Research in Developmental Disabilities 59 (2016) 35–42

Contents lists available at ScienceDirect

Research in Developmental Disabilities

Aiding the diagnosis of AD/HD in childhood: Using actigraphy

and a continuous performance test to objectively quantify
symptoms
Hannah Gilbert a , Ling Qin b , Dandan Li b , Xuehua Zhang b , Stuart J. Johnstone a,∗
a
School of Psychology, University of Wollongong, Wollongong, NSW 2519, Australia
b
Centre for Cognition and Sleeping, People’s Hospital of Guangxi Zhuang Autonomous Region, 6 Taoyuan Rd, Nanning, PR China

a r t i c l e i n f o a b s t r a c t

Article history: The current gold standard for the diagnosis of AD/HD is based on subjective reports from
Received 14 April 2016 teachers, parents, and clinicians. These measures can be problematic as they are open to
Received in revised form 14 July 2016 rater biases and also fail to account for the developmental nature of symptoms. The current
Accepted 26 July 2016
study examined the diagnostic accuracy of two objective measures, a computer-based Con-
Available online 3 August 2016
tinuous Performance Task and actigraphy (e.g. motion tracking device) in differentiating
Number of reviews completed is 2
children with AD/HD (N = 70) from healthy controls (N = 70). It was predicted that task-
unrelated movement (measured via actigraphy) during a CPT and CPT performance would
Keywords: have high classification accuracy in differentiating children with AD/HD from healthy con-
AD/HD
trols, and that the inclusion of age would increase this accuracy. Results indicated that total
Diagnosis
energy expenditure from the task-unrelated wrist and ankle movement during the CPT
Objective
CPT was higher in children with AD/HD than controls, and that CPT performance was poorer
Actigraphy in AD/HD than controls. Discriminant function analyses revealed that the CPT Full-Scale
Response Control Quotient and wrist and ankle energy expenditure provided optimal classi-
fication accuracy – correctly classifying 86% of cases, with sensitivity of 81.4% and specificity
of 91.4%. The prediction that classification accuracy would increase with the inclusion of age
was not supported by the data. When taken in conjunction with other clinical assessments,
these findings suggest that actigraphy during a CPT and CPT performance may increase the
probability of a correct AD/HD diagnosis.
© 2016 Elsevier Ltd. All rights reserved.

1. Introduction

Attention-Deficit/Hyperactivity Disorder (AD/HD) is the most common neurobehavioural disorder in school-age children,
with an estimated worldwide prevalence of 5.9–7.1% (Willcutt, 2012). The current gold standard for the diagnosis of AD/HD
in childhood is based on the subjective reports of clinicians, parents, and teachers. These reports are open to biases from
raters, and thus many have questioned their validity in the diagnostic process (Hartman, Rhee, Willcutt, & Pennington,
2007; Molina, Pelham, Blumenthal, & Galiszewski, 1998; Schultz & Evans, 2012). Accordingly, the development of objective
measures to assist diagnosis of AD/HD is an important research endeavor.

∗ Corresponding author.
E-mail addresses: hcg971@uowmail.edu.au (H. Gilbert), qlhlaa@163.com (L. Qin), vera.egg@hotmail.com (D. Li), zxh871217@163.com (X. Zhang),
sjohnsto@uow.edu.au (S.J. Johnstone).

http://dx.doi.org/10.1016/j.ridd.2016.07.013
0891-4222/© 2016 Elsevier Ltd. All rights reserved.
36 H. Gilbert et al. / Research in Developmental Disabilities 59 (2016) 35–42

Diagnostic interviews, such as the National Institute of Mental Health Diagnostic Interview Schedule for Children IV
(DISC-IV), are the preferred method for the diagnosis of AD/HD as they are in direct correspondence with the Diagnostic
and Statistical Manual of Mental Disorders (DSM) criteria (McGrath, Handwerk, Armstrong, Lucas, & Friman, 2004). The
DISC-IV is a structured psychiatric diagnostic interview that can be administered by a lay interviewer, or self-administered
via a computerized version, to children and adolescents aged 6–18 years, or their parents (McGrath et al., 2004; Wright,
Waschbush, & Frankland, 2007). However, the DISC-IV is not without limitations for it fails to address the developmental
nature of AD/HD symptoms (Curchack-Lichtin, Chacko, & Halperin, 2014). The DISC is based on current DSM-V criteria,
which provides a list of symptoms that are to be applied to all age groups. This criterion is most problematic when applied to
children in the preschool years, for symptoms of hyperactivity are predominant during this period (Lahey, Pelham, Loney, Lee,
& Wilcutt, 2005). For example, the symptom “Always on the go”, was endorsed by up to 72% of parents of preschool children,
with or without a diagnosis of AD/HD (Egger, Kondo, & Angold, 2006). Inattentive symptoms also appear to be problematic
when applied to young children. For example, “Fails to give close attention to details or makes careless mistakes” is hard
to apply to young children who are not usually engaged in structured tasks or activities (Curchack-Lichtin et al., 2014). In
order to overcome this issue, the DSM-V states that symptoms must be inconsistent with the child’s developmental level.
However, there are no clear guidelines to what represents normal behaviour at a particular age. This ambiguity may be
of particular concern for parents without access to developmental norms, and parents of young children (Byrne, Bawden,
Beattie, & DeWolfe, 2000).
Behaviour rating scales are also commonly used to assess AD/HD behavioural symptoms consistent with DSM-V criteria
(e.g. Conners’ Parent Rating Scale (CPRS), Conners’ Teacher Rating Scale (CTRS), Child Behaviour Checklist (CBCL), and NICHQ
Vanderbilt Assessment Scales). These are typically completed by multiple informants, such as parents and teachers, to
assess the child’s behaviour in multiple settings (Schultz & Evans, 2012). It is recommended that practitioners combine
these symptom reports in order to determine an accurate AD/HD diagnosis. The accuracy of such measures may be hindered
by the perceptions or biases of these informants (Pelham, Fabiano, & Massetti, 2005; Sibley et al., 2012). The validity of
teacher rating appears to be especially problematic during the high school years, as students often interact with several
teachers on a daily basis. Research has found only low to moderate inter-teacher reliability on AD/HD rating scales with
correlations ranging from 0.21 to 0.52 (Molina et al., 1998; Schultz & Evans, 2012). Teacher demographics such as age and
sex also influence the perceived severity or leniency of behavioural symptoms. For example, Schultz and Evans (2012)
reported that female and younger teachers provided more severe ratings of hyperactivity-impulsivity, than did male and
older teachers. Similar disparities have been documented between teacher and parent ratings of behaviour, with only low
to moderate correlations ranging from 0.30 to 0.50 (Barkley, 2003). Maternal biases resulting from depression and/or other
psychopathology may also influence the accuracy of these diagnostic measures. For example, several studies have reported
that mothers with elevated depressive symptoms are more likely to exaggerate their child’s behaviour than non-depressed
parents (Chi & Hindshaw, 2002; Chilcoat & Breslau, 1997; Najman et al., 2000). This finding has been extended to ratings of
AD/HD, with maternal depression predicting elevated ratings of AD/HD symptomology in children (Chi & Hindshaw, 2002).
Together these findings highlight the shortcomings of current diagnostic methods, which become most problematic when
clinicians are expected to provide an accurate diagnosis based on these reports (Schultz & Evans, 2012). As such, researchers
have considered the utility of objective methods in the diagnosis of AD/HD. For example, consideration of resting brain
electrical activity, as measured by the electroencephalogram (EEG), has been investigated as a possible diagnostic marker
of AD/HD (Helgadottir et al., 2015; Lansbergen, Arns, van Dongen-Boomsma, Spronk, & Buitelaar, 2011). This is based on
findings showing differences in the resting brain activity of children with AD/HD compared to healthy controls (e.g. Barry,
Clarke, & Johnstone, 2003; Clarke, Barry, McCarthy, & Selikowitz, 2001). EEG activity has been shown to have moderate
to high classification accuracy in distinguishing children with AD/HD from controls (Helgadottir et al., 2015), and while it
appears to be a promising tool in diagnostic practice, it can be costly to administer and requires specialised equipment and
personnel (). Another method that has been considered as an objective measure of AD/HD is a Continuous Performance Test
(CPT). A typical CPT involves the rapid presentation of stimuli on a computer screen, with participants required to respond
to a target stimulus. Several variables can be derived including reaction times and errors of commission and omission, at
various rates of stimulus delivery (Preston, Fennell, & Bussing, 2005). These variables are said to measure the core symptoms
of AD/HD, including levels of inattention, impulsivity, and vigilance (Epstein et al., 2003). Although the CPT has been widely
studied in AD/HD, most studies have only reported moderate classification accuracy with high false positive rates of up to 30%
(Forbes, 1998; Schatz, Ballantyne, & Trauner, 2001). This has led researchers to suggest that the CPT be only supplementary
in the diagnostic process (Ogundele, Ayyash, & Banerjee, 2011). AD/HD versus control group differences in performance
are likely attributable to decreased perceptual sensitivity and slower drift rates (Huang-Pollock, Karalunas, Tam, & Moore,
2012).
In order to increase the diagnostic efficiency of the CPT, wireless accelerometers known as actigraphs can provide a mea-
sure of non-task-related movement. As there is commonly an increased level of physical activity (referred to as hyperactivity)
in children with AD/HD, actigraphs have been used to quantify this movement (Reh, Schmidt, Rief, & Christiansen, 2014)
with specific measures identified as suitable candidates for molecular genetic studies seeking to identify polymorphisms
associated with the risk for AD/HD (Wood, Asherson, Rijsdijk, & Kuntsi, 2009). Recent AD/HD studies using movement indices
(i.e. traditional actigraphy measures, or more novel movement measures) have shown high discriminative validity during
structured tasks (O’Mahony et al., 2014; De Crescenzo et al., 2016), during a CPT combined with CPT performance (Brocki,
 H. Gilbert et al. / Research in Developmental Disabilities 59 (2016) 35–42 37

Tillman, & Bohlin, 2010), or over a 24 h period (Martín-Martínez et al., 2012). However, this research fails to take into account
the developmental nature of AD/HD symptoms.
Building on previous research, this study examines the effect of age on the diagnostic efficiency of both actigraphy
during a CPT, and CPT performance. This aim is in support of the developmental hypothesis of AD/HD (Berger, Slobodin,
Aboud, Melamed, & Cassuto, 2013), which posits that neuropsychological and behavioural deficits are dependent on a
child’s developmental level, with a reduction in symptoms over time (Barkley, 2003; Brocki et al., 2010). The study combines
movement data from two actigraph devices (e.g. one attached to the child’s non-dominant wrist, and one attached to the
non-dominant ankle), with performance measures from an Integrated Visual and Auditory Continuous Performance Test
(IVA-CPT). Seventy children diagnosed with either AD/HD Combined type or AD/HD Hyperactive/Impulsive type, and seventy
healthy control children participated in this study. Discriminant function analyses were performed to examine the ability
of age, movement indices via actigraphy, and IVA-CPT performance to predict group membership. It was hypothesised that
children with AD/HD would have higher levels of activity during the IVA-CPT task and poorer performance on the IVA-CPT
than healthy controls. It was further hypothesised that classification accuracy would be best when CPT and actigraphy were
considered together, and that the inclusion of age would result in improved classification accuracy.

2. Material and methods

2.1. Participants

Participants were 70 healthy control children and 70 children diagnosed with either AD/HD Combined type or AD/HD
Hyperactive/Impulsive type. The AD/HD group consisted of 64 males and 6 females, aged between 7 years 6 months and
11 years 2 months (M = 9.30, SD = 0.11). The control group consisted of 64 males and 6 females, aged between 7 years and 6
months and 10 years and 9 months (M = 9.25, SD = 0.11). Children in either group were excluded if they: (a) had an estimated
IQ of less than 80 on the Raven’s Standard Reasoning Test, (b) had any disorders of consciousness or head injuries, or (c)
had any comorbid mental disorders. All children in the AD/HD group were diagnosed with AD/HD of the Combined Type
or Hyperactive/Impulsive Type by an experienced clinician according to the Diagnostic and Statistical Manual of Mental
Disorders IV (DSM-IV; APA, 2000) diagnostic criteria, and other comprehensive assessments such as CPRS and Vanderbilt
AD/HD Parent Rating Scale. Children in the AD/HD group were recruited through the Centre for Cognition and Sleep, People’s
Hospital of Guangxi Zhuang Autonomous Region (China), after referral to the Centre for assessment of behavioural issues.
Children in the control group were recruited through a nearby local primary school. Children were asked to abstain from
taking any stimulant medication for two weeks prior to testing, to ensure that there were no significant residual medication
effects on the performance measures. Prior to the start of testing, informed consent was provided by a parent or caregiver,
and informed assent was provided by the child. The Research Ethics Committee of the People’s Hospital of Guangxi Zhuang
Autonomous Region approved the research protocol.

2.2. Procedure

Participants completed the Integrated Visual and Auditory Test (IVA-CPT) in an assessment room of the Centre for Cog-
nition and Sleep. The ambient noise of the assessment room was lower than 45 dB, and the room contained basic furniture
and minimal distractions. A one-to-one test mode was adopted, with only the participant and research assistant present in
the room. Two ActiGraphTM GT3X (ActiGraph, LLC, Pensacola, Florida) motion sensor devices were fitted to the participant
- one to the wrist of the child’s non-dominant hand, and the other to the ankle of the child’s non-dominant leg. Dominance
was determined by asking the child questions such as, “Which foot do you use when kicking a football?” and “Which hand
do you use when holding a mouse?” There were no reports of discomfort from children regarding the fit and placement of
the actigraph devices. Requirements of the IVA-CPT were then explained, and a practice block completed – if there was any
indication of the participant not understanding the requirements, the task was paused and the rules were further explained.
During completion of the IVA-CPT children were seated on a non-swivel chair to reduce excess movement. In order to
observe the child, the examiner remained seated in the room during testing without disturbance. The testing procedure
lasted for approximately 20 min. Children in the control group completed the same testing procedure in a meeting room
of the primary school, located within 800 m of the hospital. The room setup was similar to the laboratory environment,
with ambient noise lower than 45 dB. A one-to-one test mode was adopted, with only the participant and research assistant
present in the room.
Movement counts from the wrist and ankle actigraphs were converted into kilocalories (Kcal), which are units of energy
expenditure. Kcal units were calculated using a formula that integrates counts per minute and body mass in Kg. Kcal units
have been reported in the literature as measures of physical activity (e.g. light, moderate, and vigorous activity levels) in
both children with AD/HD and non-affected controls (Baerg et al., 2011; Brazendale et al., 2015). Larger Kcal units represent
greater levels of energy expenditure and physical activity.
38 H. Gilbert et al. / Research in Developmental Disabilities 59 (2016) 35–42

Table 1
Descriptive statistics for each Group and age group. Standard deviation is shown in brackets.

AD/HD Healthy control

Low age group High age group Low age group High age group

Age (years) 8.5 (0.38) 10.1 (0.55) 8.5 (0.43) 10.2 (0.48)
Kcal wrist 1.35 (0.59) 1.52 (0.74) 0.74 (0.55) 0.61 (0.47)
Kcal ankle 0.28 (0.14) 0.32 (0.27) 0.12 (0.10) 0.13 (0.11)
Full-Scale Response Control Quotient 43.9 (38.3) 51.0 (36.1) 75.6 (26.1) 86.8 (18.0)
Full-Scale Attention Quotient 46.4 (41.5) 55.2 (38.3) 72.8 (29.1) 89.9 (13.2)

2.3. Data analysis

All variables were screened for possible outliers and non-normal distributions. Normality test statistics were computed
for each variable. For one of the behavioural variables (Kcal Ankle), skewness was outside the desired range of −1 to +1.
Therefore a Log 10 transformation was conducted on this positively skewed distribution. Variables that met the assumption
of normality were analysed using a series of Univariate Analyses of Variance (ANOVAs) with Group (control, AD/HD) as the
independent factor. Levene’s statistic was significant for several of the behavioural variables (Kcal Wrist, Full-Scale Response
Control Quotient, and Full-Scale Attention Quotient), therefore violating the assumption of homogeneity of variance. In
order to account for this violation, the Welch statistic is reported for these variables. To reduce the chance of Type 1 errors, a
Bonferroni adjusted alpha level of 0.004 (i.e. 0.05/12 = 0.004) was implemented. In line with the primary aim of this research,
discriminant function analysis (DFA) was used to determine which continuous variables discriminated between the AD/HD
and control group. Firstly, a DFA was conducted using only IVA-CPT derived variables (specifically, the Full-Scale Response
Control Quotient and Full-Scale Attention Quotient) with the grouping variable (control, AD/HD). The actigraph energy
expenditure measures (Kcal Wrist, Kcal Ankle) were then added into the DFA along with the IVA-CPT variables, to determine
whether the inclusion of actigraph data would increase classification accuracy. Lastly, Age Group was entered in to the DFA to
determine whether group classification would be further increased. The Age Group variable was derived by a median split on
each group (control, AD/HD) according to age in years. Those above the median (control Mdn = 9.25, AD/HD Mdn = 9.20) were
allocated to the high Age Group, and those below the median were allocated to the low Age Group. A further exploratory DFA
was then conducted using the stepwise method, in order to determine which variables contributed most to group separation.
The predictor variables entered in this analysis were Age Group, Full-Scale Response Control Quotient, Full-Scale Attention
Quotient, Kcal Wrist, and Kcal Ankle with the grouping variable (control, AD/HD). For each DFA, classification coefficients
were calculated and tested as well as ‘leave-one-out’ cross-validation.

3. Results

Table 1 provides descriptive statistics of AD/HD and healthy groups separately for each age group.

3.1. Actigraph data – group comparison

Total energy expenditure (Kcal) from the wrist actigraph was significantly higher, Welch’s F(1, 125.87) = 61.84, p < 0.001,
␻2 = 0.55, for the AD/HD group (M = 1.44 Kcal units, SD = 0.67, 95% CI [1.30, 1.61]) than for the control group (M = 0.68 Kcal
units, SD = 0.48, 95% CI [.56, 0.80]), indicating higher levels of movement in the non-dominant “free” wrist during completion
of the IVA-CPT for the AD/HD group. Total energy expenditure (Kcal) from the ankle actigraph was also significantly higher,
F(1, 138) = 56.45, p < 0.001, ␻2 = 0.53, for the AD/HD group (M = 0.31, SD = 0.22, 95% CI [-0.70, -0.54]) than the control group
(M = 0.12, SD = 0.10, 95% CI [−1.14, -0.97]). Therefore, children in the AD/HD group had significantly higher levels of ankle
movement during the IVA-CPT than children in the control group.

3.2. IVA-CPT data – group comparison

The Full-Scale Response Control Quotient score from the IVA-CPT provides a measure of an individual’s overall ability
to regulate responses and respond appropriately. A low score on the Full-Scale Response Control Quotient is indicative
of lower levels of prudence (e.g. avoids impulsive errors of commission), consistency (e.g. response times are clustered
within a narrow range), and stamina (e.g. response times are maintained for the duration of the test). Here, the Full-Scale
Response Control Quotient score was significantly lower, Welch’s F(1, 121.50) = 45.09, p < 0.001, ␻2 = 0.49, for the AD/HD
group (M = 47.50, SD = 36.71, 95% CI [38.75, 56.25]) than the control group (M = 83.11, SD = 24.93, 95% CI [77.17, 89.06]). The
mean score for the AD/HD group fell within the Extremely Impaired range, while the mean score for the control group fell
within the Average range.
The Full-Scale Attention Quotient score of the IVA-CPT provides a measure of an individual’s overall ability to make
accurate responses, stay focused and sustain attention. A low score on the Full-Scale Attention Quotient is indicative of
lower levels of vigilance (e.g. avoids inattentive errors of omission), focus (e.g. consistency in processing speed) and speed
 H. Gilbert et al. / Research in Developmental Disabilities 59 (2016) 35–42 39

(e.g. rapid response times). Here, the IVA-CPT Full-Scale Attention Quotient score was also significantly lower, Welch’s F(1,
121.58) = 33.44, p < 0.001, ␻2 = 0.43, for the AD/HD group (M = 50.96, SD = 39.70, 95% CI [41.49, 60.42]) than the control group
(M = 84.14, SD = 27.01, 95% CI [77.70, 90.58]). The mean score for the AD/HD group fell within the Extremely Impaired range,
while the mean score for the control group fell within the Average range.

3.3. Discriminant function analyses

3.3.1. DFA using IVA-CPT measures

The two IVA-CPT quotient scores were entered together in a DFA. Group centroids (group means on the discriminant func-
tion) were significantly different between the groups (control = 0.568, AD/HD = -0.568, Wilk’s  = 0.75, 2 = 38.82, p < 0.001).
As illustrated in Fig. 1a, these variables correctly classified 70% of cases. The sensitivity (i.e. proportion of AD/HD group that
were correctly identified as such) of the classification function was 58.6%, with a specificity (i.e. proportion of controls that
were correctly identified as such) of 81.4%.

3.3.2. DFA using IVA-CPT and actigraph measures

The two actigraph measures (Kcal Ankle, and Kcal Wrist) were entered together into a DFA with the two IVA-CPT quo-
tient scores. Group centroids were significantly different between the groups (control = − 0.988, AD/HD = 0.988, Wilk’s
 = 0.50, 2 = 93.60, p < 0.001). As illustrated in Fig. 1b, these variables correctly classified 84% of cases. The sensitivity of the
classification function was 82.9%, with a specificity of 91.4%

3.3.3. DFA using IVA-CPT, actigraph, and age group measures

The variable Age Group was entered into the discriminant analysis along with the IVA-CPT scores, and the actigraph mea-
sures. Group centroids were significantly different between the groups (control = −0.992, AD/HD = 0.992, Wilk’s  = 0.50,
2 = 93.86, p < 0.001). As illustrated in Fig. 1c, these variables correctly classified 82% of cases. The sensitivity of the classifi-
cation function was 80%, with a specificity of 90%.

3.3.4. Stepwise DFA

The stepwise DFA simultaneously entered the five dependent variables (Full-Scale Response Control Quotient, Full-Scale
Attention Quotient, Kcal Wrist, Kcal Ankle, and Age Group) into the analysis. Of these variables, only three significantly con-
tributed to group separation – these were Full-Scale Response Control Quotient, Kcal Wrist, and Kcal Ankle, Wilk’s  = 0.50,
2 = 93.93, p < 0.001. Group centroids were significantly different between the groups (control = -0.988, AD/HD = 0.988).
The canonical correlation between the groups and the discriminant function formed by the three variables was moderate,
Rc = 0.71. Overall, the IVA-CPT Full-Scale Response Control Quotient score contributed most to group separation (r 2 = 0.66),
followed by the actigraph measures Kcal Wrist (r 2 = 0.53) and then Kcal Ankle (r 2 = 0.41). The stepwise discriminant function
correctly classified 86% of cases, with a sensitivity of 81.4%, and a specificity of 91.4%.

4. Discussion

Based on previous studies using actigraphy (e.g. Brocki et al., 2010; O’Mahony et al., 2014; Reh et al., 2014) it was
hypothesised that children with AD/HD would have higher levels of activity during a CPT task than typically developing
children. As hypothesised, movement indices from the wrist and ankle actigraphs indicated that children with AD/HD had
significantly higher levels of movement during the IVA-CPT than controls. This finding is consistent with research showing
increased levels of activity in children with AD/HD (e.g. Brocki et al., 2010; Reh et al., 2014). A CPT is said to measure the
core symptoms of AD/HD including levels of inattention and impulsivity (Epstein et al., 2003). This finding is supported
by studies showing poorer CPT performance in children with AD/HD, compared to typically developing children (Epstein
et al., 2003; O’Mahony et al., 2014; Vaughn et al., 2011). Furthermore, it was hypothesised that children with AD/HD would
have poorer performance on the IVA-CPT than controls and this was supported. The IVA-CPT quotient scores indicated that
children with AD/HD had significantly lower levels of response control (i.e. an overall ability to regulate responses and
respond appropriately) and attention (i.e. an overall ability to make accurate responses, stay focused, and sustain attention),
than children in the control group. More specifically, the Full-Scale Response Control Quotient Scale scores indicated that
children in the AD/HD group had lower levels of prudence (e.g. avoids impulsive errors of commission), consistency (e.g.
response times are clustered within a narrow range), and stamina (e.g. response times are maintained for the duration of the
test). The Full Scale Attention Quotient scores indicated that children in the AD/HD group had lower levels of vigilance (e.g.
avoids inattentive errors of omission), focus (e.g. consistency in processing speed), and speed (e.g. rapid response times).
Previous research using CPT performance has reported only moderate classification accuracy in distinguishing children
with AD/HD from controls (Forbes, 1998; Schatz et al., 2001). Therefore, it was hypothesised that IVA-CPT performance would
have only moderate classification accuracy. Results from the discriminant function analysis supported this hypothesis, with
the two IVA-CPT scores correctly classifying 70% of cases. This finding supports an argument made by several researchers
that the CPT should only be supplementary in the diagnostic process (Ogundele et al., 2011).
It was further hypothesised that the inclusion of actigraphy (e.g. movement indices), along with IVA-CPT performance
would further increase the classification accuracy of the discriminant function, in line with O’Mahony et al. (2014). The
40 H. Gilbert et al. / Research in Developmental Disabilities 59 (2016) 35–42

Fig. 1. Classification accuracy of the discriminant function analyses. Optimal classification accuracy was achieved with the combination of IVA-CPT and
actigraphy measures.

results supported this hypothesis, with the two IVA-CPT scores and the two actigraph measures correctly classifying 84.3%
of cases. A plausible explanation for this increase in classification accuracy may be the ability of actigraphy to effectively
capture the hyperactivity component of AD/HD, which is often pronounced in the Combined type and Hyperactive/Impulsive
presentations of AD/HD.
 H. Gilbert et al. / Research in Developmental Disabilities 59 (2016) 35–42 41

In accordance with the developmental hypothesis of AD/HD, it was predicted that the inclusion of age into the discriminant
function would further increase the classification accuracy reported above. Results from the discriminant function analysis
indicated that age had a negative impact on classification accuracy, with a slight reduction to 82.1% of cases correctly
classified, and did not contribute significantly to group separation in the stepwise DFA. This result is perhaps surprising
considering the documented change found in AD/HD symptoms with chronological age (e.g. Brocki et al., 2010), but it
should be noted that this study examined participants in the child age-range, rather than spanning the child and adolescent
age-range. This finding does lend support to research showing a general decline in AD/HD symptomology, which does not
occur until later in adolescence (e.g. 11–18 years of age; Biederman, Mick, & Faraone, 2000).
A limitation of the present study is a lack of clinical comparison, with children in the control group being excluded if
they presented with any clinical symptoms of hyperactivity or impulsivity. In clinical practice, a clinician may be presented
with an array of childhood psychological disorders and their associated symptoms. Therefore, it is necessary for future
research to examine the diagnostic utility of actigraphy during a CPT, and CPT performance in differentiating children with
AD/HD from children with other psychological disorders. Related to this point, comparing a clinical population with healthy
controls in diagnostic test evaluation may lead to an overestimation of the sensitivity and specificity of the test, referred to
as spectrum bias (Elie, Coste, & French Society of Clinical Cytology Study Group, 2008). It must be noted that test accuracy
could differ between subgroups of patients with different characteristics, including presence and severity of symptoms. A
further limitation of the present study is the limited age range, with children aged between 7 and 11 years of age. Based on
previous research showing symptom decline later in adolescence, future research should examine the influence of age on
the diagnostic utility of actigraphy during a CPT, and CPT performance using a sample of adolescent children (e.g. between
11 and 18 years of age). Lastly the sample used in this study is taken from a Chinese population. While research suggests that
AD/HD is a cross-cultural disorder, with similar symptom structure between eastern and western cultures (Norvilitis and
Fang, 2005; Yang, Schaller, & Parker, 2000), further replication studies using cross-cultural samples are needed to confirm
the results of the present study.

5. Conclusion

The results of the present study provide further evidence that quantified movement (via actigraphy) during a CPT, and
CPT performance can have high classification accuracy in distinguishing children with AD/HD from controls. Furthermore,
the use of these objective measures in clinical practice may have considerable diagnostic utility in children aged 7–11
years. Not only do these objective measures rule out sources of rater bias found in other subjective diagnostic measures of
AD/HD, they are also time efficient and inexpensive to implement. Our results indicate that actigraphy and a CPT may assist
clinicians in evaluating a child’s behaviour, and may increase the probability of a correct diagnosis when, importantly, used
in conjunction with other clinical data. While this study suggests the implementation of objective measures to aid the clinician
in the diagnostic process, it does by no means suggest that these measures replace careful DSM-V criteria, as assessed by a
skilled clinician.

Funding sources

The research was funded by the Guangxi S&T Program (1355005-5-5) and the Guangxi Talent Highland Development
and Promotion Program of Appropriate Health Techniques (006008) grants to LQ. These funding sources did not play any
role in study design; in the collection, analysis and interpretation of data; in the writing of the report; or in the decision to
submit the article for publication.

Conflict of interest

The authors declare that there is no conflict of interest.

References

Baerg, S., Cairney, J., Hay, J., Rempel, L., Mahlberg, N., & Faught, B. E. (2011). Evaluating physical activity using acceleromerty in children at risk of
developmental coordination disorder in the presence of attention deficit hyperactivity disorder. Research in Developmental Disabilities, 32(4),
1343–1350.
Barkley, R. A. (2003). Issues in the diagnosis of attention-deficit/hyperactivity disorder in children. Brain and Development, 25(2), 77–83.
Barry, R. J., Clarke, A. R., & Johnstone, S. J. (2003). A review of electrophysiology in attention-deficit/hyperactivity disorder: I. Qualitative and quantitative
electroencephalography. Clinical Neurophysiology, 114, 171–183.
Berger, I., Slobodin, O., Aboud, M., Melamed, J., & Cassuto, H. (2013). Maturational delay in ADHD: Evidence from CPT. Frontiers in Human Neuroscience, 7,
1–11.
Biederman, J., Mick, E., & Faraone, S. V. (2000). Age-dependent decline of symptoms of attention deficit hyperactivity disorder: Impact of remission
definition and symptom type. American Journal of Psychiatry, 157(5), 816–818.
Brazendale, K., Chandler, J. L., Beets, M. W., Weaver, R. G., Beighle, A., Huberty, J. L., & Moore, J. B. (2015). Maximising children’s physical activity using the
LET US Play principles. Preventative Medicine, 76, 14–19.
Brocki, K. C., Tillman, C. M., & Bohlin, G. (2010). CPT performance, motor activity, and continuous relations to ADHD symptom domains: A developmental
study. European Journal of Developmental Psychology, 7(2), 178–197.
42 H. Gilbert et al. / Research in Developmental Disabilities 59 (2016) 35–42

Byrne, J. M., Bawden, H. N., Beattie, T. L., & DeWolfe, N. A. (2000). Preschoolers classified as having attention-deficit hyperactivity disorder (ADHD): DSM-
IV symptom endorsement pattern. Journal of Child Neurology, 15(8), 533–538.
Chi, T. C., & Hindshaw, S. P. (2002). Mother-child relationships of children with ADHD: The role of maternal depressive symptoms and depression-related
distortions. Journal of Abnormal Child Psychology, 30(4), 387–400.
Chilcoat, H. D., & Breslau, N. (1997). Does psychiatric history bias mother’s reports? An application of a new analytic approach. Journal of the American
Academy of Child and Adolescent Psychiatry, 36(7), 971–979.
Clarke, A. R., Barry, R. J., McCarthy, R., & Selikowitz, M. (2001). Excess beta activity in children with attention-deficit/hyperactivity disorder: An atypical
electrophysiological group. Psychiatry Research, 103(2–3), 205–218.
Curchack-Lichtin, J. T., Chacko, A., & Halperin, J. M. (2014). Changes in ADHD symptom endorsement: Preschool to school age. Journal of Abnormal Child
Psychology, 42, 993–1004.
De Crescenzo, F., Licchelli, S., Ciabattini, M., Menghini, D., Armando, M., Alfieri, P., . . . & Vicari, S. (2016). The use of actigraphy in the monitoring of sleep
and activity in ADHD: A meta-analysis. Sleep Medicine Reviews, 26, 9–20.
Egger, H. L., Kondo, D., & Angold, A. (2006). The epidemiology and diagnostic issues in preschool attention-deficit/hyperactivity disorder: A review. Infants
and Young Children, 19, 109–122.
Elie, C., Coste, J., & French Society of Clinical Cytology Study Group. (2008). A methodological framework to distinguish spectrum effects from spectrum
biases and to assess diagnostic and screening test accuracy for patient populations: Application to the Papanicolaou cervical cancer smear test. BMC
Medical Research Methodology, 8, 7.
Epstein, J. N., Erkanli, A., Conners, C. K., Klaric, J., Costello, J. E., & Angold, A. (2003). Relations between continuous performance test performance measures
and ADHD behaviours. Journal of Abnormal Child Psychology, 31(5), 543–554.
Forbes, G. B. (1998). Clinical utility of the test of variables of attention (TOVA) in the diagnosis of attention-deficit/hyperactivity disorder. Journal of Clinical
Psychology, 54(4), 461–476.
Hartman, C. A., Rhee, S. H., Wilcutt, E. G., & Pennington, B. F. (2007). Modeling rater disagreement for ADHD: Are parents or teachers biased? Journal of
Abnormal Child Psychology, 35(4), 536–542.
Helgadottir, H., Gudmundsson, O. O., Baldursson, G., Magnusson, P., Blin, N., Brynjolfsdottir, B., . . . & Johnsen, K. (2015). Electroencephalography as a
clinical tool for diagnosing and monitoring attention deficit hyperactivity disorder: A cross-sectional study. BMJ Open, 5(1), 1–9.
Huang-Pollock, C. L., Karalunas, S. L., Tam, H., & Moore, A. N. (2012). Evaluating vigilance deficits in ADHD: A meta-analysis of CPT performance. Journal of
Abnormal Psychology, 121(2), 360–371.
Lahey, B. B., Pelham, W. E., Loney, J., Lee, S. S., & Wilcutt, E. (2005). Instability of the DSM-IV subtypes of ADHD from preschool through to elementary
school. Archives of General Psychiatry, 62(8), 896–902.
Lansbergen, M. M., Arns, M., van Dongen-Boomsma, M., Spronk, D., & Buitelaar, J. K. (2011). The increase in theta/beta ration on resting-state EEG in boys
with attention-deficit/hyperactivity disorder is mediated by slow alpha peak frequency. Progress in Neuropsychopharmacology & Biological Psychiatry,
35(1), 47–52.
Martín-Martínez, D., Casaseca-de-la Higuera, P., Alberola-López, S., Andrés-de Llano, J., López-Villalobos, J. A., Ardura-Fernández, J., & Alberola-López, C.
(2012). Nonlinear analysis of actigraphic signals for the assessment of the attention-deficit/hyperactivity disorde (ADHD). Medical Engineering and
Physics, 34, 1317–1329.
McGrath, A. M., Handwerk, M. L., Armstrong, K. J., Lucas, C. P., & Friman, P. C. (2004). The validity of the ADHD section of the diagnostic interview for
children. Behaviour Modification, 28(3), 349–374.
Molina, B., Pelham, W. E., Blumenthal, J., & Galiszewaski, E. (1998). Agreement among teachers’ behaviour ratings of adolescents with a childhood history
of attention deficit hyperactivity disorder. Journal of Clinical Child Psychology, 27(3), 330–339.
Najman, J. M., Williams, G. M., Nikles, J., Spence, S., Bor, W., O’Callaghan, M., & Andersen, M. J. (2000). Mother’s mental illness and child behaviour
problems: Cause-effect association or observation bias? Journal of the American Academy of Child and Adolescent Psychiatry, 39(5), 592–602.
Norvilitis, J. M., & Fang, P. (2005). Perception of ADHD in China and the United States: A preliminary study. Journal of Attention Disorders, 9(2), 413–424.
O’Mahony, N., Florentino-Liano, B., Carballo, J. J., Baca-Garcia, E., & Rodriguez, A. A. (2014). Objective diagnosis of ADHD using IMUs. Medical Engineering
and Physics, 36(7), 922–926.
Ogundele, M. O., Ayyash, H. F., & Banerjee, S. (2011). Role of computerised continuous performance task tests in ADHD. Progress in Neurology and
Psychiatry, 15(3), 8–13.
Pelham, W. E., Fabiano, G. A., & Massetti, G. M. (2005). Evidence-based assessment of attention deficit hyperactivity disorder in children and adolescents.
Journal of Clinical Child and Adolescent Psychiatry, 34(3), 449–476.
Preston, A. S., Fennell, E. B., & Bussing, R. (2005). Utility of a CPT in diagnosing ADHD among a representative sample of high risk children: A cautionary
study. Child Neuropsychology, 11(5), 459–469.
Reh, V., Schmidt, M., Rief, W., & Christiansen, H. (2014). Preliminary evidence for altered motion tracking-based hyperactivity in ADHD siblings.
Behavioural and Brain Functions, 10(7), 1–8.
Schatz, A. M., Ballantyne, A. O., & Trauner, D. A. (2001). Sensitivity and specificity of a computerized test of attention in the diagnosis of attention-
deficit/hyperactivity disorder. Assessment, 8(4), 357–365.
Schultz, B. K., & Evans, S. W. (2012). Sources of bias in teacher ratings of adolescents with ADHD. Journal of Educational and Developmental Psychology, 2(1),
151–162.
Sibley, M. H., Pelham, W. E., Jr., Molina, B. S., Gnagy, E. M., Waschbusch, D. A., Garefino, A. C., . . . & Karch, K. M. (2012). Diagnosing ADHD in adolescence.
Journal of Consulting and Clinical Psychology, 80(1), 139–150.
Vaughn, A. J., Epstein, J. N., Rausch, J., Altaye, M., Langberg, J., Newcorn, J. H., . . . & Wigal, T. (2011). Relation between outcomes on a continuous
performance test and ADHD symptoms over time. Journal of Abnormal Child Psychology, 39(6), 853–864.
Willcutt, E. G. (2012). The prevalence of DSM-IV attention-deficit/hyperactivity disorder: A meta-analytic review. Neurotherapeutics, 9, 490–499.
Wood, A. C., Asherson, P., Rijsdijk, F., & Kuntsi, J. (2009). Is overactivity a core feature in ADHD? Familial and receiver operating characteristic curve
analysis of mechanically assessed activity level. Journal of the American Academy of Child & Adolescent Psychiatry, 48(10), 1023–1030.
Wright, K. D., Waschbusch, D. A., & Frankland, B. W. (2007). Combining data from parent ratings and parent interview when assessing ADHD. Journal of
Psychopathology and Behavioural Assessment, 29(3), 141–148.
Yang, K. N., Schaller, J. L., & Parker, R. (2000). Factor structures of Taiwanese teacher’s ratings of ADHD: A comparison with U.S. studies. Journal of Learning
Disabilities, 33(1), 72–82.