Master’s Thesis, 30 ECTS
The Program for Master of Science in Psychology Oriented Towards Sports, 300 ECTS
Spring 2020
Supervisor: Professor Louise Rönnqvist
MOVEMENT INITIATION AND EXECUTION IN 6 – 8 YEAR OLD CHILDREN BORN
PRETERM: EFFECTS OF GESTATIONAL AGE AND
PHYSICAL ACTIVITY
Klara Stjernholm & Lisa Wennergren Gros
BW: Birth weight
CBCL: Child Behaviour Checklist
DCD: Developmental Coordination Disorder EF: Executive functions
EPT: Extremely preterm
FSIQ: Full Scale Intelligence Quotient FT: Full term
GA: Gestational age MPT: Moderately preterm
MVPA: Moderate to Vigorous Physical Activity PA: Physical Activity
PT: Preterm
SI: Processing speed index VPT: Very preterm
WISC-IV: Wechsler Intelligence scale for Children, Fourth edition
Abstract
The purpose of this study was to explore differences in movement initiation and execution, and their associations with amount of physical activity (PA) and cognitive abilities, in 6 to 8 year old children born PT compared to peers born at term. The sample consisted of in total 78 children divided in subgroups, 16 born very preterm (VPT), 24 born moderate preterm (MPT) and 38 age matched controls born at term (FT) with no diagnosed cognitive or motor impairments. Three-dimensional (3D) kinematic recordings of wrist movements during two bimanual tapping tasks (horizontal and vertical) were finalized and kinematic outcome measures were investigated in relation to PA and performance on WISC-IV. Children born VPT showed significantly longer Latency times and longer Duration of movement execution compared to children born MPT and FT. No significant within group correlations between PA and movement performance were found. Duration of movement execution and Total duration of movement execution were negatively associated with Full-scale intelligence quotient (FSIQ) and processing speed index (SI) in the PT group. Early school aged children born VPT need longer planning time to initiate and execute goal directed bimanual movements, compared to peers born MPT and FT. Kinematic performance did not evidently associate with amount of PA, although kinematics, by means of longer Latency time and Duration of movement execution, negatively associated with SI.
Keywords: Prematurity, movement planning, kinematics, physical activity, cognitive abilities
Abstrakt
Syftet med denna studie var att undersöka skillnader i initiering och utförande av en rörelse samt deras associationer med fysisk aktivitet och kognitiv förmåga hos förtidigt födda barn i åldrarna 6 till 8 år samt åldersmatchade fullgångna jämnåriga barn. Urvalet bestod av total 78 barn uppdelade i subgrupper, 16 väldigt förtidigt födda, 24 moderat förtidigt födda och 38 åldersmatchade fullgångna kontroller utan några kända kognitiva eller motoriska nedsättningar.
Tredimensionella (3D) kinematiska registreringar av handledsrörelser under två bimanuella knapptrycksuppgifter (horisontell och vertikal) genomfördes. Utfallet av rörelsemätningarna studerades i association till fysisk aktivitet och resultat på WISC-IV. Väldigt förtidigt födda barn visade längre latenstid och längre duration av rörelse jämfört med moderat förtidigt födda och fullgångna barn. Inga signifikanta inom grupps korrelationer mellan fysisk aktivitet och rörelseutförande hittades. Utförandets duration samt hela utförandets duration associerade negativt med IQ kvot och processhastighet (SI) hos för tidigt födda barn. Väldigt förtidigt födda barn i tidig skolålder behöver mer tid till rörelseplanering, initiering och utförande av viljestyrda målinriktade rörelser jämfört med jämnåriga barn födda senare i graviditeten.
Kinematiskt utfall associerade inte signifikant med fysisk aktivitet, däremot associerade kinematiskt utfall, i form av längre latenstid och duration av utförande, negativt med kognitiv processhastighet.
Nyckelord: Prematuritet, rörelseplanering, kinematik, fysisk aktivitet, kognitivförmåga
Movement initiation and execution in 6 - 8 year old children born preterm: effects of gestational age and physical activity
The number of children born preterm (PT) has increased over the years and due to improvements in perinatal care, children are surviving at lower gestational age (GA) (Hebestreit & Bar-Or, 2001; Wolke, Johnson, & Mendonça, 2019). A birth before completion of 37 weeks of gestation is defined as a PT birth. A PT birth is divided into four subgroups defined by GA: late preterm (34-36 weeks of gestation); moderate preterm (MPT; 32-33 weeks of gestation); very preterm (VPT; < 32 weeks of gestation) and extremely preterm (EPT; < 28 weeks of gestation). The definition of a full term (FT) birth is ≥ 37 - 42 weeks of gestation (Wolke et al., 2019). A PT birth is an increasing risk factor for neurodevelopmental problems, cognitive difficulties, low academic performances as well as motor impairments (Allotey et al., 2017; Hebestreit & Bar-Or, 2001; Réveillon, Borradori Tolsa, Monnier, Hüppi, & Barisnikov, 2016). Motor impairments due to neurodevelopmental abnormalities are frequently described in children born PT, whereas up to 50% of children born < 30 weeks of gestation are reported with mild to severe motor impairments at school age (Spittle et al., 2016).
A study by Domellöf, Johansson, Farooqi, Domellöf and Rönnqvist (2013) showed how general cognitive abilities measured by WISC-IV, Full Scale IQ, relates to how effectively children born PT organize their movements in a unimanual precision task measured by kinematic registrations. Their results suggest that lower IQ score links to less effectively organized arm movements in children born VPT through longer durations and distances of movement compared to children born MPT and FT. A major part of cognitive abilities is executive functions (EF). Within the concept of EF there are planning and response inhibition.
To be able to organize your movements effectively, both pre- and online planningare crucial.
Less developed movement planning skills have been found in children born EPT at the age of 7 as well as in children born PT with DCD at the age of 8, compared to FT born peers. These problems have been found to relate to cognitive processing problems and affect school performance, learning and everyday life (Goyen, Lui, & Hummell, 2011; Lönnberg et al., 2018). Response inhibition is defined as an inhibition process of delaying, actively suppress, or interrupt a response (Réveillon et al., 2016). When examining EF in 6-year-old boys born FT with normal birth weight (BW) it was found that fetal maturity (i.e., GA) and fetal growth (i.e., BW in relation to GA) affected different types of EF, however neither GA nor BW were associated with inhibitory control (Phua et al., 2012). On the other hand, Réveillon et al. (2016) report results indicating that GA has an influence on response inhibition. According to Réveillon et al. (2016) children born PT (< 35 weeks of gestation) showed general difficulties in inhibiting a response compared to term born peers. Results were explained by an increase of impulsive behaviour and impairments in inhibition (Réveillon et al., 2016). In a recent meta- analysis higher rates of ADHD diagnoses are reported in children born PT compared to children born FT, where results indicate a correlation between lower GA and higher risk of ADHD symptoms (Allotey et al., 2017). Attention problems are frequently described in the PT population, and in the VPT population in particular. IQ and EF are found to be independent predictive factors for attention problems, where IQ is shown to have a stronger effect on inattention in children born VPT compared to term born peers (Aarnoudse-Moens, Weisglas- Kuperus, Duivenvoorden, van Goudoever, & Oosterlaan, 2013). Beyond, Aarnoudse-Moens et al. (2013) show that impaired EF explain attention problems even better.
Moreover, not only does GA seem to affect EF, so does also physical activity (PA) (Van der Niet et al., 2015). A positive relationship has been found between total volume of PA (consisting mainly of light intensity exercise) and planning performance, where both daily and intense PA might enhance planning skills in children aged 8 to 12 years. Van der Niet et al.
(2015) also suggested that more time spent in sedentary behaviour relates to worse performance regarding inhibitory control. Previous studies have also found that PA benefits both the structure of the brain as well as its functioning regarding cognitive skills in children, including EF and school performances (Khan & Hillman, 2014; Van der Niet et al., 2015). PA has also been established as a protective factor for health and wellbeing (Khan & Hillman, 2014). The association between PA and motor skills in children and adolescents has been investigated but the results in the field are ambiguous (Cliff, Okely, Smith, & McKeen, 2009). Cliff et al. (2009) found a positive association between 3 to 5 year old boy’s object control skills (catch, throw and kick) and volume of moderate to vigorous PA (MVPA). Matarma et al. (2018) found no relation between volume of MVPA and motor skills. According to Hebestreit and Bar-Or (2001) PA has a positive effect on coordination and supports the development of motor skills.
There is also research suggesting that organized physical training programs designed to improve motor skills can increase fine motor skills and agility (Alesi, Bianco, Luppina, Palma,
& Pepi, 2016; Qi, Tan, Sui, & Wang, 2018). Alesi et al. (2016) results further strengthened the evidence that PA might have positive effects on the development of EF.
Accordingly, PA may have a positive impact on motor skills. Due to the frequently present impairments in motor functioning seen in children born PT it has been suggested that exercise, in any form, is particularly important to encourage in this group (Hebestreit & Bar- Or, 2001). Hebestreit and Bar-Or (2001) also stress the importance of encouraging a variety of sports challenging and evolving multiple motor skills in the PT population. In a cohort study by Spittle, Cameron, Doyle, Cheong, & Victorian Infant Collaborative Study Group (2018) the frequency of non-cerebral palsy motor impairments in children born EPT (< 28 weeks of gestation) seem to increase over time. This increase in motor impairments is conferred as caused by a change in the children’s home and school environment, including an increase of sedentary behaviour. Thus, to increase the understanding of the environmental impact, there is a need for further research. Since motor impairments are often present in children born PT it is reasonable to think that children born PT would engage less in PA and sports compared to their term born peers. The research in this area show little or no difference in PA patterns in children born PT compared to children born at term (Kaseva et al., 2012; Lowe, Watkins, Kotecha, &
Kotecha, 2016; Spiegler, Mendonca, & Wolke, 2019). Kaseva et al. (2012) have suggested that BW is the main factor explaining differences in leisure time PA participation in adults, where very low BW predicting significantly lower participation. Spittle et al. (2018) raise the question of PA working as a protective factor for motor impairments in children born PT, which gives reason to study the impact of PA on subtle motor impairments in this population.
By use of kinematic registrations Domellöf, Johansson and Rönnqvist, (2018) found that children born PT display less developed movement organization at the age of 4 in terms of slower movements compared to FT controls. This is also supported by Domellöf et al. (2013). These differences did, however, not persist at the age of 8 (Domellöf et al., 2018).
Johansson, Domellöf and Rönnqvist (2014) studied the effects of timing training in youth with diplegic cerebral palsy, 12 to 16 years of age. With kinematic movement registrations, they could detect subtle motor improvements after completed intervention (Johansson et al., 2014).
Using the same kinematic registrations, the aim of this study was to investigate upper-limb movement initiation (by means of onset latency: time between the auditory “go” signal to the first hand motion), execution and inhibitory control (by number of false starts) during a goal directed, sequential tapping task in children born PT compared to FT born controls.
Additionally, we wanted to investigate the possible relationship between kinematic outcomes, cognitive abilities, GA and the children’s amount of regular PA. We aimed to study if PA could be used as a protective factor and a future intervention against subtle motor impairment in this vulnerable population. To further broaden the understanding of early motor impairments and/or
motor developmental delays in children born PT is crucial in order to limit disruption in motor skill development and its possible secondary impacts on PA participation and performance (Spittle et al., 2016). The use of kinematic measures in clinical settings is expected to improve early detection of impairments. It can also serve practitioners in both customizing and monitoring progression and effectiveness of interventions to this extra vulnerable population (Johansson et al., 2014; Spittle et al., 2016).
Well-developed EF play an important role concerning successful planning and organization of movements as well as the ability of inhibitory control (Domellöf et al., 2013;
Phua et al., 2012; Réveillon et al., 2016). It seems like organized sport activities may positively enhance the development of EF (Khan & Hillman, 2014; Van der Niet et al., 2015; Alesi et al., 2016) and that organized regular motor training improves motor skills (Cliff et al., 2009;
Hebestreit & Bar-Or, 2001; Alesi et al., 2016; Tan et al., 2018). We expected to see GA, amount of PA and cognitive abilities to correlate with inhibitory control and planning performance separately and positively, measured through movement initiation and execution. Furthermore, we hypothesized that latency time would be shorter and number of false starts would be lower the higher the GA (FT vs. PT, as well as VPT vs. MPT), as possible effects of planning, inhibition and/or attention problems.
Specific questions:
1. As parts of motor planning and movement initiation, how are movement inhibition and kinematic execution related?
2. Do latency time and number of false starts differ between the groups based on GA (FT vs. PT, as well as VPT vs. MPT), with latency time becoming shorter and number of false starts becoming lower the higher the GA?
3. Is there a relationship between kinematic performance and GA/BW? Does regular amount of PA relate to kinematic performance, with a higher amount of physical activity resulting in a stronger kinematic performance?
4. Is the correlation between kinematic performance and cognitive ability more noticeable than that between kinematic performance and physical activity?
Method Participants
Data from in total 78 children between the age of 6 to 9 years (mean age = 7.7, range:
6.2-8.7 years) was included in this study. Of them, 38 children were born FT (≥ 37 weeks of gestation) and 40 children were born PT. The children born PT were divided into two groups, consisting of 16 children born VPT (≤ 33 weeks of gestation) and 24 children born MPT (≤ 36 weeks of gestation), see Table 1. Of these, 78 of the participants completed CBCL and 65 completed WISC-IV. The participants in this study were all part of a quasi-longitudinal, comprehensive research project at Umea University (PI: L. Rönnqvist) that aims to investigate neurobehavioral and psychological development and functioning of children born PT compared to age matched children born FT. No child reported neuropsychological diagnoses or were labelled with any known motor deficiencies prior to testing, all children took part of regular schooling and had similar socioeconomic backgrounds.
Table 1
Participant characteristics (mean and range: min-max), gestational age (GA; measured in weeks), birth weight (BW; measured in grams) and age of testing (measured in years)
FT (N=38)
MPT (N=24)
VPT (N=16)
Measure Mean Min-Max Mean Min-Max Mean Min-Max
GA 40.3 38.0-41.9 34.4 33.6-35.4 28.3 22.9-32.4
BW 3762 2940-4790 2208 1367-2962 1078 404-1848
Age at testing 7.68 6.17-8.75 7.55 6.17-8.66 7.66 6.33-8.75
Note. FT = full term, MPT = moderately preterm, VPT = very preterm, GA = gestational age and BW = birth weight
Ethical considerations. Participation in this study was voluntary. Informed consent was obtained in writing from both parents and verbally from the children. The study from which our data was collected was approved by the ethical committee of Umea Regional Medical Ethical Board and conducted in accordance with the Declaration of Helsinki. Data was decoded to ensure anonymity.
Procedure
The participant’s movement performance was tested by a sequential, Extension, Flexion, Adduction, Abduction (EFAA) tapping task, in either a horizontal or vertical direction (for graphic illustration see Figure 1). The task was to perform goal directed upper limb movements, either by one hand at a time (unimanual condition) or with both hands simultaneously (bimanual condition). Markers were fixated with skin friendly adhesive tape on respectively left and right shoulder, elbow, wrist, knuckle of index fingers and forehead. This thesis focused on data from the bimanual condition where the participant had to coordinate their hand movements. Data was extracted from the markers on the wrists of both arms.
Throughout the task the participant was seated in a chair in front of a test platform with 10 integrated easy to press light-switches, see Figure 2. Beginning with the index fingers placed on specified starting points, the participant was instructed to start its hand movement on a sound signal and to press three light-switch buttons (with the right/left of both index-fingers) in a sequential order. The light-switches were to be pressed either vertically from the bottom to the top (extension) or horizontally from outside to midline (adduction). Kinematics were measured by optoelectronic registrations recorded by a 6-camera 3D optoelectronic registration system, at a pre-set recording resolution of 120 frames per second (ProReflex, Qualisys Inc., Gothenburg, Sweden). The kinematic measurement was synchronized with video recordings.
The participant cognitive function was measured through WISC-IV, all ten ordinary subtests were included and administered by a trained psychologist. The parents of the participants completed the Child Behaviour Checklist (CBCL/6-18). Background data regarding GA and BW was collected.
Figure 1. Graphic illustration of the EFAA table, all the measures given in centimetres.
Figure 2. Task setup and markers fixated on shoulders, elbows, wrists, knuckle of index fingers and forehead.
Measures. The kinematic parameters used in this study are described in Table 2. Child Behaviour Checklist (CBCL) is a standardized questionnaire constructed to identify behavioural problems and social competence among children and adolescents between the ages 6-18, assessment forms completed by a guardian (Bordin, et al., 2013). Background data from CBCL/6-18 was used to assess PA rate, where parents’ rate how much time their child spends in sports compared to peers and how skilled the child is in each sport compared to peers. This is estimated with a 3-level scale (1 = less than average, 2 = average and 3 = more than average).
A composite value of parents rating is used (PA rate) in this study. To further assess how much the child participated in sport a measure of total number of sports is included.
The Swedish version of Wechsler Intelligence Scale for Children-Fourth Edition (WISC-IV) was used to assess the children's cognitive functioning. It is a neuropsychological test which is used to assess cognitive functioning in children aged 6:0 to 16:11 (Pearson, 2003).
The assessment gives information about general intelligence through the full-scale. The scale
can also be divided into four indexes: perceptual and verbal functioning, processing speed and working memory. The measure of full-scale IQ (FSIQ) and processing speed index (SI) are used as they were the indexes correlating with kinematic parameters.
Table 2
Description of measures
Variable Description
Number of false starts (N) The number of false starts for each participant
Latency time (ms) The time it takes to react on “sound” go signal, and to plan a goal directed movement, to the exact moment (frame) whereas the movement begins
Velocity peak (mm/s) Highest velocity/speed of movement
Deceleration (%) Part of movement spent in deceleration phase in relation to acceleration
Duration of movement execution (ms)
Duration of movement to first button press (acceleration and deceleration) - latency time excluded
Total duration of
movement execution (ms)
Duration from go-signal to first button press - latency time included
FSIQ (WISC-IV total score)
Full-scale intelligence quotient
Processing speed index Measure of processing speed, attention, cognitive flexibility and perceptual speed
PA rate A composite value of how much time the child spends participating in sport and skillfulness compared to peers calculated from CBCL (parent rating)
Tot sport (N) Total number of different sport activities the child takes part in (on weekly basis)
Kinematic data analysis. All kinematic data were smoothed using a 13-frame fit to window filter (QTM-software). Data was analysed and extracted from the 3D (x, y, z) movement coordinates in QTM (adjusted Qualisys program). Latency time, duration from onset of movement until first velocity peak and end of execution were extracted by 120 Hz/sec and converted to milliseconds (ms). Peak velocity profile was extracted in mm/s, illustrated in Figure 3. Regarding latency time, a cut of at 100 ms for a false start was set. Justification for the 100 ms criteria was based on fastest possible auditory–movement onset reaction time for human beings with previous research showing that online movement adjustments can occur in as little as 110 ms (Holmes & Dakwar, 2015). The criteria for end of execution was selected as the end of the deceleration phase, where the lowest velocity reached below 100 mm/s was used.
In trials whereas no exact lowest velocity could be found, a plateau where the velocity didn’t change more than 5 mm/s within a 5 frames period was used as an additional criteria.
Reliability testing of data extraction was done for the parameters Latency time, Velocity peak and Total duration of movement execution. A stratified randomized sample of participants was used for the reliability testing. The extracted data came from one extension and one adduction task from 3 of the children born FT and 3 of the children born PT, in total 24 trials.
The data was independently extracted and compared. Inter estimator reliability was calculated
and resulted in a statistically significant correlation (r(72) = 1.00, p = .001). The high inter estimator reliability may be due to the authors initially practicing extracting data together.
Figure 3. Example of a velocity profile of the movement of a wrist (onset to touch), with velocity in mm/s and time in frames (1 sec = 120 frames). Illustration of variables; 1) Latency time (frames), 2) Velocity peak (mm/s), 3) end of Deceleration, 4) Duration of movement execution (frames) and 5) Total duration of movement execution.
Statistical data-analysis. Individual mean values were calculated from trial level, resulting in a mean value for each hand/side and task. Statistical analysis was made using STATISTICA 11 software. Series of factorial-ANOVAs were used for analysing kinematic outcomes (group [FT vs. MPT vs. VPT] x task; [horizontal vs. vertical] x side [right vs. left]).
For subgroup comparisons of background variables separate one-way ANOVAs were applied.
Normality of distribution for all data was initially assessed visually from histograms and was verified acceptable considering our sample and small groups. Since Levene’s test of homogeneity was significant between the three groups regarding some kinematic data, confidence interval for each kinematic parameter was reported and Welch F test was conducted. For continuous variables, means and SD are reported. Follow-up comparison by use of Scheffés post hoc test was applied whereas significant main and interaction effects were observed. The group differences present regarding the kinematic parameters are concurrent irrespective of which hand performs the task and which task that is performed. Since the horizontal task took longer time to finish (in comparison to the vertical task) it was found to be the more complex and decisive task to execute, outcomes of the horizontal task were chosen for added correlation analysis. To analyse the influence of false starts on kinematic outcome,
independent sample T-test were performed. To explore associations between GA, BW, kinematic outcomes, PA and cognitive abilities, Pearson’s correlations were performed for the groups PT and FT separately. Outliers were found only in the FT group. In the FT group three outliers were found by means of Latency time (Latency time > 719 s). These outliers were corrected for since such latency times are longer than expected and can be due to lack of attention on task. Loss of data, when corrected for False starts on trial level, were 38 trials in the FT group, 21 trials in the MPT group and 9 trials in the VPT group.
Results Group differences
Kinematic outcomes. Table 3 gives the information of group comparison of kinematic outcomes. Further analysis by means of 3x2x2 factorial ANOVAs regarding group differences with subgroups included (FT, MPT, VPT) and each kinematic parameter separately. Only significant results of interest are reported in the continuous text. For all outcome results see Table 3 (and Table 7A found in the Appendix A, for specificity).
Regarding onset Latency time, a significant main effect for groups was found, F(2, 288)
= 11.4, p = .001, ηp2 = .074. Scheffés post hoc test showed that Latency time was significantly longer (p = .001) for children born VPT (M = 323 ms, SD = 143) in comparison to children born FT (M = 256 ms, SD = 118) and children born MPT (p = .001) (M = 249 ms, SD = 83).
Latency time did not differ significantly between FT and MPT. Regarding onset Latency time, no significant interaction effects for group*task, group*side nor group*task*side were found, see Appendix A. Levene’s test of homogeneity of variance was significant (p = .001).
Regarding Duration of movement execution, a significant main effect for groups was found, F(2, 298) = 5.72, p = .004, ηp2 = .037. Scheffés post hoc test showed that Duration of movement execution was significantly longer (p = .037) for children born VPT (M = 646, SD
= 138) in comparison to children born FT (M = 598, SD = 131) and children born MPT (p = .027) (M = 590, SD = 111). Duration of movement execution did not differ significantly between FT and MPT. Regarding Duration of movement execution, no interaction effects for group*task, group*side nor group*task*side were found, see Appendix A.
Regarding Total duration of movement execution, a significant main effect for groups was found, F(2, 291) = 11.4, p = .001, ηp2 = .072. Scheffés post hoc test showed that Total duration of movement execution was significantly longer (p = .001) for children born VPT (M
= 968, SD = 207) in comparison to children born FT (M = 867, SD = 192) and children born MPT (p = .001) (M =837, SD = 134). Total duration of movement execution did not differ significantly between FT and MPT. Regarding Total duration of execution, no interaction effects for group*task, group*side nor group*task*side were found, see Appendix A. Levene’s test of homogeneity of variance was significant (p = .012). For further, more specific outcome information on all outcomes from the 3x2x2 factorial ANOVAs, see Appendix A, Table 7A.
Inhibitory control and false starts. The number of False starts in the group of children born FT was 67 out of 475 trials (M = 1.76, SD = 2.78), a frequency of 14%. In the group of children born MPT the number of False starts was 65 out of 262 trials (M = 2.71, SD = 2.99), a frequency of 25% and in the group of children born VPT the number of False starts was 17 out of 197 trials (M = 1.06, SD = 1.48), a frequency of 8.6%. Regarding number of False starts, a one-way ANOVA showed no significant group differences (p = .15), see Table 3.
Independent sample t-test showed no significant differences on any kinematic parameters between those who did have False starts and those who did not have any False starts neither in the FT (Latency time: p = .93; Velocity peak: p = .83; Deceleration: p = .37; Duration of movement execution: p = .43; Total duration of movement execution p = .49) nor in the PT
group (Latency time: p = .072; Velocity peak: p = .38; Deceleration: p = .30; Duration of movement execution: p = .81; Total duration of movement execution p = .17).
Table 3
Mean, (SD) and 95% Confidence Interval of Latency time, Velocity peak, Deceleration (%), Duration of movement execution, Total duration of movement execution and False starts, presented by subgroups.
Kinematic parameter
FT
(N=38) 95% CI MPT
(N=23)
95% CI VPT
(N=16)
95% CI p Effect size ηp2
False starts (N) 1.76 (2.78)
.82-2.61 2.71 (2.99)a
1.45-3.97 1.06 (1.48)
.27-1.85 .15 .050
Latency time H (ms)
253 (97.4)
228-264 243 (78.1)
218-266 324 (172)
267-390 .062 .072
Latency time V (ms)
265 (96.7)
242-284 265 (88.1)
230-283 318 (112)
277-358 .016 .055
Velocity peak H (mm/s)
1019 (152)
987-1056 1017 (119)
973-1052 1002 (124)
943-1050 .92 .002
Velocity peak V (mm/s)
622 (123)
596-649 565 (103)
534-596 590 (101)
553-626 .026 .048
Deceleration H (%)
53.8 (4.95)
52.9-55.2 53.8 (7.02)
51.2-55.6 53.9 (8.47)
50.6-57.5 1.00 .001
Deceleration V (%)
53.1 (6.68)
52.3-55.0 53.9 (5.96)
51.9-55.3 55.5 (7.47)
52.8-58.2 .21 .020
Duration of movement execution H (ms)
663 (123)
635-691 655 (95.2)
623-685 719 (112)
687-768 .19 .044
Duration of movement execution V (ms)
534 (99.4)
513-557 525 (72.5)
501-543 564 (111)
525-604 .18 .022
Tot duration of movement execution H (ms)
918 (171)
880-953 903 (131)
859-943 1040 (212)
978-1128 .031 .090
Tot duration of movement execution V (ms)
814 (199)
765-850 776 (102)
746-807 882 (171)
821-944 .027 .047
Note. FT = full term, MPT = moderately preterm, VPT = very preterm, ms = milliseconds, mm/s = millimetre per second, CI = Confidence interval, Xa = N(24). False starts represent the composite value from both the horizontal and vertical task, otherwise H = Horizontal task, V = vertical task. Significant p-values are indicated in bold. *No significant effect of side (right vs. left) nor any interactions were found, hence side is not reported separately.
Physical activity and cognitive abilities. Regarding number of sports, a one-way ANOVA showed a significant main effect for groups; F(2, 75) = 4.13, p = .020, ηp2 = .099.
Scheffés post hoc test showed a significant difference (p = .020) between MPT and VPT, with children born MPT participating in a higher number of sports compared to children born VPT.
Regarding Number of sports there was no significant difference between FT and MPT, nor between FT and VPT. Regarding PA rate, a significant main effect for groups was found, F(2, 73) = 3.88, p = .025, ηp2 = .096. Scheffés post hoc test showed a significant difference (p = .032) between MPT and VPT, with children born MPT having a higher PA rate compared to
children born VPT. PA rate did not differ significantly between FT and MPT, nor between FT and VPT. Regarding FSIQ based on the WISQ-IV scores, a significant main effect for groups was found: F(2, 62) = 5.04, p = .009, ηp2 = .14. Scheffés post hoc test showed that FSIQ was lower for children born VPT compared to children born FT (p = .010). FSIQ did not differ significantly between children born FT and MPT, nor between children born MPT and VPT, see Table 4.
Table 4
Group comparison Mean, (SD) and significance level for the three groups FT, MPT and VPT on background variables: Speed index, FSIQ, Tot sport and PA rate.
Variable FT
(N = 38)
MPT (N = 24)
VPT
(N = 15) p ηp2
Speed index 99.6 (12.9)a 100 (14.9)b 96.1 (13.8) .63 .014
FSIQ 103 (10.2)a 97.8 (13.0)b 92.0 (9.99) .009 .099
Tot sport 2.16 (1.13) 2.54 (1.38) 1.44 (1.03)c .020 .096
PA rate 40.5 (9.81) 42.0 (10.1) 33.4 (7.30)d .025 .050
Note. FT = full term, MPT = moderately preterm, VPT = very preterm, FSIQ = full-scale intelligence quotient, PA = physical activity, Xa = N(34), Xb = N(16), Xc = N(16), Xd = N(14). Significant p-values are indicated in bold.
Correlations between background variables and kinematics
The correlation outcome analysis between GA, BW, SI, FSIQ, Tot sport, PA rate and all kinematic parameters for the group of children born PT and FT respectively is shown in Table 5 and 6.
Full term. Among children born FT a significant positive moderate correlation was found between Latency time and Total duration of movement execution (r(34) = .71, p = .001), and between Latency time and Deceleration (%) (r(34) = .41, p = .016). No significant correlations could be seen between Latency time and Velocity peak, nor between Latency time and Duration of movement execution, see Table 5. No significant correlations were found between GA and/or BW and any kinematic parameters in the FT group, nor between FSIQ, SI or PA rate and any of the kinematic parameters, see Table 5.
Preterm. For the children born PT a significant modest negative correlation was found seen between Latency time and False starts (r(28) = -.40, p = .034). A significant positive correlation was found between Latency time and Total duration of movement execution (r(28)
= .88, p = .001). No significant correlations were found between Latency time and any of the other kinematic parameters, see Table 6. Significant correlations between GA and/or BW and several kinematic parameters were found in the PT group. For the children born PT there was a significant, moderate, negative correlation between GA and Latency time (r(28) = -.43, p = .024), and GA and Total duration of movement execution (r(28) = -.47, p = .012). No significant correlations could be found between PA rate nor Tot sport and any of the kinematic parameters, see Table 6. Significant correlations were found between both FSIQ and SI and several kinematic parameters, see Table 6. A moderate negative correlation was found between FSIQ and Duration of movement execution (r(28) = -.48, p = .010). A modest negative correlation was found between FSIQ and Total duration of movement execution (r(28) = -.40, p = .037). Moderate negative correlations were found between SI and Duration of movement execution (r(28) = -.54, p = .003) and between SI and Total duration of movement execution
(r(28) = -.46, p = .014). In addition, a significant moderate positive correlations was found between GA and Tot sport (r(28) = .41, p = 028.), between BW and Tot sport (r(28) = .46, p = . 013) and between BW and PA rate (r(28) = .42, p = .028), see Table 6.
Table 5
Within group correlation considering FT group, kinematic measure horizontal task only, r and (p-value) reported.
Full term N = 34
Variable
GA BW FSIQ SI PA
rate
Tot sport
FS LT VP D DoME
BW .44
(.009)
FSIQ -.11
(.52) -.31 (.078)
SI .15
(.41) -.26 (.14)
.72 (.001)
PA rate .20
(.25) -.16 (.37)
-.038 (.832)
.15 (.41)
Tot sport (N) .083
(.64)
-.040 (.82)
.012 (.95)
.19 (.28)
.58 (.001)
False starts (N) -.077 (.66)
.21 (.23)
.034 (.85)
.026 (.89)
-.089 (.62)
.076 (.67)
Latency time (ms) .16 (.41)
.087 (.63)
-.12 (.50)
-.032 (.86)
-.058 (.75)
-.14 (.44)
-.20 (.26)
Velocity peak (mm/s) -.043 (.81)
.026 (.88)
.043 (.81)
.033 (.852)
-.053 (.76)
-.11 (.55)
.13 (.48)
-.095 (.59)
Deceleration (%) -.005 (.98)
-.27 (.13)
-.015 (.93)
-.043 (.81)
.15 (.40)
-.21 (.24)
-.22 (.21)
.41 (.016)
-.22 (.21)
Duration of movement execution (ms)
.019 (.92)
.22 (.20)
-.20 (.25)
-.30 (.081)
-.15 (.39)
.014 (.94)
-.23 (.19)
.15 (.39)
-.69 (.001)
.009 (.96)
Tot duration of movement execution (ms)
.11 (.53)
.20 (.25)
-.20 (.25)
-.22 (.21)
-.13 (.45)
-.052 (.77)
-.29 (.097)
.71 (.001)
-.55 (.001)
.25 (.15)
.80 (.001)
Note. GA = gestational age, BW = birth weight, FSIQ = full-scale intelligence quotient, SI = speed index, PA = physical activity, FS = False starts, LT = Latency time, VP = Velocity peak, D = Deceleration, DoME = Duration of movement execution, ms = milliseconds, mm/s = millimetre per second. Significant results are indicated in bold.
Table 6
Within group correlation considering PT group, kinematic measure horizontal task only, r and (p-value) reported
Preterm N = 28
Variable
GA BW FSIQ SI PA
rate Tot sport
FS LT VP D DoME
BW .91
(.001)
FSIQ .32
(.092) .46 (.013)
SI .22
(.25) .20 (.31)
.75 (.001)
PA rate .33
(.085) .42 (.013)
.13 (.52)
.15 (.45)
Tot sport (N) .41
(.028) .46 (.013)
.20 (.31)
-.040 (.84)
.63 (.001)
False starts (N) .25 (.20)
.20 (.30)
.089 (.65)
.17 (.38)
-.060 (.76)
-.054 (.78)
Latency time (ms) -.43 (.024)
-.31 (.11)
-.19 (.33)
-.26 (.18)
.008 (.97)
-.12 (.53)
-.40 (.034)
Velocity peak (mm/s) .14 (.49)
.038 (.85)
-.060 (.76)
-.20 (.30)
.26 (.18)
.064 (.75)
.18 (.36)
-.22 (.28)
Deceleration (%) .29 (.88)
-.12 (.54)
.071 (.72)
.26 (.18)
-.20 (.31)
-.10 (.60)
-.065 (.74)
.21 (.29)
.26 (.18)
Duration of movement execution
-.35 (.070)
-.28 (.15)
-.48 (.010)
-.54 (.003)
-.22 (.25)
.025 (.90)
-.35 (.068)
.30 (.12)
-.37 (.052)
.41 (.032)
Tot duration of movement execution (ms)
-.47 (.012)
-.35 (.065)
-.40 (.037)
-.46 (.014)
-.092 (.64)
-.12 (.56)
-.46 (.014)
.88 (.001)
-.36 (.061)
-.084 (.67)
.70 (.001)
Note. GA = gestational age, BW = birth weight, FSIQ = full-scale intelligence quotient, SI = speed index, PA = physical activity, FS = False starts, LT = Latency time, VP = Velocity peak, D = Deceleration, DoME = Duration of movement execution, ms = milliseconds, mm/s = millimetre per second. Significant results are indicated in bold.
Discussion
The aim of this study was to investigate bimanual movement initiation and kinematic execution, and how these parameters associate with GA, amount of PA and cognitive abilities, in 6 to 8 year old children born PT compared to aged matched children born at term.
Kinematic performance
Our results provide support for the hypothesis of shorter latency time the higher the GA, with results showing that children born PT spend more time planning for execution of movement by means of longer latency times. Children born PT also show longer duration of
movement and longer duration of the task in total, indicating slower movements and thus less developed movement organization (Domellöf et al., 2018; 2013) compared to children born FT. Yet, the differences in movement performance found are represented by the children born VPT only, with weak to moderate effect sizes. These findings are in agreement with Domellöf et al. (2018) annotation of importance to not handle children born PT with different GA’s as a single group. In our results children born VPT show a tendency, though not significant, to spend more time in deceleration (in relation to acceleration). A longer time spent in deceleration meaning you are not fully knowing where you are going in the execution of a movement. These findings suggest that children born VPT need more time to pre-plan their movement, though without showing any improvements in movement organization, indicating less developed movement planning skills. Thus, in agreement Goyen et al. (2011) and Lönnberg et al. (2018) findings, adding that poorer movement planning performance are shown, not only in children born EPT or with stated DCD, but also in apparently typically developed children born VPT. This is supporting the idea that being born at lower GA is considered a greater risk factor for developmental motor impairments (Spittle, et al., 2016;
Allotey et al., 2017; Hebestreit & Bar-Or, 2001). Strengthening the hypothesis that decreasing GA within the PT population is related to problems with movement planning, our correlations indicate that within the group of children born PT, lower GA associate with subtle motor deficiencies in terms of worsen kinematic performances, with longer Latency times and Duration of movement execution. It is likely that these poorer movement planning and initiation performances found in the VPT group are represented by a neurodevelopmental delay or consequences of neurodevelopmental abnormalities due to a PT birth (Spittle et al., 2016;
Domellöf et al., 2013).
In contrary to our expectation, execution of the vertical task demonstrates more evident group differences. Considering the vertical task being less cognitive and motor skill demanding compared to the horizontal task, it suggests that movement impairments due to a VPT birth can be seen already at the initiation of a goal directed movement in little need of planning. When it comes to kinematic performance, cognitive ability and PA, children born MPT seem to be largely comparable to children born FT. Hence, children born prematurely is heterogeneous and should not be seen as one single group with the same strengths and weaknesses.
Just as we hypothesised, Réveillon et al. (2016) did find a positive relationship between GA and inhibitory control in children born PT. In the present study, however, no evident correlations in any group, nor any group differences regarding GA and inhibitory control (by number of False starts), were found. This is consistent with findings by Phua et al. (2012), suggesting all children at this age have the same problems with inhibitory control. Though not significant, an increased incidence of impulsive behaviour (by number of False starts) is found in the MPT group. An explanation to that can be the increasing risk of ADHD symptoms with decreasing GA (Allotey et al., 2017). The low prevalence of impulsive behaviour in children born VPT, together with their slower movements, could be interpreted as them being more passive, careful or inattentive. Since impaired EF has been found to explain attention problems (Aarnoudse-Moens et al., 2013), and according to our findings of lower scores on SI and lower IQ-scores in the VPT group, we are suggesting that the possible attention problems in the children born VPT could explain the longer latency times found. Our results indicate that the occurrence of impulsive behaviour does not seem to affect the ability of movement planning and execution per se. When task is carried out correctly (without False starts), the movement is well planned and executed.
The impact of physical activity and cognitive abilities on movement kinematics
Our results are supporting the previous findings of lower BW predicting lower PA participation in adults (Kaseva et al., 2012), with a positive correlation between PA rate and BW and the number of sport activities a child participates in and BW, in the PT group already at early school age. Additionally, our results indicate that the number of sport activities a child participates in seems to increase with increasing GA in the PT group, in contrast to findings by Spiegler et al. (2019). However, less engagement in PA does not seem to affect movement performance in our case. In contrast to the hypothesis that increasing sedentary behaviour could explain an, over time, increase of minor motor deficiencies in children born PT (Spittle et al., 2018; Van der Niet et al., 2015), our results show a higher representation of impulsive behaviour among the children being most physically active in the PT group. This suggests that PA may not improve inhibitory control and that worse inhibitory control seems not to impact on PA participation in early school age children born PT. One explanation is that, since children born MPT are more physically active, they also tend to be more physically reactive with shorter latency times, faster movements, and thus more impulsive behaviour.
Although it seems favourable for children born PT to engage in PA (Hebestreit & Bar- Or, 2001), our results show that children born VPT who would probably benefit from PA the most, are those participating the least. The hypothesis of PA working as a protective factor regarding movement performance for children born PT (Spittle et al., 2018) are not supported by our findings. Instead our results suggest that the amount of PA and initiation and execution of a goal directed upper limb movement are not evidently found related in early school aged children, thus supporting the findings by Matarma et al. (2018). The subtle, probable neurodevelopmental delays, found in early school aged children born VPT regarding movement planning might have implications on which type of sport the child chooses to take part in. This could explain the tendency of increasing participation in individual sports compared to team sports with decreasing GA seen in our sample (Rudberg & Granström, 2016).
Align with what Goyen et al. (2011) discussed our results suggest that PA and sports in general do not improve specific kinematic performance. Goyen et al. (2011) have suggested that specific interventions are needed to enhance motor performance. We argue that an organized motor training program targeting the impairments found in movement planning and speed might still be beneficial in children born PT (Alesi et al., 2016; Johansson et al., 2014; Qi et al., 2018). Perhaps children born VPT need more PA to have the same benefits from PA as children born MPT and at term. Since the children born MPT take part in a greater variety of sports and spend more time in PA compared to both children born VPT and FT, an alternative explanation for them showing comparable results to their term born peers could be that children born MPT have compensated for complications/delays in motor-development due to a PT birth following Hebestreit and Bar-Or (2001) recommendations. Further, parents to children born PT in Sweden have access to information about motor impairments related to a PT birth and hence, are encouraged to promote their child to take part in PA to stimulate motor development.
Thus, this could explain the higher participation in, and amount of, PA seen in the MPT group, making our results less generalizable to other countries. Parental and socioeconomic factors may also be an alienating factor to these findings.
According to our results FSIQ correlates with kinematic performance, aligning with the hypothesis that movement planning and initiation are action-perception-cognition dependent.
Our results partly confirm the finding that children born PT show lower cognitive abilities compared to children born FT (Allotey et al., 2017; Wolke et al., 2019), yet this was only true for children born VPT. The positive correlations between FSIQ and kinematic outcomes in children born PT support previous findings (Domellöf et al., 2013), adding the indication of a
relation between cognitive abilities and kinematic performance already at the pre-planning and early stages of initiation of a movement. Since FSIQ has been found to relate to kinematic performance, decision was made to study which sub index of WISC-IV measuring EF could be underlying the association between FSIQ and kinematic outcomes. Higher score on SI negatively correlating with Duration of movement execution and Total duration of movement execution, indicates that processing speed can explain some of the variance in kinematic performance, to no surprise since the task in hand challenges concentration, speed and ability to inhibit. Hence, not only does general cognitive level affect how effectively a child born PT carry out their movements, but also does EF in terms of processing speed when studying subtle movement kinematics. This adds more specificity to the previous found relationship between deficits in movement performance and cognition, supporting Goyen et al. (2011) hypothesis that problems in motor planning in children born EPT could be linked to cognitive processing problems.
Strengths and limitations
With many risk factors to consider when studying a vulnerable population such as children born PT, a strength of this study is the relatively homogenous sample. This made the exploration of subtle differences within healthy school aged children possible as well as dividing the sample into groups and subgroups. Additionally, the children born PT has an age and sex matched FT control making the results more valid. The possibility to assess subtle movement performance through kinematic registrations, gives this study an upper hand in assessing differences not available to the naked eye. Assessing cognitive abilities when examining differences in movement performances makes it possible to examine underlying factors of motor skills. Each participant had several trials on each task during the test situation, making it possible to redo the task if not understood correctly or when registration instruments failed.
A limitation of our study is the lack of inclusion of possible confounders. While environmental factors such as parental education, sensitivity of parenting and parental attitudes might weaken the associations between movement performance and EF (Wolke et al., 2019;
Spiegler et al., 2019), the associations found are not invalid. With our small sample, IQ and PA are considered the variables most relevant to control for. Additionally, the inclusion criteria of the ongoing research project from which our sample is taken (PI: L. Rönnqvist) were no known specific disorders such as brain injury and cerebral palsy. This inclusion criteria made it possible for us to investigate the effect of the premature birth itself. However, such disorders are common among children born PT (Wolke et al., 2019), which affects the generalizability of our results, and is important to note when comparing this study with other studies. Given our small sample of children born VPT and small effect sizes, the generalisability of our results should be made with caution at least outside of Sweden. Kinematic performance varies a lot within the group of children born VPT which is common when studying motor performance (Lönnberg et al., 2018). The EFAA test were not designed to primarily assess inhibitory control or impulsive behaviour. To further examine differences in inhibitory control and PA’s impact on kinematic performance more specific tests supplemented with kinematic registration would be preferable. Though, this test demonstrates and assess bimanual, goal directed movement performance, something that might be more generalizable to daily life actions since bimanual movements are frequent daily life actions. To our knowledge there are few studies examining pre-planning and early initiation of a movement through a bimanual task. We contribute with further knowledge of motor developmental deviations in children born PT, which is important for the refinement of kinematic registrations as a diagnostic instrument in a clinical setting.
Our data of PA patterns are based on parental ratings making the outcome sensitive to
subjective interpretation regarding their child’s and peers’ participation and skilfulness. The amount of PA a child participates in during the school day makes our measure of PA even more uncertain. More objective and sensitive measures could result in stronger correlations between amount of PA and kinematic outcomes, since children being more active during school time are often more active during leisure time (Philblad, 2017). To study PAs impact on movement performance further we suggest a more comprehensive PA measure to be used, with additional objective measures of daily activity such as accelerometer or logbook. Further research in this area should focus on children’s sport performance, including more sensitive measures of skilfulness, which kinematic measurements on sport activities may offer. Thus, to see if and how subtle movement developmental delays negatively affects sport performance or achievements.
Conclusions
We introduce further evidence that children born VPT at 6 to 8 years of age express less developed movement planning skills compared to peers born MPT or at term. Thus, by adding new evidence of poorer movement performances already at the pre-planning and early initiation stages of goal-directed bimanual movements. We did not find any support for associations between movement initiation and execution and children’s general PA. However, poorer movement performance does not seem to affect the children’s sport participation reported by their parents but may still affect the kinematic quality of sport performance and execution in terms of precision and timing. Instead the deficits found regarding pre-planning and early initiation of movements might be explained by cognitive processing speed. The implications on daily life and sport performance are not of focus in this study, but something for future research to focus on. However, sport participation and amount of PA might influence motor impairments in a sample where diagnosed motor impairments are present. Motor training programmes focusing on more specific motor skills as well as challenging cognitive processing, including perception-action-planning, might promote the individual child’s motor and EF development, and hence minimizing negative impact of a VPT birth.
Reference list
Aarnoudse-Moens, C., Weisglas-Kuperus, N., Duivenvoorden, H., van Goudoever, J., &
Oosterlaan, J. (2013). Executive Function and IQ Predict Mathematical and Attention Problems in Very Preterm Children. PLoS ONE, 8. doi:10.1371/journal.pone.0055994 Alesi, M., Bianco, A., Luppina, G., Palma, A. & Pepi, A. (2016). Improving children's
coordinative skills and executive functions: the effects of a football exercise program.
Perceptual and motor skills, 122, 27-46. doi:10.1177/0031512515627527
Allotey, J., Zamora, J., Cheong‐See, F., Kalidindi, M., Arroyo‐Manzano, D., Asztalos, E., . . . Thangaratinama, S. (2017). Cognitive, motor, behavioural and academic performances of children born preterm: a meta‐analysis and systematic review involving 64 061 children.
BJOG: An International Journal of Obstetrics & Gynaecology, 125, 16-25. doi:101111/1471- 0528-14832
Bordin, I. A., Rocha, M. M., Paula, C. S., Teixeira, M. C. T., Achenbach, T. M., Rescorla, L. A.
& Silvares, E. F. (2013). Child Behavior Checklist (CBCL), Youth Self-Report (YSR) and Teacher's Report Form (TRF): an overview of the development of the original and Brazilian versions. Cadernos de Saúde Pública, 29, 13-28. doi:10.1590/S0102-311X2013000100004 Cliff, D. P., Okely, A. D., Smith, L. M. & McKeen, K. (2009). Relationships between
fundamental movement skills and objectively measured physical activity in preschool children. Pediatric exercise science, 21, 436-449. doi:10.1123/pes.21.4.436
Domellöf, E., Johansson, A. M., Farooqi, A., Domellöf, M. & Rönnqvist, L. (2013). Relations among upper-limb movement organization and cognitive function at school age in children born preterm. Journal of Developmental & Behavioral Pediatrics, 34, 344-352.
doi:10.1097/DBP.0b013e318287ca68
Domellöf, E., Johansson, A. M., & Rönnqvist, L. (2018). Developmental progression and side specialization in upper-limb movements from 4 to 8 years in children born preterm and fullterm. Developmental neuropsychology, 43, 219-234.
doi:10.1080/87565641.2018.1426765
Goyen, T. A., Lui, K., & Hummell, J. (2011). Sensorimotor skills associated with motor dysfunction in children born extremely preterm. Early human development, 87, 489-493.
doi:10.1016/j.earlhumdev.2011.04.002
Hebestreit, H. & Bar-Or, O. (2001). Exercise and the child born prematurely. Sports Medicine, 31, 591-599. doi:10.2165/00007256-200131080-00004
Holmes, N. P. & Dakwar, A. R., (2015). Online control of reaching and pointing to visual, auditory, and multimodal targets: effects of target modality and method of determining correction latency. Vision research, 117, 105-116. doi:10.1016/j.visres.2015.08.019
Johansson, A. M., Domellöf, E., & Rönnqvist, L. (2014). Timing training in three children with diplegic cerebral palsy: short- and long-term effects on upper-limb movement organization and functioning. Frontiers In Neurology, 5. doi:10.3389/fneur.2014.00038
Kaseva, N., Wehkalampi, K., Strang-Karlsson, S., Salonen, M., Pesonen, A.-K., Räikkönen, K.,
… Kajantie, E. (2012). Lower Conditioning Leisure-Time Physical Activity in Young Adults Born Preterm at Very Low Birth Weight. PLoS One, 7. doi:10.1371/journal.pone.0032430 Khan, N. A., & Hillman, C. H. (2014). The Relation of Childhood Physical Activity and
Aerobic Fitness to Brain Function and Cognition: A Review, Pediatric Exercise Science, 26, 138-146. doi:10.1123/pes.2013-0125
Lowe, J., Watkins, W. J., Kotecha, S. J. & Kotecha, S. (2016). Physical activity and sedentary behavior in preterm-born 7-year old children. PloS one, 11.
doi:10.1371/journal.pone.0155229
