Measurement of upper limb movement acceleration and functions in children with cerebral palsy

Abstract

OBJECTIVE:

The main purpose of this study was to measure the peak acceleration of the upper limb (UL) during reaching, and to calculate correlations between peak acceleration data and functional test results in children with cerebral palsy (CP).

METHODS:

We recruited 15 children with CP (8 boys and 7 girls) and measured peak acceleration and function as revealed by the Jebsen Taylor Hand Function Test (JTHF), the Quality of Upper Extremity Skills Test (QUEST), the Box and Blocks Test (BBT), and the ABILHAND-Kids questionnaire. We calculated correlations between peak acceleration data and scores on the functional tests.

RESULTS:

The peak acceleration of the more-affected UL was significantly higher than that of the less-affected UL ( $p<$ 0.05). The peak acceleration data were positively correlated with JTHFT scores. On the other hand, the peak acceleration data were negatively correlated with QUEST, BBT, and ABILHAND-Kids scores. The test-retest reliability of the peak acceleration was excellent, with an intra-class correlation coefficient (ICC) 0.87–0.98.

CONCLUSIONS:

Peak acceleration data correlated with UL functional test results; as this proved to be reliable, the tri-axial accelerometer is a clinically useful assessment tool for evaluating UL movement. Therefore, our results suggest that measurement of acceleration using a tri-axial accelerometer is appropriate when clinicians quantify UL movement during therapeutic rehabilitation in clinical settings.

Keywords

Correlation peak acceleration reaching reliability tri-axial accelerometer

1. Introduction

Reaching is defined as voluntary and purposeful movement of the upper limb (UL) to make contact with objects [1]. To achieve reaching, there must be harmony between the musculoskeletal (e.g., muscles and tendons) and neurological (e.g., brain and spinal cord) systems [2]. In addition, reaching is essential for completion of daily living activities such as eating and playing with toys (children) [3]. However, the reaching movement of children with cerebral palsy (CP) is only partial, due to motor deficits such as muscle weakness, atypical muscle tone, altered movement coordination, and inappropriate postural adjustment [4]. Thus, it is especially important to evaluate reaching when determining functional level or UL status in clinical practice.

A recent study of arm reaching in children with CP demonstrated that the affected-arm moved a shorter distance and more slowly than the less-affected arm [5]. Another study examined differences in shoulder and hand displacement between the involved and less-involved arm during hitting task with a rod to round target by mild to moderate hemiplegic CP patients. This study found that hitting task by the involved arm was of longer duration, had a lower maximal velocity of movement, and showed a loss of smoothness, compared with the less-involved arm [6].

More recently, several investigators have tried to find a valid and reliable way to evaluate UL function and/or performance. Nevertheless, most research on assessment of UL movement in children with CP has used three-dimensional (3D) motion analysis systems employing body-mounted sensors [7, 8]. However, processing and analyzing data collected via 3D motion analysis systems is expensive, complex, and time-consuming. On the other hand, collection of kinematic data via accelerometers is a low-cost and reliable technique when used to evaluate UL movement in both stroke patients [9] and children with CP [7]. Aboelnasr et al. [7] investigated difference of reaching movement between children with spastic hemiparetic CP and typically developing children using 3D motion analysis systems. They found slower and less smooth movement in children with hemiparetic CP than typically developing children.

However, there is a lack of evidence on UL movement acceleration during reaching in children with CP. Furthermore, no study has yet examined the relationship between UL movement acceleration and UL functions in this population. Therefore, the first purpose of our study was to compare peak acceleration between the more- and less-affected ULs during reaching in children with CP. The second aim of our work was to investigate the relationship between UL peak acceleration and UL functions measured using the JTHFT, QUEST, BBT, and ABILHAND-Kids evaluation tools. The main research questions were whether a tri-axial accelerometer could be used to measure peak acceleration of the UL in children with CP, and whether correlations were or were not evident between the UL peak acceleration and functional test scores.

2. Methods

2.1 Subjects

Fifteen children with CP (8 boys and 7 girls; mean age, 9.0 $\pm$ 1.1 years; mean height, 130.5 $\pm$ 6.3 cm; mean weight, 29.3 $\pm$ 9.5 kg) were recruited from Gyeongju city. Prior to recruiting the subjects, we estimated the target sample size by performing power analysis using G-Power 3.1 version software (University of Dusseldorf, Dusseldorf, Germany). Based on an effect size of 3.73, a 1- $\beta$ error (power) of 0.08, and an $\alpha$ error of 0.05, the target sample was estimated to be eight children with CP. Thus, we recruited 15 children with CP to allow for dropout. The inclusion criteria were (1) a diagnosis of CP by a neurologist, (2) an ability to understand the instructions of the researchers, and (3) a Gross Motor Function Classification System (GMFCS) grade of above level III, (4) able to extend and reach UL independently. The exclusion criteria were (1) a UL Modified Ashworth Scale (MAS) spasticity score of above grade 2, (2) orthopedic surgery on the dominant UL within the prior 6 months, and (3) a House Functional Classification System (HFCS) grade of below 4 [10]. The anthropometric and clinical characteristics of all subjects are shown in Table 1. All experimental protocols of this study were approved by the Institutional Review Board for Human Studies of Inje University (approval no. 2017-01-012). Prior to participating in this study, we received written informed consent from the parents of all children.

Table 1
Clinical and anthropometric characteristics of the subjects

Children	Age	Gender	Height (cm)	Weight (kg)	Dignosis	HFCS ${}^{\text{a}}$	Sensory ${}^{\text{b}}$
1	10	B	128.2	23.8	Diplegia	4	Intact
2	9	B	132.7	33.7	Hemiplegia	5	Intact
3	7	B	133.6	35.7	Hemiplegia	5	Intact
4	7	G	125.0	28.0	Diplegia	5	Intact
5	9	B	135.3	30.0	Hemiplegia	6	Intact
6	9	G	131.6	35.2	Diplegia	5	Intact
7	7	G	119.3	35.4	Hemiplegia	7	Intact
8	10	B	127.5	25.2	Hemiplegia	6	Intact
9	10	B	145.9	45.9	Hemiplegia	6	Intact
10	9	G	131.5	31.3	Diplegia	6	Intact
11	10	G	128.7	25.3	Diplegia	5	Intact
12	9	G	125.4	23.3	Hemiplegia	6	Intact
13	10	G	125.8	29.5	Hemiplegia	6	Intact
14	9	B	129.6	25.6	Diplegia	4	Intact
15	10	B	137.8	36.6	Hemiplegia	6	Intact

HFCS ${}^{\text{a}}$ : House Functional Classification System; Sensory ${}^{\text{b}}$ : Touch, pressure, and position sense.

2.2 Apparatus

We used a tri-axial accelerometer (Fitmeter, Fit.Life Inco., Suwon, Korea) in this study. The device was 35 mm in length $\times$ 35 mm in width $\times$ 13 mm in height, and weighed approximately 13.7 g. It was firmly taped to the third metacarpal bone. Each child was seated on a stool facing a table. He or she was required to flex the elbow at approximately 90 ${}^{\circ}$ with slight pronation of the forearm. In the reaching task, all children extended the elbow to touch a small red ball (diameter: 20 cm, position: shoulder height) at a self-selected pace, and then returned to the starting position. All measurements were performed three times and 10 s of rest was allowed between each trial. Data were collected at 128 Hz. The movement acceleration curve was used to evaluate peak acceleration in the medio-lateral (ML) and vertical (VT) directions. These were identified and calculated using Fitmeter Manager 2 software (Fit.Life Inco., Suwon, Korea).

2.3 Clinical measurements

Initially, the parents of all children filled out personal information forms and then a principal researcher collected clinical information including the HFCS score and data on touch, pressure, and position sensing from all children. Following collection of clinical information, all children sat beside the therapeutic table without any back support, to facilitate data collection, as follows. (1) The Jebsen Taylor Hand Function Test (JTHFT) was used to assess UL and hand function. This test comprises seven subtests including writing, card-turning, holding common small objects, simulated feeding, playing checkers, picking up large light objects, and picking up large heavy objects. The shorter the task execution time means the higher the quality of UL function. The JTHF exhibits moderate to excellent reliability in healthy subjects [intra-rater reliability: intra-class correlation coefficient (ICC) (1,1) $=$ 0.516–0.814; inter-rater reliability: ICC (3,1) $=$ 0.584–0.892] [12]. A researcher measured the time taken to finish each subtest (using a stopwatch) and allowed 10 s of rest between each task. The test was run three times and the mean value used for data analysis. (2) The Quality of Upper Extremity Skills Test (QUEST) was also performed to assess the quality of UL functions. The QUEST evaluates dissociated movements (64 items), grasping ability (24 items), weight-bearing (50 items), and protective extension (36 items). The total QUEST score ranges from 0 to 100. Scores close to 100 indicate that UL function is of good quality. The QUEST exhibits moderate to strong reliability in children with CP [intra-rater reliability: ICC (1,1) $=$ 0.84–0.90; inter-rater reliability: ICC (3,1) $=$ 0.84–0.90] [13]. Initially, a researcher trained in pediatric evaluation measured the raw scores and then transformed these using the prescribed formula. All assessment procedures were carried out in accordance with the QUEST manual [14]; The Box and Blocks Test (BBT) was applied to study UL and hand functions. The tools for testing consist of 150 colored wooden blocks (2.5 cm $\times$ 2.5 cm $\times$ 2.5 cm) and two boxes (25.4 cm ${}^{2}$ ). A recent BBT reliability study showed that the inter- and intra-rater reliabilities were ICC $=$ 0.99 and 0.96, respectively, in neurologic patients [15]. Prior to the test, all children were allowed a practice trial for familiarization. When the test began, each child sat on an adjustable chair facing the tools on a table. The children then moved the blocks from one box to the other for 60 s. A researcher counted the number of transferred blocks. The test was performed three times and the average value was used for data analysis. (4) The ABILHAND-Kids questionnaire was used to evaluate manual skill. The test includes 21 sub-items related to daily life. The ABILHAND-Kids has been shown to be of high reliability in children with CP (R $=$ 0.94) [16]. All parents completed the questionnaire (using three-point ordinal scales) and a researcher then evaluated the data using the appropriate Rasch formula.

2.4 Statistical analysis

All data are given as mean values with standard deviations (SDs). The peak acceleration of the more-affected and less-affected ULs were compared using the Mann-Whitney U-test. Pearson’s correlation was used to seek relationships between UL peak acceleration and scores on the functional assessment tools (JTHFT, QUEST, BBT, and ABILHAND-Kids). The ICC was calculated to estimate test-retest reliability. Effect sizes of 0.20–0.39 are regarded as “weak”; 0.40–0.59 as “moderate”; 0.60–0.79, as “strong”; and $>$ 0.80 as “very strong”. The statistical level of significance adopted was a two-tailed p-value $<$ 0.05. All data were analyzed using PASW statistics software (ver. 20; Norusis/SPSS Inc., Chicago, IL, USA).

3. Results

The ML peak acceleration of the more- and less-affected ULs were 0.80 $\pm$ 0.13 m/s ${}^{2}$ and 0.60 $\pm$ 0.13 m/s ${}^{2}$ , respectively (U $=$ 30.00, $p=$ 0.001). The VT peak acceleration of the more- and less-affected ULs were 0.69 $\pm$ 0.14 m/s ${}^{2}$ and 0.53 $\pm$ 0.12 m/s ${}^{2}$ , respectively (U $=$ 46.00, $p=$ 0.006) (Table 2). The detailed data from the JTHFT, QUEST, BBT, and ABILHAND-Kids evaluations are shown in Table 3. The ML and VT peak acceleration data correlated positively with the JTHFT scores. On the other hand, the ML and VT peak acceleration data correlated negatively with QUEST, BBT, and ABILHAND-Kids scores. The ML acceleration data correlated significantly with all variables ( $p<$ 0.05), and the VL acceleration data correlated significantly with JTHFT, BBT, and ABILHAND-Kids scores ( $p<$ 0.05). However, no significant correlation was found between VT acceleration data and QUEST scores ( $p>$ 0.05). (Table 4). The test-retest reliability of the peak acceleration was excellent (ICC $=$ 0.87–0.98).

Table 2
Comparison of peak acceleration

ML ${}^{\text{a}}$ peak acceleration				VT ${}^{\text{b}}$ peak acceleration
More-affected side	Less-affected side	U	$p$	More-affected side	Less-affected side	U	$p$
0.80 $\pm$ 0.13	0.60 $\pm$ 0.13	30.00	0.001 ${}^{*}$	0.69 $\pm$ 0.14	0.53 $\pm$ 0.12	46.00	0.006 ${}^{*}$

ML ${}^{\text{a}}$ : medio-lateral; VT ${}^{\text{b}}$ : vertical.

Table 3

Outcomes of upper limb functional tests

Functional tests	Outcomes
JTHFT ${}^{\text{a}}$ (s)	191.31	$\pm$ 23.05
QUEST ${}^{\text{b}}$ (%)	67.41	$\pm$ 6.76
BBT ${}^{\text{c}}$ (EA)	46.13	$\pm$ 8.53
ABILHAND-Kids (logit)	1.51	$\pm$ 0.37

JTHFT ${}^{\text{a}}$ : Jebsen Taylor Hand Function Test; QUEST ${}^{\text{b}}$ : Quality of Upper Extremity Skills Test; BBT ${}^{\text{c}}$ : Box and Blocks Test.

Table 4

Correlation between upper limb peak acceleration and functional tests

	ML ${}^{\text{a}}$ acceleration		VT ${}^{\text{b}}$ acceleration
	$r$	$p$	$r$	$p$
JTHFT ${}^{\text{c}}$	0.756 ${}^{*}$	0.001	0.601 ${}^{*}$	0.018
QUEST ${}^{\text{d}}$	$-$ 0.792 ${}^{*}$	0.000	$-$ 0.482	0.069
BBT ${}^{\text{e}}$	$-$ 0.692 ${}^{*}$	0.004	$-$ 0.657 ${}^{*}$	0.008
ABILHAND-Kids	$-$ 0.679 ${}^{*}$	0.005	$-$ 0.672 ${}^{*}$	0.006

ML ${}^{\text{a}}$ : medio-lateral; VT ${}^{\text{b}}$ : vertical; JTHFT ${}^{\text{c}}$ : Jebsen Taylor Hand Function Test; QUEST ${}^{\text{d}}$ : Quality of Upper Extremity Skills Test; BBT ${}^{\text{e}}$ : Box and Blocks Test.

4. Discussion

In this study, we measured UL peak acceleration in the ML and VT directions during reaching in children with CP. Moreover, we calculated the relationships between UL peak acceleration and functions as measured by JTHFT, QUEST, BBT, and ABILHAND-Kids tools. We found that the ML and VT peak acceleration in the more-affected UL were lower than the ML and VT peak acceleration. Additionally, the UL peak acceleration in the ML and VT directions were significantly correlated with JTHF, QUEST, BBT, and ABILHAND-Kids scores.

Most importantly, the ML and VT peak acceleration of the more-affected UL were 30% and 25% higher than the ML and VT peak acceleration of the less-affected UL, respectively ( $p<$ 0.05). Our outcomes are consistent with those of previous studies dealing with reaching in populations with neurologic deficits [3, 5, 17]. Michaelsen et al. compared peak acceleration between paretic and non-paretic ULs in chronic stroke patients. In this study, they found that the peak acceleration of the paretic and non-paretic ULs were 0.45 and 0.71 m/s ${}^{2}$ , respectively. Thus, the figure for the paretic UL was 36% lower than that for the non-paretic UL [17]. A recent study compared cumulative reaching distance, reaching smoothness, and reaching velocity of the non-preferred and preferred UL in hemiplegic CP patients [5]. It was found that cumulative reaching distance, reaching smoothness, and reaching velocity of the non-preferred UL differed significantly compared with the preferred UL (being reduced by 10%, 55%, and 17%, respectively) [5].

Significantly positive (JTHFT) and negative (QUEST, BBT, and ABILHAND-Kids) correlations were observed between peak acceleration data (both ML and VT directions) and UL functional variables. Additionally, moderate to good correlations with functional test results were apparent: JTHFT $=$ 0.756; QUEST $=-$ 0.792; BBT $=-$ 0.692; ABILHAND-Kids $=-$ 0.679. The JTHFT measures the time taken to complete tasks and the QUEST, BBT, and ABILHAND-Kids findings represent the quality of performance. Thus, a longer JTHFT time means a poorer UL function, whereas higher scores on the QUEST, BBT, and ABILHAND-Kids evaluations indicate higher UL quality. Our research findings support those of previous studies, as follows. Thrane et al. [18] investigated correlations between raw acceleration data and Fugl-Meyer Assessment scores ( $r=-$ 0.851, $p<$ 0.001). Additionally, Wang et al. [19] found a moderate correlation between acceleration data and ABILHAND-Kids scores ( $r=$ 0.54, $p<$ 0.01).

The test-retest reliability of UL peak acceleration data derived using a tri-axial accelerometer was associated with an ICC $=$ 0.87–0.98. This value was somewhat lower than that of a previous study exploring the reliability of acceleration measurements during reaching in children with CP ( $r=$ 0.92) [20]. We think that the reason for this difference is that we recruited children of younger age than those of the previous study.

Taken together, our findings demonstrate that peak acceleration data derived using a tri-axial accelerometer were reliable and valid, and correlated with the results of JTHF, BBT, QUEST, and ABILHAND-Kids tests in children with CP. Clinically, we recommend that clinicians should assess peak acceleration using a tri-axial accelerometer when evaluating UL functions and coordination. This would aid in the planning of treatment. In particular, as the tri-axial accelerometer is small, light, and portable, it can be used to evaluate dynamic movements in a real-world setting. A laboratory is not required.

This study had several limitations that should be reflected in future research. First, our sample size was limited. As children with CP have different clinical characteristics, we had difficulty fulfilling our inclusion criteria. Therefore, further studies should be conducted on larger samples. Second, we recruited children aged 6–11 years as in the previous studies. Additional research is needed for children ages 4–5 to generalize our findings. Finally, we did not consider trunk control. In the future, it will be necessary to research the relationship between trunk control and UL using a tri-axial accelerometer.

5. Conclusions

The ML and VT peak acceleration data of the UL were positively correlated with JTHFT data and negatively correlated with QUEST, BBT, and ABILHAND-Kids data. We also found that evaluation using a tri-axial accelerometer was reliable. Therefore, our results suggest that measurement of peak acceleration using a tri-axial accelerometer would be helpful when clinicians seek to measure the effects of therapeutic rehabilitation on the UL in clinical settings.

Footnotes

Conflict of interest

None to report.

References

Archambault

Ferrari-Toniolo

Caminiti

Battaglia-Mayer

. Visually-guided correction of hand reaching movements: The neurophysiological bases in the cerebral cortex. Vision Res. 2015; 110(Pt B): 244-56.

Jayaneththi

Viloria

Wiedemann

Jarrett

McDaid

. Robotic assessment of neuromuscular characteristics using musculoskeletal models: A pilot study. Comput Biol Med. 2017; 86: 82-9.

Chang

. Kinematical measure for spastic reaching in children with cerebral palsy. Clin Biomech (Bristol, Avon). 2005; 20(4): 381-8.

Rosenbaum

Paneth

Leviton

Goldstein

Bax

Damiano

, et al. A report: the definition and classification of cerebral palsy April 2006. Dev Med Child Neurol Suppl. 2007; 109: 8-14.

Ronnqvist

Rosblad

. Kinematic analysis of unimanual reaching and grasping movements in children with hemiplegic cerebral palsy. Clin Biomech (Bristol, Avon). 2007; 22(2): 165-75.

Van Thiel

Steenbergen

. Shoulder and hand displacements during hitting, reaching, and grasping movements in hemiparetic cerebral palsy. Motor Control. 2001; 5(2): 166-82.

Aboelnasr

Hegazy

Altalway

. Kinematic characteristics of reaching in children with hemiplegic cerebral palsy: A comparative study. Brain Inj. 2017; 31(1): 83-9.

Lenfeldt

Johansson

Domellof

Riklund

Ronnqvist

. Alterations in white matter microstructure are associated with goal-directed upper-limb movement segmentation in children born extremely preterm. Hum Brain Mapp. 2017 [Epub ahead of print].

Subramanian

Yamanaka

Chilingaryan

Levin

. Validity of movement pattern kinematics as measures of arm motor impairment poststroke. Stroke. 2010; 41(10): 2303-8.

10.

Lazzari

Politti

Belina

Collange Grecco

Santos

Dumont

AJL

, et al. Effect of transcranial direct current stimulation combined with virtual reality training on balance in children with cerebral palsy: A randomized, controlled, double-blind, clinical trial. J Mot Behav. 2017; 49(3): 329-36.

11.

Mackey

Walt

Stott

. Deficits in upper-limb task performance in children with hemiplegic cerebral palsy as defined by 3-dimensional kinematics. Arch Phys Med Rehabil. 2006; 87(2): 207-15.

12.

Culicchia

Nobilia

Asturi

Santilli

Paoloni

De Santis

, et al. Cross-cultural adaptation and validation of the Jebsen-Taylor hand function test in an Italian Population. Rehabil Res Pract. 2016; 2016: 897-917.

13.

Haga

van der Heijden-Maessen

van Hoorn

Boonstra

Hadders-Algra

. Test-retest and inter- and intrareliability of the quality of the upper-extremity skills test in preschool-age children with cerebral palsy. Arch Phys Med Rehabil. 2007; 88(12): 1686-9.

14.

DeMatteo

. QUEST: quality of upper extremity skills test: DeMatteo 1992.

15.

Platz

Pinkowski

van Wijck

Kim

di Bella

Johnson

. Reliability and validity of arm function assessment with standardized guidelines for the Fugl-Meyer Test, Action Research Arm Test and Box and Block Test: a multicentre study. Clin Rehabil. 2005; 19(4): 404-11.

16.

Arnould

Penta

Renders

Thonnard

. ABILHAND-Kids: a measure of manual ability in children with cerebral palsy. Neurology. 2004; 63(6): 1045-52.

17.

Michaelsen

Gomes

Marques

Rodrigues

Borges

, Jr Claudino

, et al. Using an accelerometer for analyzing a reach-to-grasp movement after stroke. Motriz Revista de Educacao Fisica. 2013; 19(4): 746-52.

18.

Thrane

Emaus

Askim

Anke

. Arm use in patients with subacute stroke monitored by accelerometry: association with motor impairment and influence on self-dependence. J Rehabil Med. 2011; 43(4): 299-304.

19.

Wang

Lin

Chung

Pei

Teng

. Validity, responsiveness, and clinically important difference of the ABILHAND questionnaire in patients with stroke. Arch Phys Med Rehabil. 2011; 92(7): 1086-91.

20.

Yoo

Lee

Cha

You

. Augmented effects of EMG biofeedback interfaced with virtual reality on neuromuscular control and movement coordination during reaching in children with cerebral palsy. Neuro Rehabilitation. 2017; 40(2): 175-85.