the ADL, items of the
AFAST domains are more numerous and allow a
greater level of detail and a more accurate staging,
thus permitting a more precise programming of care
plans, support or stimulation of individual procedural memory. Table 2 highlights the graded
items (in brackets the values of individual items)
that investigate personal/oral hygiene: a score of 0
indicates complete autonomy for example combing,
brushing teeth, shaving, use of sanitary pads, while
a score of 6 expresses complete dependence. The

intermediate scores 15 are obtained depending on
whether or not the ID person under examination
performs one or more of the illustrated tasks (with
hints, with verbal or gestural cues or physical guidance), thus detailing the current level of residual
abilities of the person in that precise functional
domain.
For the current study, an iterative process of
forward translations, back translations and evaluation of translation correspondence by a bilingual
expert was conducted to achieve conceptual equivalence between the original and Italian translation of
AFAST, according to the procedures recommended
for translation of psychometric scales (Steiner &
Norman, 1995).
Other instruments used for the validation study
of AFAST-I
For the verification of concurrent validity of
AFAST-I, ADL (Katz et al. 1970) and Instrumental
Activities of Daily Living (IADL) (Lawton & Brody
1969) scales were administered. Scores range from
0 to 6, and 0 to 8 (05 for males), respectively, with
higher scores reflecting higher level of autonomy.
The two scores of the 50-item DMR-I questionnaire a screening tool for dementia in the ID (De
Vreese et al. 2007) respectively, the Sum of Cognitive Scores (DMR-SCS: short- and long-term
memory, orientation to time and space) and the
Sum of Social Scores (DMR-SOS: verbal communication, mood, activities and interests, some prob-
Table 2 Example of AFAST: the seven levels of independence in personal/oral hygiene
1. Able to perform all personal hygiene tasks

2. Able to perform all personal hygiene tasks within regular routines, may show difficulty in performing tasks if routine is changed
(for example, hospitalised, moved)
3. Able to perform all personal hygiene tasks but requires occasional reminders from staff to complete the task
4. Able to perform personal hygiene tasks but requires frequent reminders from staff to complete the task, may need staff
guidance (verbal and point cues) in some parts of some tasks (for example, may forget steps), may still be proficient in one
area and lose ability in another area
5. Requires staff supervision (verbal and point cues) to complete some personal hygiene tasks and staff assistance (light, moderate
physical cues) to complete others
6. May still be able to perform some steps of some personal hygiene tasks with staff assistance but depends on staff to meet
other personal hygiene needs
7. Depends on staff to meet all personal hygiene needs
AFAST, Alzheimers Functional Assessment Tool.

375
lematic behaviours such as anger, delusional

thinking, etc.), were used to test the convergent
validity of AFAST-I. Higher SCS (range 044) and
SOS (range 066) represent a higher frequency of
behaviours that are considered to be indicative of
dementia. The discriminant validity of AFAST-I
was checked by the use of the sub-scale of an Italian
version of the Assessment for Adults with Developmental Disabilities (De Vreese et al. 2011). This
informant-based scale quantifies the frequency of
11 behavioural excesses which have occurred in the
last two weeks (e.g. day and/or night wandering,
agitation-aggression, disinhibition, uncooperative
behaviour, crying, aimless disruptive vocalisations),
rated on a scale from 0 (has not occurred in the
past two weeks) to 6 (more than once an hour/all
of the time) with a maximum total score of 66.
All these scales were administered independently
at the same time period to the entire study sample,
except for the sub-scale of AADS-I, which, due to
organisational problems, could be applied to only 41
subjects.
Statistical analyses
Data of the psychometric validation of AFAST-I
and of the deepening of its clinical significance were
analysed using SPSS (version 15.01 for Windows).
Regarding the latter aspect of the study, the raw
AFAST-I and DMR-I scores were standardised to
compare and obtain comparable data, as they are
scales with different ratios. As there are no specific
studies of normative samples that investigate the
functional abilities of adults/seniors with ID in Italy
(De Vreese et al. 2009; Gomiero et al. 2011), we
have not made use of percentile ranks, but of
equivalent scores (ES). The ES transform the raw
scores by dividing them into 5 equitably adjusted
regions, where at 95% probability, the worst 5% of
the population of reference is expressed as an ES of
0. Finally, we have processed all the data collected
in our sample, except for AADS-I and dementia
diagnosis for the reasons mentioned above, with a
data mining method, based on a particular artificial
adaptive system, the Auto Contractive Map
(AutoCM), developed at Semeion Research Center
(Buscema 2007), that is able to highlight any kind
of consistent patterns and/or systematic relationships and hidden trends and associations among
variables (i.e. in terms of many-to-many rather than

dyadic associations) even if obtained from a small
sample of subjects.
Results
Composition of the study sample
In the present sample none of the comparisons
between the variables displayed in Table 1 reached
a level of statistical significance, except for a notably
higher frequency of women (2 = 9.9, P = 0.003)
and for a higher predictable average age of
institutionalised persons compared with those
living in community housing or at home
(mean SD = 56.7 4.7 vs. 54.1 5.4 vs.
45.5 9.7, F2,58 = 16.6, P < 0.0001).
Descriptive analysis of AFAST-I

The average total score ( SD) of AFAST-I was
equal to 11.97 ( 9.95) with a range from 0 to 36
with no significant difference between males and
females, non-DS and DS, degree of ID severity,
and care settings. The analysis by ancova, taking
the sum of AFAST-I as a dependent variable, level
of ID severity as an independent variable, ID diagnosis as causal factor, and demographic data and
care setting as covariates, showed that the scale is
not significantly affected either by age (F1,56 = 2.52,
P = 0.118) or gender (F1,56 = 1.54, P = 0.219). The
independence of age and sex is also confirmed by a
statistical non-significance of the correlation coefficients (age: Pearsons r = 0.024, P = 0.854; gender:
Pearsons r = 0.053, P = 0.684). However, care
setting appears to be a significant covariate
(F1,51 = 6.18, P = 0.01). A similar ancova was also
performed taking as independent variable the presence or absence of cognitive impairment. Level of
ID was considered as a causal factor, and ID diagnosis, age, gender and care setting as covariates.
Mean ( SD) total AFAST-I score significantly differed among the 4 levels of dementia [no dementia:
12.9 ( 9.9) vs. possible dementia: 15.5 ( 11.4) vs.
definite dementia: 23.7 ( 7.4) vs. decline with complications: 10] with care setting resulting again a
significant covariate (F1,34 = 7.41, P = 0.014). Interestingly the interaction dementia diagnosis x degree

376
Table 3 Pearson correlation matrix among the single items and the total AFAST-I score
AFAST
Toileting
Dining
Walking
Bathing
Dressing
Hygiene
Awareness
Use of toilet
Dining
Walking/motor
Bathing
Dressing
Personal/oral hygiene
Awareness
Total score
1
0.54
0.65
0.61
0.75
0.62
0.58
0.82
1
0.47
0.55
0.64
0.53
0.62
0.76
1
0.69
0.64
0.60
0.43
0.78
1
0.73
0.82
0.57
0.87
1
0.76
0.71
0.90
1
0.51
0.85
1
0.76
AFAST-I, Italian version of the Alzheimers Functional Assessment Tool.
Table 4 Intraclass correlation coefficients for the single AFAST-I

items
AFAST-I domains
Inter-rater
reliability
Intra-rater
reliability
Use of toilet
Dining
Walking/motor
Bathing
Dressing
Personal/oral hygiene
Awareness
0.95
0.79
0.93
0.85
0.82
0.84
0.86
0.86
0.82
0.94
0.81
0.81
0.87
0.64
AFAST-I, Italian version of the Alzheimers Functional Assessment Tool.
of ID severity did not reach a statistical level of significance (F3,34 = 0.59, P = 0.627).
Internal consistency
Cronbachs for the total AFAST-I score was 0.92
(standardised). All items correlated with the total
score above the criterion value of 0.40 (Steiner &
Norman, 1995) (Table 3).
Reliability
Reliability was assessed with intraclass correlation
coefficients (ICC), which were equal to 0.96 and
0.93, respectively, for inter-rater reliability and testretest reliability (intra-rater). As shown in Table 4,
all ICC for the individual seven domains exceed the
criterion value of 0.70 (Steiner & Norman, 1995),
with the exception of the inter-rater reliability for

the domain environmental awareness.
Validity
A good concurrent validity was observed, respectively with both the ADL (Pearsons r = 0.87) and
the IADL (Pearsons r = 0.69) scales (Table 5).
The Pearson correlation coefficients among total
AFAST-I scores and the SCS and SOS of the
DMR-I (De Vreese et al. 2007), largely reaching
levels of statistical significance, are also indicative of
a good convergent validity of AFAST-I (Table 5).
Discriminant validity was tested by correlating
total AFAST-I scores with the frequency sub-scale
of behavioural excesses collected with the AADS-I.
As expected, the coefficient obtained between the
two measures was extremely low (Pearsons
r = 0.21, P = 0.183), confirming the specificity of
AFAST-I. Indeed, and as can be seen in Fig. 1,
there is a super imposable profile of co-variation of
the three measures used to check the convergent
validity of AFAST-I, in contrast to what happens
for the non-cognitive measure used to verify its discriminant validity (FAADS-I).
Insights into the clinical significance of AFAST-I

Analysis with traditional statistical procedures
By analysing and comparing the data collected for
the domains related to cognitive aspects (DMRSCS), social behaviour (DMR-SOS) and functional
status (AFAST-I), we asked ourselves whether it
was possible to identify other relationships among

377
AFAST-I
DMR-I SCS
DMR-I SOS
ADL
IADL
AFAST-I
DMR-I
SCS
DMR-I
SOS
ADL
IADL
1
0.63
0.61
0.87
0.69
1
0.75
0.63
0.67
1
0.57
0.54
1
0.70
Table 5 Pearson correlation matrix for

the concurrent and convergent validity of
AFAST-I
The correlation coefficient with the ADL and IADL scales are negative because scoring is
opposite to those of the AFAST-I and DMR-I. All correlation coefficients reach statistical
significance at the 0.01 level after Bonferronis correction.
AFAST-I, Italian version of the Alzheimers Functional Assessment Tool; DMR-I, Italian
version of the Dementia Questionnaire for Persons with Intellectual Disabilities; SCS, Sum
of Cognitive Scores; SOS, Sum of Social Scores. IADL, Instrumental Activities of Daily
Living; ADL, Activities of Daily Living.
Figure 1 Performance scores expressed in percentages in the different scales of each individual subject. FAADS, Frequency of occurrence
of Behavioral Excesses indexed by the Assessment for Adults with Developmental Disabilities Questionnaire; AFAST, Alzheimers
Functional Assessment Tool; SCS, Sum of Cognitive Scores; SOS, Sum of Social Scores; DMR, Dementia Questionnaire for Persons with
Intellectual Disabilities.

378
these three domains after standardisation of their

scores into ES in order obtain comparable data
(Table 6).
Figure 2 shows the distribution of the five ES
regions for AFAST-I, DMR-SCS and DMR-SOS.
For the sake of clarity, our analysis will focus only
on those subjects who obtained an ES equal to 0,
which, in clinical practice, is considered frankly
pathological.
As may be noted from Fig. 2 there is a higher
number of subjects (11 vs. 4) who obtained a pathological score in DMR-SCS than in AFAST-I, indicating on the one hand that dementia-related
cognitive decline generally precedes functional
Table 6 Intervals of conversion of the raw AFAST-I and DMR-I

scores in equivalent scores (ES)
Equivalent
scores
AFAST-I
DMR-I
SCS
DMR-I
SOS
4
3
2
1
0
11
12 < 18
18 < 25
25 < 32
32
23
24 < 28
28 < 33
33 < 38
38
15
16 < 27
27 < 39
39 < 51
51
AFAST, Alzheimers Functional Assessment Tool SCS, Sum of

Cognitive Scores; SOS, Sum of Social Scores; DMR, Dementia
Questionnaire for Persons with Intellectual Disabilities.
decline in basic activities (AFAST-I), and on

the other hand, that the occurrence of certain
(problem) behaviours is less specific in capturing
cognitive impairment leading to functional loss and
vice versa. This finding is also confirmed by the distribution pattern of the average raw AFAST and the
two DMR scores according to the five classes of ES
(Fig. 3), where a greater co-linearity of the trend
lines is present between AFAST-I and DMR-SCS
compared with DMR-SOS.
Analysis with artificial neural networks
As already outlined above, the rationale of using
an artificial neural network, such as AutoCM, is
related to the limits of adequately analysing data
in the presence of complex and mostly non linear
conditions, especially when derived from small
study samples. In non-technical terms, AutoCM
after the training phase the weights allow a direct
interpretation: specifically, they are proportional
to the strength of many-to-many associations
across all variables. This permits a further, useful
processing: association strengths may be easily
visualised by transforming weights into physical
distances, i.e. couples of variables whose connection weights are higher get relatively nearer, and
vice versa. By applying a simple mathematical filter
such a minimum spanning tree to the matrix of
distances, a graph is generated, whose use has
Equivalent scores
0
1
ES 0 = threshold where
5% of the worst scores
are located considered as
frankly pathological
2
3
4
0
10
15
20
25
30
35
Number of subjects (Total = 61)
AFAST
DMR SCS
DMR SOS
Figure 2 Frequency distribution of the equivalent scores in three

ID-specific measures. AFAST, Alzheimers Functional Assessment
Tool; SCS, Sum of Cognitive Scores; SOS, Sum of Social Scores;
DMR, Dementia Questionnaire for Persons with Intellectual
Disabilities.
Figure 3 Mean raw scores of the three ID-specific measures for

each of the four equivalent scores. AFAST, Alzheimers Functional
Assessment Tool; DMR, Dementia Questionnaire for Persons with
Intellectual Disabilities; SCS, Sum of Cognitive Scores; SOS, Sum
of Social Scores.

379
already been tested in the medical field (Buscema

& Grossi 2008; Buscema et al. 2008) and that is
termed connectivity map. This representation then
allows a very intuitive visual mapping of the
complex web of connection schemes among variables, and greatly facilitates the identification of
the variables that play a key role in the schemes,
i.e. that turn out to be hubs of the graph. The
AutoCM matrix of connections preserves non
linear associations, while at the same time capturing elusive connection schemes among clusters
that are often overlooked by traditional cluster
analyses, and pinpointing complex similarities
among variables on various dimensions role, connectivity, essentially, and so on. We are fully
aware of the fact that the AutoCM algorithms are
novel and therefore not entirely understood so far
in all of their properties and implications and that
further research is called for to explore them. But
at the same time we are convinced that their
actual performance in the context of well-defined,
well understood problem provides an encouraging
test to proceed in this direction. In this paper, we
introduce a methodology, which allows for basic
improvements in both robustness of use in badly
specified and/or computationally demanding problems, and output usability and intelligibility. In
particular, AutoCMs spatialise the correlation
among variables by constructing suitable embedding space where a visually transparent and
cognitively natural notion such as closeness
among variables reflects accurately their associations. The closeness (from a minimum of 0 that
indicates the absence of association, to a theoretical maximum of 1) can be converted into a compelling graph-theoretic representation that picks all
and only the relevant correlations and organises
them into a coherent picture. Such representation
is not actually constructed through some form of
cumbersome aggregation of two-be-two association
between couples of variables, but rather by building a complex global picture of the whole pattern
of variation. In this contest we cannot fully exploit
the topological meaningfulness of graph-theoretic
representations, but in that actual paths connecting
nodes (variables) in the representation, carry a
definite meaning in terms of logical interdependence in explaining the data sets variability. (for
further mathematical detail and another similar
application in the Alzheimer disease field, see also

Buscema et al. 2008).
We now report two brief examples which do not
have the claim to generalisation because significant
only for the present study sample. In any case, the
analysis of data with the AutoCM algorithm highlights a number of interesting findings.
The topological analysis of Fig. 4 allows us to
identify within the sparsely branched network
derived from the entire study population, a central
cluster including four variables (DMR-SCS, age
and two AFAST-I items dressing and bathing),
directly connected to the central node of the map,
represented by another AFAST-I item: personal/
oral hygiene is therefore a well-defined hub whose
association strengths are all above 0.9. Connections
weights with other variables such as ID diagnosis,
degree of ID severity and type of care setting are
located in the periphery and may therefore be considered marginal variables.
Figure 5 shows, still by way of example, the map
relative to the 11 subjects in our sample who
obtained the worst scores in DMR-SCS (i.e.
PE = 0). In this subsample, the key role played by
gender can be expected since nine of the 11 subjects
were females. A second hub that interconnects and
supports the network is represented by use of toilet
connected directly with three AFAST-I items,
namely personal/oral hygiene dining and dressing, but also with the variable DMR-SOS. The
association strengths among these five variables are
all very high. At difference with the AutoCM map
of the total study sample, ID diagnosis and level
of ID still mediated by gender (see above) get
relatively nearer the hub use of toilet indicating
their greater influential role in this connectivity
map.
Discussion
This study presents data from a psychometric validation of an Italian version of AFAST designed specifically to quantify the degree of disability in basic
daily activities in adults/seniors with ID and (suspected) dementia available in Italy.
The scale under examination evidences excellent
internal consistency, which is confirmed by itemtotal correlations with all coefficients well above the
380
Figure 4 Network of all the variables collected from the total study sample. ID diagnosis (DS vs. non-DS); Level of ID (profound, severe,
moderate, mild); Sex: female vs. male; Residence: three types of care setting; ADL, Activities of Daily Living; IADL, Instrumental
Activities of Daily Living; SCS, Sum of Cognitive Scores of DMR; SOS, Sum of Social Scores of DMR. Hygiene, personal/oral hygiene.
Numbers represent the strength of the connection with values ranging from 1 (maximum) to 0 (absent).
Figure 5 Network of the variables collected from the 11 subjects with a pathological DMR-SCS (ES = 0). For the legend, see Fig. 4.

381
criterion value of 0.40. This means that AFAST-I

appropriately measures different aspects of the same
construct, despite the presence of the domain environmental awareness, which is not usually included
in the scales that assess efficiency in basic daily
activities.
AFAST-I also shows excellent inter- and intrarater reliabilities, respectively indicative of a good
concordance of interpretation of the degree of functional dependency between two independent (professional) informants and of its stability over a short
interval of time. The finding that ICC for the item
environmental awareness in the test-retest reliability is slightly below the criterion value of 0.70 presumably derives from a marked variability of the
level of infra- and interdian alertness which is frequent in subjects with ID, especially if suffering
from (very) severe dementia, even in the absence of
a clear delirium. Therefore, the scale AFAST-I
proves to be statistically homogeneous and reliable,
and among other things not influenced by confounding variables such as age and gender.
The analysis to verify the statistical validity of
AFAST-I has also produced excellent results, and
all in the expected direction. A high and statistically
significant correlation coefficient (Pearsons
r = 0.87) between AFAST and a measure considered a gold standard such as the ADL (Katz et al.
1970), supports the criterion-related validity (concurrent) of AFAST. Significant correlations among
AFAST-I scores and the two sub-scales of the
DMR-I (SCS: Pearsons r = 0.63; SOS: Pearsons
r = 0.61), which measure cognitive function and
social behaviour in people with ID and (suspected)
dementia, substantiate its convergent validity. The
correlation coefficient between DMR-SCS and
AFAST is also in line with the literature, which
reports a strong correlation between cognitive function and functional status only in the advanced
stages of Alzheimers dementia (Reed et al. 1989;
Auer et al. 1994). By contrast, a low and nonsignificant correlation coefficient (Pearsons
r = 0.21) between AFAST-I and a measure of a
completely different construct such as the frequency
of positive behavioural disorders, as indexed by the
AADS, confirms the discriminant validity of the
scale in question. This low correlation between
functional status and dementia-related problem
behaviour is similar to that observed in the general
population with (Alzheimers) dementia (Glinas

2007). Therefore, from these data we can deduce
that AFAST-I has a good statistical validity both in
terms of criterion-related (concurrent) and construct (convergent and discriminant) validities.
The results obtained by a second level analysis of
the clinical significance of AFAST-I by means of
both traditional statistical methods and a novel
algorithm of ANN, deserve some comments.
After transforming the raw scores of the three
main measures (AFAST-I, SCS and SOS of the
DMR-I) of the study into ES, two major elements
emerged:
1 it appears clearly from Fig. 2 that a cognitive disturbances as indexed by the DMR-SCS, is not necessarily associated with severe functional disability.
Since the 11 subjects (of which the majority has a
diagnosis of DS) with PE = 0 on the DMR-SCS
scale had no comorbid organic or sensory deficits
limiting the efficiency in performing basic tasks, we
can infer that their cognitive impairment precedes
the loss of the basic functional abilities in a much
similar way to what happens in the general population suffering from Alzheimers dementia, in agreement with the retrogenesis theory of Alzheimers
dementia (Auer et al. 1994; Reisberg et al. 2002;
Glinas 2007);
2 the finding that no subject obtained a PE equal to
0 on the DMR-SOS, emphasises that the presence
of language problems, mood disorders, apathy, and
some behavioural problems, is not always indicative
of a dementing illness leading to functional disability and vice versa. As a matter of fact, Fig. 3 clearly
shows that the trend lines of AFAST-I distributed
along the five PE regions, are much more in line
with the index of cognitive functioning (SCS) than
that of social behaviour (SOS), in accordance also
with the abovementioned low correlation coefficient
between AFAST-I and FAADS.
This comparative analysis of PE between cognitive,
behavioural and functional scales has both theoretical and practical implications as it may make an
overall comparison among these domains possible,
considered key domains in dementia, ameliorating
diagnostic work-up and post-diagnosis staging and
rendering (rehabilitation) intervention scheduling
more easily and presumably more efficacious.
As mentioned above, the AutoCM, is an algorithm that allows us to re-design the structure of the

382
variables that are most connected and that become

the focal nodes of functional abilities. Figure 4
shows how variables that affect the functional abilities are mutually linked in a network of nonrandom connections, in which the most significant
value is the one related to the functional domain
personal/oral hygiene and which therefore should
be monitored/supported/reactivated with great care
and intensity in ageing ID. Instead, variables such
as gender, care setting, and instrumental daily skills
(IADL) and even level of ID severity weigh less (i.e.
are less relevant) at least in the present sample, and
they would thus seem to be less predictive of a possible future functional decline. The lack of predictability of IADL is probably due to the fact that they
are very often compromised lifespan because of the
ID.
In the 11 subjects who scored pathologically in
the DMR-SCS, the AFAST item use of toilet
appears as another focal point (hub). The diagnosis
of ID (in particular DS) and social behaviour
(DMR-SOS) also acquire a greater weight within
the network. We emphasise that the application of
this algorithm, in particular, and artificial neural
networks in general, is intended to be only an
exemplification of the future possibilities of analysis
that can come to distinguish, both individually and
in groups, the weights of different (apparently
independent) variables. This novel way of analysing
data may allow us to develop more appropriate
projects of assistance or of reactivation at the
individual level or in small groups. Planning
interventions based on criterial surveys that can
simultaneously take into account a vastly higher
number of factors, seems to us more appropriate
and very fruitful for progress in all areas of complex
design, especially where clinical interpretation plays
a crucial role. Direct clinical experience, competence, professionality and evidenced-basedmedicine, while remaining important, are based on
indices of central tendency, which, at times, are
unable to grasp the individual weight that is
hidden in the available data (Gomiero et al.
2011).
Some caveats should be noted. First, the sample
size is small and a replication of the findings in a
larger group with a more proportionate distribution
not only of gender, DS and non-DS adults and
seniors but also of ID individuals with and without
definite dementia, is needed. A larger sample size

will help to better understand whether AFAST may
consistently distinguish DAD patients from nondementing DS subjects. Second, diagnosis of
dementia was made clinically. Because of low
specificity of the actual diagnostic criteria we
could not exclude co-occurrence of degenerative
conditions or cerebral vascular suffering other
than DAD.
Conclusions
The evaluation of the functional status in subjects
with ID and dementia is crucial both in the diagnostic and post-diagnostic phase. AFAST-I seems
to fulfil the necessary requirements to become a tool
of choice for the evaluation of the basic tasks in
adults/seniors with ID and (suspected) dementia.
First, the loss of dependency of each functional
domain explored reflects the typical hierarchical
decline in the course of DAD. Secondly, the administration is of short duration and easily understood
by the professional carers. In addition, the scale
assesses not only the presence or absence of a loss
of some daily basic abilities, but it contemplates
several difficulty levels of autonomy based on the
need for reminders, verbal or gestural suggestions or
physical guidance, thus providing useful information
to set up formal or informal interventions aimed at
reactivating (e.g. excess disability due to organic or
psychiatric comorbidity or to inappropriate care
setting) or at maintaining over time residual abilities. Finally, the AFAST-I appears to be a valuable
tool and reliable from a psychometric point of view.
Further studies are needed to test AFASTs ability
to capture accurately disease progression in a longitudinal perspective and to be just as useful for
evaluating the efficiency and effectiveness of pharmacological treatments or non-pharmacological
interventions.
Acknowledgements
The authors thank professor Ricard Straub for his
valuable linguistic advice, ANFFAS Trentino
Onlus, the Social Cooperatives Villa Maria and
Laboratorio Sociale, and the Foundation Sospiro,
for their kind collaboration. A special thank also

383
goes to Semeion Research Center (and Dr Enzo

Grossi) who kindly granted the use of the software
of AutoCM algorithms.
Conflict of interest
None
References
Agnoli F. & Furlan S. (2008) La differenza che fa la
differenza: dalla significativit statistica alla significativit
pratica [The difference that makes the difference: from
statistical significance to practical significance].
Psicologia Clinica e dello Sviluppo 2, 21146.
American Psychiatric Association (2000) Diagnostic and
Statistical Manual of Mental Disorders Text Revised, 4th
edn. American Psychiatric Association, Washington,
DC.
Auer S. R., Sclan S. G., Yaffee R. A. & Reisberg B.
(1994) The neglected half of Alzheimers disease: cognitive and functional concomitants of severe dementia.
Journal of American Geriatric Society 42, 126672.
Aylward E. H., Burt D. B., Thorpe L. U., Lai F. &
Dalton A. J. (1997) Diagnosis of dementia in individuals
with intellectual disability. Journal of Intellectual Disability Research 41, 15264.
Burt D. B., Primeaux-Hart S., Loveland K. A., Cleveland
L. A., Lewis K. R., Lesser J. et al. (2005) Aging in
adults with intellectual disabilities. American Journal of
Mental Retardation 4, 26884.
Buscema M. (ed.) (2007) Squashing theory and contractive map network, semeion technical paper #32, Rome.
Buscema M. & Grossi E. (2008) The semantic connectivity map: An adaptive self-organizing knowledge discovery method in data bases. Experience in gastroesophageal reflux disease. International Journal of Data
Mining and Bioinformatics 2, 362404.
Buscema M., Grossi E., Snowdon D. & Antuono P.
(2008) Auto-contractive maps: an artificial adaptive
system for data mining. An Application to Alzheimer
Disease. Current Alzheimer Research 5, 4819.
De Vreese L. P., Mantesso U., Scarazzini M., Menegatti

C. & Gomiero T. (2007) Dementia Questionnaire for
Persons with Intellectual Disabilities (DMR): uno studio
di attendibilit della versione italiana della scala [Questionnaire for Persons with Intellectual disabilities
(DMR): a reliability study of the Italian version of the
scale]. Psicogeriatria 2, 1927.
De Vreese L. P., Caffarra P., Savar R., Cerutti R.,
Franceschi M., Grossi E. et al. (2008) Functional Disability in Early Alzheimers disease A validation study
of the Italian Version of the Disability Assessment for
Dementia Scale. Dementia and Geriatric Cognitive Disorders 25, 18694.
De Vreese L. P., De Bastiani E., Mantesso U. & Gomiero
T. (2009) La Nuova Longevit della Disabilit Intellettiva.
[The New Longevity of Intellectual Disabilities.]. Liguori
Editori, Naples.
De Vreese L. P., Mantesso U., De Bastiani E., Marangoni
A. C. & Gomiero T. (2011) Psychometric evaluation of
the Italian version of the AADS questionnaire: a
caregiver-rated tool for the assessment of behavioural
deficits and excesses in persons with intellectual disabilities and dementia. International Psychogeriatrics 23,
112432.
De Vreese L. P., Mantesso U., De Bastiani E., Weger E.,
Marangoni A. & Gomiero T. (2012) The impact of
dementia-derived non-pharmacological intervention procedures on cognition and behaviour in older adults with
intellectual disabilities: a three-year follow-up study.
Psychosocial and environmental interventions in older
adults with intellectual disabilities. Journal of Policy and
Practice in Intellectual Disabilities 9, 92102.
Glinas I. (2007) Functional autonomy. In: Clinical Diagnosis and Management of Alzheimers Disease, 3rd edn
(ed. S. Gauthier), pp. 17788. Informa Health, London.
Gomiero T., De Vreese L. P. & Mantesso U. (2006)
Diagnosi, intervento e gestione centrata sulla persona:
adulti e anziani con sindrome di Down e altra disabilit
intellettiva [Diagnosis, intervention and person-centered
care: adults and elderly with Down syndrome and other
intellectual disabilities]. Spazi e Modelli Percorsi Evolutivi
per la Disabilit Psicofisica 3, 2538.
Gomiero T., Croce L., Grossi E., De Vreese L. P.,
Buscema M., Mantesso U. et al. (2011) A short version
of SIS (Support Intensity Scale): the utility of the application of Artificial Adaptive Systems. US-China Education Review 2, 196207.
Coppus A. M. W., Evenhuis H. M., Verberne G.-J.,

Visser F. E., Oostra B. A., Eikelenboom P. et al. (2008)
Survival in elderly persons with Down syndrome.
Journal of American Geriatric Society 56, 231116.
Katz S., Down T. D., Cash H. R. & Grotz R. C. (1970)

Progress in the development of the index of ADL. The
Gerontologist 10, 2030.
Croce L. (2007) Lanziano con Disabilit Intellettiva: il

rischio di deterioramento cognitivo e le possibili
strategie dintervento. [Elderly with intellectual disabilities: risk of cognitive decline and possible intervention
strategies.]. American Journal of Mental Retardation 5,
35764. (Italian edition).
Lancioni G., Groeneweg J., Bosco A. & Basili G. (2008)

Significativit statistica e significativit clinica in ricerca
applicata con persone con disabilit gravi e multiple
[Statistical significance and clinical significance in
applied research with persons with severe and multiple
disabilities]. Psicologia Clinica e dello Sviluppo 3, 26172.

384
Lawton M. P. & Brody E. M. (1969) Assessment of older

people: self-maintaining and instrumental activities of
daily living. The Gerontologist 9, 17986.
Marcantonio E. R., Simon S. E., Bergmann M. A., Jones
R. N., Murphy K. M. & Morris J. N. (2003) Delirium
symptoms in post-acute care: prevalent, persistent, and
associated with poor functional recovery. Journal of the
American Geriatrics Society 51, 49.
Prasher V. P. (2005) Alzheimers disease and Dementia in
Down Syndrome and Intellectual Disabilities. Radcliff Publishers, Abingdon.
Reed B. R., Jagust W. J. & Seab J. (1989) Mental status as
a predictor of daily function in progressive dementia.
The Gerontologist 29, 8047.
Reisberg B., Franssen E. H., Souren L. E., Auer S. R.,
Akram I. & Kenowsky S. (2002) Evidence and mechanisms of retrogenesis in Alzheimers and other
dementias: management and treatment import. American Journal of Alzheimers Disease and Other Dementias
17, 20212.
Silverman W., Schupf N., Zigman W., Devenny D.,
Miezejeski C., Shubert R. et al. (2004) Dementia in
adults with mental retardation: Assessment at a single
point in time. American Journal of Mental Retardation

109, 11125.
Sparrow S. S., Balla D. A. & Cicchetti D. V. (1984) Vineland Adaptive Behavior Scales. American Guidance
Service, Circle Pines, MN. (Italian edition by G.
Balboni & L. Pedrabassi. Florence: Organizzazioni
Speciali, 2003).
Steiner D. L. & Norman G. R. (1995) Health Measurement
Scales: A practical Guide to Their Development and Use.
Oxford Medical Publications, Oxford.
Sullivan W. F., Berg J. B., Bradley E., Cheetham T.,
Denton R., Heng J. et al. (2011) Primary care of adults
with developmental disabilities. Canadian Family Physician 52, 54153.
Thompson B. (2003) Statistica, pratica, clinica: quanti tipi
di significativit deve considerare chi opera nel counseling? [Statistics, practice, clinics: how many types of significance should be considered in counselling?].
Bollettino di Psicologia Applicata 240, 313.
World Health Organization (1992) The ICD-10 Classification of Mental and Behavioural Disorders. Diagnostic Criteria for Research. WHO, Geneva.
Accepted 25 November 2013
This document is a scanned copy of a printed document. No warranty is given about the
accuracy of the copy. Users should refer to the original published version of the material.

Functional Abilities and Cognitive Decline in Adult and

Transféré par

Informations du document

Titre original

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Functional Abilities and Cognitive Decline in Adult and

Transféré par

Droits d'auteur :

Formats disponibles

bs_bs_banner

Journal of Intellectual Disability Research

volume 59 part 4 pp 370384 april 2015