Clinical Neuropsychiatry (2014) 11, 3, 103-107
THE FIVE DIGITS TEST ON THE ASSESSMENT OF PSYCHIATRIC PATIENTS
WITH HETEROGENEOUS EDUCATIONAL BACKGROUNDS:
EVIDENCES OF VALIDITY ON THE ASSESSMENT OF BIPOLAR DISORDER
Jonas Jardim de Paula, Suzana Abrantes, Fernando Silva Neves, Leandro Fernandes Malloy-Diniz
Objective: to analyze the validity of the Five Digits Test (FDT), an adapted version of the Stroop Color-Word Test
(SCWT) on Bipolar Disorder patients (BD).
Method: we assessed 35 BD patients diagnosed in a clinical interview by a psychiatrist using the MINI-Psychiatric
Interview. The participants were euthymic during the cognitive assessment. Besides the FDT, the participants performed
the SCWT, the Five Points Test (FPT) and the Rey Auditory Verbal Learning Test (RAVLT) as measures of criterion-
related, convergent and divergent validity, respectively. We adopted partial correlations, controlling for age, education,
sex and symptoms of depression and mania, to analyze the association between the FDT times and errors with the other
cognitive measures. Group comparisons with a paired control sample were used to assess the FDT discriminant validity.
Results: we found signicant correlations between the FDT and the SCWT and between the FDT and the FPT,
but not between the FDT and the RAVLT. The shared variance of the measures sustained the construct validity for the
assessment of processing speed and executive functions. Group comparisons indicate large effect sizes for the FDT
timed measures and errors in the executive functions components. The test accuracy for group classication was 90%.
Conclusions: the results suggest that the FDT is a valid measure of processing speed and executive functions for the
assessment of BD patients, and might be an alternative to the SCWT for patients with low formal education.
Key words: Five Digits Test (FDT), Stroop Color-Word Test (SCWT), Bipolar Disorder patients (BD), MINI-
Psychiatric Interview, Rey Auditory Verbal Learning Test (RAVLT)
Declaration of interest: none
Jonas Jardim de Paula1,2,3, Suzana Abrantes4, Fernando Silva Neves2,5, Leandro Fernandes Malloy-Diniz2,3,5
1 Departamento de Psicologia, Faculdade de Ciências Médicas de Minas Gerais, Belo Horizonte, Brazil.
2 INCT de Medicina Molecular, Faculdade de Medicina, Universidade Federal de Minas Gerais, Av Alfredo Balena, 190, Belo
Horizonte-MG, CEP 30130-100, Brazil.
3 Laboratório de Investigações em Neurociência Clínica (LINc) – Universidade Federal de Minas Gerais, Belo Horizonte, Brasil.
4 Private Practice
5 Departamento de Saúde Mental, Faculdade de Medicina, Universidade Federal de Minas Gerais, Belo Horizonte, Brasil.
Jonas Jardim de Paula
Avenue Alfredo Balena, 190, Belo Horizonte-MG, CEP 30130-100, Brazil. Telephone/Fax: 55-31-9990-2760.
The assessment of processing speed and executive
functions in Bipolar Disorder (BD) is an important
aspect on their clinical evaluation. Decits in selective
attention (a component of one Executive Function
called “Inhibitory Control”) is related to several clinical
outcomes including impairments in everyday life
activities (O’Shea et al. 2010), psychosocial functioning
(Solé et al. 2012), and suicide behavior (Malloy-Diniz
et al. 2011). One of the most used paradigms for this
cognitive function is the interference control, assessed
by Stroop-like tests, as the Stroop Color-Word Test
(SCWT). It involves the time and errors in colour naming,
naming the colours of graphed words and naming the
colours of graphed colour-names in incongruent (Red-
Green) conditions. The rst two components are usually
related to processing speed while the last is related to
executive functions (Troyer et al. 2006).
The SCWT has several limitations in the assessment
of patients with lower formal education, learning
disabilities or non-clinical difculties in word reading
(Cox et al. 1997). The prociency in reading is an
important factor for the interpretation of the SCWT.
In Brazil, the educational background of most adults,
especially those in public health care units is often
decient, with difculty in reading uency and
comprehension of small texts or even isolated words.
Submitted may 2014, accepted may 2014
FirSt epub ahead oF print 2014 may cn100006
© 2014 Giovanni Fioriti Editore s.r.l. 103
Jonas Jardim de Paula et al.
104 Clinical Neuropsychiatry (2014) 11, 3
An alternative to the classical SCWT is the numeric
or counting versions of the test. These are Mini-Verbal
or Language-Free versions of the SCWR: they use
a very simple content (number, colours or gures)
but the same design as classical tests do (Sedó and
The Five Digits Test (FDT) is a numeric-Stroop
test, which is relatively independent of reading
routines (Sedó 2007). The test involves four different
components. On the rst one, the patient must read one
out of ve numbers in a congruent condition (“1”, he/
she must say “one”; or “3-3-3”, he/she must say three),
exposed on small squares (FDT Reading). On the
second component, he/she must tell how many asterisks
(1 to 5) are in each square (“*”, he/she must say “one”;
or “***” he/she must say “three”) (FDT Counting).
The third component involves an interference
condition where one to ve numbers are exposed in
an incongruent condition (“2-2-2” or “5-5-5-5”), and
the subject must not say the name of the number, but
tell how many numbers are in each stimulus (in the
example, for “2-2-2” he/she must say “three” and for
“5-5-5-5” he/she must say “four”) (FDT Choosing).
The last component is similar to the third one, but for
every ve stimuli there is one marked square where
he/she must not say how many numbers are in each
square, but read one of the numbers (FDT Shifting).
The time and errors of each of the four conditions are
used as measures of “automatic attentional processes”
(Reading and Counting) and “controlled attentional
processes” (Choosing and Shifting). The two rst
components are more related to simple attention and
processing speed, while the two last are more dependent
on selective attention and executive functions.
Although the test is a promising measure of these
constructs for the Brazilian population, we found
no studies assessing its psychometric properties
in psychiatric patients. In this brief report, we aim
to assess the FDT construct, criterion-related and
discriminant validity in BD patients.
We assessed 35 euthymic BD patients from a
specialized health unit of mental health in a university
hospital. We performed the diagnosis based on the
DSM-IV-TR criteria by a psychiatric interview
conducted by a certied clinical psychiatrist. On
this assessment, the professional also applied the
Portuguese version of the MINT International
Neuropsychiatric Interview (Amorim 2000). The mean
age of the patients was 46.21 (±13.12) years and 71%
had eleven or less years of formal education. The BD
patients had been euthymic for at least four weeks
according to the clinical interview. The mean Beck
Depression Inventory (Beck et al. 1961) score was 8.83
(±2.44) and the Young Mania Rating Scale (Young et
al. 1978) was 6.26 (±2.38), scores not suggestive of
prominent symptoms of depression or mania during
the neuropsychological assessment, characterizing the
euthymic phase of BD.
We used four different neuropsychological
measures. Our interest was to assess the FDT construct
and criterion-related validity, so we chose a criterion
measure, a convergent measure (another test of
processing speed and executive functions) and a
divergent measure (a test of another cognitive domain).
Criterion-Related validity: For this purpose, we
selected the SCWT Victoria version (Troyer et al.
2006). The test has three components: Box, Neutral
Words and Colour Words. The executive demands of
the tasks increase along their three parts. We used the
SCWT times to assess the criterion-related validity of
the FDT timed measures and the total errors of each
SCWT component for the FDT error measures.
Convergent validity: We used the Five Points
Test (Regard et al. 1982) for the assessment of this
psychometric property. The Five Points Test is a non-
verbal uency task where the patient must draw a
series of gures using straight lines in several squares
containing ve dots. The test demands, besides the
motor uency, planning, monitoring and cognitive
shifting, in order to produce new drawing but avoid
errors or repetitions. We selected two measures from
this test: the total of correct drawings as a convergent
measure for the FDT times (a more pure measure of
uency and processing speed) and the total number of
errors as a convergent measure for the FDT errors (a
more pure measure of executive functions, including
inhibitory control and cognitive shifting).
Divergent Validity: For the assessment of this
psychometric property, we selected the Brazilian
Portuguese version of the Rey Auditory-Verbal
Learning Test (Malloy-Diniz et al. 2007). This test is
a classical measure of episodic memory, validated for
the Brazilian adult population. As a general estimate
of episodic memory, we selected the sum of words
(RAVLT Total) of the ve learning trials.
Discriminant validity: For the assessment of
this psychometric property, we assessed a paired
(1:1) control group without evidences of cognitive,
psychiatric or neurological disorders. The control group
was a convenience sample of healthy adults matched
to the BD patients in the distribution of sex (χ²=0.08,
p=0.771), age (χ²=0.56 p=0.757) and formal education
(χ²=0.58, p=0.445). The control group did not report
psychiatric disorders or neurological diseases and had
scores on the Standard Raven Progressive Matrices
Test (Raven, 1981) above the percentile 10 according
to the Brazilian population norms.
We explored the relationship of the FDT measures
with the other tests computing partial correlations
of the neuropsychological measures controlling
for the effects of age, education and symptoms of
depression and mania. We selected this procedure
since sociodemographic variables may inuence
neuropsychological test performance and might
inate the correlations of the cognitive constructs.
Measures that do not follow a normal distribution
underwent logarithmic, square root, square or cubic
transformations to better t the analysis assumptions.
Since multiple correlations were computed and the
sample size was relatively small, a more conservative
p-value was selected to avoid type-I errors (p=0.01). The
BD and control groups were compared by independent
samples t-tests, and the Cohen’s d was adopted as a
measure of effect sizes. We also performed a stepwise
binary logistic regression model to analyze the FDT
accuracy in the group classication. The test times and
errors were entered as independent predictors.
The Five Digits Test in Bipolar Patients
Clinical Neuropsychiatry (2014) 11, 3 105
timed measures with the SCWT Color-Words time were
signicant, and the shared variance ranged from 33%
to 57%. The pattern of errors was more specic, with
signicant correlations between the FDT Reading errors
only with SCWT Dots and Neutral Words (30 and 38%
of shared variance, respectively). FDT Counting errors
had no correlation with the criterion measure. The
executive components’ errors of the FDT (Choosing and
Shifting) were only associated with the SCWT Color-
Word component (69% and 49% of shared variance,
respectively). For the convergent and divergent validity
The BD participants description is shown on
table 1. Table 2 shows the correlations of the
neuropsychological measures divided by the validity
assessed (criterion-related, convergent and divergent).
On the criterion-related validity analysis, we found
signicant correlations between the FDT automatic
components times (Reading and Counting) with all
the timed measures of the SCWT, ranging from 22%
to 57% of shared variance. The correlations of all FDT
Table 1. Bipolar Disorder patient’s description
Participants description % or M(SD)
Gender Male 20%
Age Group 18 to 35 years 29%
36 to 50 years 26%
51 to 65 years 46%
Formal Education Less than 12 71%
12 years or more 29%
Marital Status Single 51%
Beck Depression Inventory 8.83 (2.44)
Young Mania Rating Scale 6.26 (2.38)
SCWT 1 (Time) 20.38 (5.67)
SCWT 2 (Time) 26.62 (10.76)
SCWT 3 (Times) 39.74 (18.66)
SCWT - Interference Score 16.24 (13.97)
SCWT 1 (Errors) 0.21 (0.73)
SCWT 2 (Errors) 0.15 (0.70)
SCWT 3 (Errors) 1.94 (4.68)
Five Points Test (Unique Drawings) 19.63 (2.21)
Five Points Test (Errors) 0.87 (2.21)
RAVLT (Sum of Words) 40.94 (9.79)
M: Mean, SD: Standard-Deviation, SCWT: Stroop Color-Word Test, RAVLT: Rey Auditory-Verbal Learning Test
Table 2. Partial correlations and shared variance (r²) between the Five Digits Test with the Stroop Color-Word,
Five Points Test and Rey Auditory Verbal-Learning Tests
FDT Times SCWT Dots SCWT
Neutral-W SCWT Color-W Five Points Test RAVLT
Reading (T)¹ 0.579* (34%) 0.659* (43%) 0.754* (57%) -0.471* (22%) -0.164 (3%)
Counting (T)¹ 0.554* (31%) 0.715* (51%) 0.612* (37%) -0.440* (19%) -0.047 (0%)
Choosing (T)¹ 0.394 (16%) 0.474* (22%) 0.573* (33%) -0.238 (6%) -0.204 (4%)
Shifting (T)¹ 0.393 (15%) 0.476* (23%) 0.528* (28%) -0.177 (3%) -0.082 (1%)
Reading (E)² 0.550* (30%) 0.618* (38%) 0.38 (14%) 0.508* (26%) 0.034 (0%)
Counting (E)² 0.375 (14%) 0.408 (17%) 0.315 (10%) 0.491* (24%) -0.121 (1%)
Choosing (E)² 0.117 (1%) 0.559* (31%) 0.831* (69%) 0.746* (56%) 0.114 (1%)
Shifting (E)² 0.119 (1%) 0.146 (2%) 0.698* (49%) 0.526* (28%) 0.097 (1%)
Note: 1 – Correlations with SCWT times and Five Points Test unique drawings. 2- Correlations with SCWT errors and Five
Points Test errors. * p<0.01
FDT: Five Digits Test, T: Times, E: Errors, SCWT: Stroop Color Word Test, W? words, RAVLT: Rey Auditory-Verbal Learning
Test sum of words
Jonas Jardim de Paula et al.
106 Clinical Neuropsychiatry (2014) 11, 3
on the convergent task. Finally, we found a divergent
pattern of association between the FDT and the RAVLT,
an evidence of divergent validity. Signicant group
differences were found for the FDT times (with very
large effect sizes), for FDT Choosing and Shifting
Errors (with large effect sizes), but not for Reading and
Counting errors. In a binary logistic regression model,
the FDT Shifting time predicted group membership
with high accuracy (90%).
The results are convergent with other research
ndings, which examined the association of the
FDT measures with other cognitive tests. Sedó and
DeCristoforo (2001) revised correlational studies of
the FDT with other measures and found a signicant
association with the SCWT. The effect sizes reported
by the authors were very similar to ours, suggesting
a transcultural validity. Sedó (2007) also reviewed
other correlational studies and reported signicant
correlations with other attentional and executive
functions tests with moderate or high effect sizes. In
the Brazilian population, Magalhães (2013) examined
the association of the FDT with other measures related
to inhibitory control in healthy adults. Her results
indicated that the FDT along with the SCWT and other
tests of attention composed a latent factor related to
interference control, a subcomponent of the inhibitory
control system. A study with mood disorders carried
in older adults with major depressive disorders also
revealed signicant group differences between patients
and controls, with moderate to high effect sizes (de
Paula et al. 2011).
This study has important limitations, which we
must address. We had a small sample size. Although
the participants were well characterized and some
constraint were adopted for the statical analysis, the
sample size is not sufcient to identify associations
with small effect sizes. In addition, other factors, which
may have inuenced the results, such as premorbid
IQ, number of BD episodes, reading and counting
prociency, were not controlled. Despite these
limitations, the results show evidence of validity for a
new attention/executive functions test, which may be
more adequate for the assessment of Brazilian adults
with psychiatric disorders than the traditional ones due
to their heterogeneous educational background.
we found correlations between the Five Points Test
unique drawing (a more “pure” measure of uency and
processing speed) with the two rst FDT components
(22 and 18% of shared variance respectively), but not
with the Choosing and Shifting components. However,
the Five Points Test errors were signicantly correlated
with all the FDT errors, but more with the Choosing
Errors (56% of shared variance). On the divergent
validity analysis, no signicant correlations were found
between any measures of the FDT with the RAVLT sum
Table 3 shows the BD and control groups
performance in the FDT, as well as the group
comparisons. We found large effect sizes for all FDT
timed measures (Cohens’s d ranging from 2.8 to 3.3)
and for errors of Choosing (d=1.1) and Shifting (d=1.0).
Comparisons on “automatic” attentional process errors
were not signicant. When all the timed measures along
with the Choosing and Shifting Errors were entered on
a forward stepwise binary logistic regression model,
the procedure yielded signicant results on its last step
(χ²=62.81, df=1, p<0.001), and answered for 79% of
variance according to the Negelkerke pseudo-R² test.
The FDT Shifting time alone (β= -0.32, SE=0.09,
p<0.001, Wald=12.43) had an overall classication
rate of 90%, 86% for the BD group and 94% for the
control group. The results suggest a sensitivity of 0.94,
specicity of 0.87, predictive positive value of 0.86 and
predictive negative value of 0.94 for the FDT Shifting
time in the distinction between controls and BD patients.
Our results provide a series of evidences about
the construct validity of the FDT for the assessment
of patients with BD. When correlated with a criterion
measure of selective-attention/executive functions
(SCWT) we found a large fraction of shared variance
between the tests, both in terms of time and errors.
When correlated to the convergent measure, a non-
verbal uency test (Five Points Test), we found
signicant associations between the automatic timed
components of the test and the production of drawings;
and the errors on the FDT with incorrect drawings
Table 3. Group comparisons on the FDT measures between the Bipolar Disorder group and a paired control group
(matched by gender, age and education)
FDT Measures Bipolar Disorder (n=35) Controls (n=35) Independent samples t-test
M (SD) Min-Max M (SD) Min-Max t d
Reading (T) 29.83 (8.50) 16-55 19.94 (2.59) 14-24 6.6* 3.3
Counting (T) 37.91 (12.27) 19-73 23.43 (2.94) 19-29 6.8* 3.1
Choosing (T) 57.94 (21.17) 31-120 35.06 (5.09) 26-46 6.2* 2.8
Shifting (T) 79.71 (29.98) 41-153 44.17 (5.53) 35-55 6.9* 2.9
Reading (E) 0.29 (0.96) 0-4 0.00 (0.00) 0-0 1.8 -
Counting (E) 0.49 (1.54) 0-8 0.00 (0.00) 0-0 1.9 -
Choosing (E) 1.97 (3.71) 0-17 0.29 (0.57) 0-2 2.7* 1.1
Shifting (E) 3.09 (5.35) 0-28 0.97 (1.12) 0-3 2.3* 1.0
T: Time, E: Errors, M: Mean, SD: Standard-Deviation, Min: Minimum, Max: Maximun, , d: Cohen’s d (effect size)
The Five Digits Test in Bipolar Patients
Clinical Neuropsychiatry (2014) 11, 3 107
assessment of type I bipolar patients. Journal of Affective
Disorders 112, 1, 231-236.
O’Shea R, Poz R, Michael A, Berrios GE, Evans JJ,
Rubinsztein JS (2010). Ecologically valid cognitive tests
and everyday functioning in euthymic bipolar disorder
patients. Journal of Affective Disorders 125, 1, 336-340.
Raven J (1981). Manual for Raven’s Progressive Matrices and
Vocabulary Scales. Harcourt Assessment, San Antonio,
Regard M, Strauss E, Knapp P (1982). Children’s production
on verbal and non-verbal uency tasks. Perceptual and
Motor Skills 55, 1, 839-44.
Malloy-Diniz LF, Lasmar VA, Gazinelli LS, Fuentes D,
Salgado JV (2007). he Rey Auditory-Verbal Learning
Test: applicability for the Brazilian elderly population.
Revista Brasileira de Psiquiatria 29, 4, 324-329.
Sedó MA (2007). Five Digits Test. TEA Ediciones, Madrid.
Sedó MA, DeCristoforo L (2001). All-language verbal
tests free from linguistic barriers. Revista Espanola de
Neurosicología 3, 3, 68-82.
Solé B, Bonnin CM, Balanzá-Martínez V, Tabarés-Seisdedos
R, Popovic D, Martínez-Arán A, Vieta E (2012).
Neurocognitive impairment and psychosocial functioning
in bipolar II disorder. Acta Psychiatriatrica Scandinavica
125, 4, 309-317.
Troyer AK, Leach L, Strauss E (2006). Aging and response
inhibition: Normative data for the Victoria Stroop Test.
Aging, Neuropsychology, and Cognition: A Journal on
Normal and Dysfunctional Development 13, 1, 20-35.
Young RC, Biggs JT, Ziegler VE, Meyer DA (1978). A rating
scale for mania: reliability, validity and sensitivity. The
British Journal of Psychiatry 133, 429-435.
Grant INCT-MM (FAPEMIG: CBB-APQ-00075-09/
Amorim P (2000). Mini International Neuropsychiatric
Interview (MINI): validation of a short structured
diagnostic psychiatric interview. Revista Brasileira de
Psiquiatria 22, 3, 106-115.
Beck AT, Ward CH, Mendelson M, Mock J, Erbaugh J (1961).
An inventory for measuring depression. Archives of
General Psychiatry 4, 561-571.
Cox CS, Chee E, Chase GA, Baumgardner L, Schuerholz LJ,
Reader MJ, Morh J, Denckla MB (1997). Reading prociency
affects the construct validity of the Stroop Test Interference
Score. The Clinical Neuropsychologist 11, 2, 105-110.
de Paula JJ, Ávila RT, Costa DS, Moraes EN, Bicalho MA,
Nicolato R, Corrêa H, Sedó M, Malloy-Diniz LF (2011).
Assessing processing speed and executive functions
in low educated older adults: the use of the Five Digits
Test in patients with Alzheimer’s disease, mild cognitive
impairment and major depressive disorders. Clinical
Neuropsychiatry: Journal of Treatment Evaluation 8, 6,
Magalhães SS (2013). Estrutura fatorial do controle inibitório
no envelhecimento: Comparação entre amostras de
adultos e idosos [dissertation]. Curitiba: Universidade
Federal do Paraná.
Malloy-Diniz LF, Neves FS, Abrantes SS, Fuentes D, Corrêa
H (2009). Suicide behavior and neuropsychological