1
Physical fitness, motor skill and physical activity relationships in Grade 4 to 6 children
2 3
Richard Larouche1,2, M.Sc.; Charles Boyer1, M.A.; Mark Stephen Tremblay1,2,3, Ph.D., FACSM;
4
Patricia Longmuir1,3, Ph.D.
5 6
1
7
Research Institute, 401 Smyth Road, Ottawa, Canada, K1H 8L1
8
2
9
6N5
10
3
Healthy Active Living and Obesity Research Group, Children’s Hospital of Eastern Ontario
School of Human Kinetics, University of Ottawa, 125 University Private Ottawa, Canada, K1N
Department of Pediatrics, University of Ottawa, 401 Smyth Road, Ottawa, Canada, K1H 8L1
11 12
E-mail
13
(
[email protected]); Mark S Tremblay (
[email protected]); Patricia E Longmuir
14
(
[email protected])
addresses:
Richard
Larouche
(
[email protected]);
15 16
Author for correspondence:
17
Richard Larouche, M.Sc., Ph.D. candidate
18
Children’s Hospital of Eastern Ontario Research Institute
19
401 Smyth Road, Room R242
20
Ottawa, ON, K1H 8L1
21
Phone: 1-613-737-7600 ext 4191
22
Fax: 1-613-738-4800
23
E-mail:
[email protected]
1
Charles
Boyer
24
Abstract
25
The present study sought to quantify the relationships among physical activity (PA), health-
26
related fitness and motor skill in children (Grades 4 to 6), and to determine whether specific tests
27
of fitness or motor skill are independently associated with objectively-measured PA level. 491
28
students (56.4% female) wore a Digi-Walker pedometer for 7 consecutive days. Standardized
29
protocols were used to assess health-related fitness (BMI percentile, waist circumference, 20m
30
shuttle run, plank, handgrip, and trunk flexibility). Motor skill was evaluated using a validated
31
obstacle course. Pearson correlations (with Holm adjustments for multiple comparisons) initially
32
assessed associations among PA, health-related fitness and motor skill. Multi-variable linear
33
regression was used to determine which factors were significantly associated with daily step
34
counts, while adjusting for gender, age, testing season and socio-economic status. Step counts
35
were significantly correlated with predicted aerobic power (r=0.30), obstacle course time (r=-
36
0.27), obstacle course score (r=0.20), plank isometric torso endurance (r=0.16), and handgrip
37
strength (r=0.12), but not with waist circumference (r=-0.10), trunk flexibility (r=0.10) or
38
overweight status (rho=-0.06). In the multi-variable model, predicted aerobic power, obstacle
39
course time, testing season, gender and the predicted aerobic power by gender interaction were
40
significantly associated with step counts, explaining 16.4% of the variance. Specifically, the
41
relationship between predicted aerobic power and step counts was stronger in girls. These
42
findings suggest that aerobic fitness and motor skill are independently associated with children’s
43
PA. Future longitudinal studies should evaluate whether interventions to enhance aerobic fitness
44
and motor skill could enhance daily PA among children of this age.
45
Keywords: children, pedometer, health-related fitness, motor proficiency, correlates, multi-
46
variable model
2
47
Résumé
48
Cette étude visait à quantifier les associations entre l’activité physique (AP), la condition
49
physique et la motricité chez les enfants de la 4è à la 6è année et à déterminer si certains tests de
50
condition physique ou de motricité sont associés indépendamment à l’AP mesurée
51
objectivement. 491 étudiants (56,4% de filles) ont porté un podomètre Digi-Walker pendant 7
52
jours consécutifs. Des protocoles standardisés ont été employés pour mesurer la condition
53
physique (percentile d’IMC, circonférence de taille, test navette, planche, force de préhension et
54
flexibilité du tronc). Les habiletés motrices ont été évaluées avec un parcours d’obstacles validé.
55
Un modèle de régression multiple a été utilisé pour identifier les facteurs associés
56
significativement au nombre de pas par jour en contrôlant pour le genre, l’âge, la saison et le
57
niveau socioéconomique. Le nombre de pas par jour était significativement corrélé avec la
58
puissance aérobie estimée (r=0,30), le temps pour compléter le parcours d’obstacles (r=-0,27), le
59
score sur le parcours d’obstacles (r=0,20), l’endurance musculaire évaluée avec la planche
60
(r=0,16) et la force de préhension (r=0,12), mais pas avec la circonférence de taille (r=-0,10), la
61
flexibilité du tronc (r=-0,10) ou l’embonpoint (rho=-0,06). Dans le modèle multivarié, la
62
puissance aérobie estimée, le temps pour compléter le parcours d’obstacles, la saison, le genre et
63
l’interaction entre la puissance aérobie et le genre étaient significativement associés au nombre
64
de pas par jour, expliquant 16,4% de la variance. La relation entre la puissance aérobie estimée et
65
le nombre de pas par jour était plus forte chez les filles. Ces résultats suggèrent que la puissance
66
aérobie et la motricité sont indépendamment associées avec l’AP des enfants. De futures études
67
longitudinales devraient évaluer si des interventions visant à améliorer la puissance aérobie et la
68
motricité peuvent accroître l’AP des enfants de cet âge.
69
3
70
Mots clés : enfant, podomètre, condition physique, habiletés motrices, corrélats, modèle
71
multivarié
4
72 73
INTRODUCTION Population studies indicate that most North American children fail to meet current
74
physical activity (PA) recommendations of 60 minutes of daily moderate-to-vigorous physical
75
activity (Colley et al. 2011; Troiano et al. 2008). Moreover, during the last few decades
76
significant decreases in childhood physical fitness and large increases in the prevalence of
77
overweight and obesity have been reported in many countries (Lobstein et al. 2004; Tomkinson
78
et al. 2003; Tremblay et al. 2010). Even in children and youth, insufficient physical activity and
79
poor cardiovascular fitness are independent cardiovascular disease risk factors (Ekelund et al.
80
2012). These risk factors are also likely to track from childhood to adulthood (Bao et al. 1994).
81
This epidemiological evidence clearly suggests that PA should be promoted at an early
82
age. While some controlled trials have been shown to achieve short-term increases in PA (van
83
Sluijs et al. 2007), little is known regarding the long-term effectiveness of PA interventions
84
(Trudeau and Shephard 2005). This underscores a need for a better understanding of the
85
correlates of PA, and the extent to which they vary with characteristics such as gender and age
86
(Bauman et al. 2002, 2012). The correlates of PA can be divided into different ‘levels of
87
influence’ (e.g. individual factors, interpersonal factors, built environment, etc.) following an
88
ecological approach (Bauman et al. 2012). Within this categorization, one’s physical fitness and
89
motor skill would fit among the individual factors.
90
An earlier literature review found generally weak associations between self-reported PA
91
and aerobic fitness (Morrow and Freedson 1994). The limited validity and reliability of self-
92
report instruments has been proposed as an explanation for these findings (Welk et al. 2000).
93
Recently, the use of objective measures of PA has become increasingly common, and a recent
94
systematic review reported a consistent low-to-moderate association between accelerometry-
5
95
measured PA and cardiovascular fitness (Dencker and Andersen 2011). In parallel, another
96
systematic review concluded that there is strong evidence supporting an association between PA
97
and fundamental motor skills (Lubans et al. 2010). However, earlier studies were generally
98
concerned with how PA influences fitness outcomes (or motor skill) and have typically assessed
99
the link between PA and different outcomes separately. Hence, there remains a lack of data
100
examining multiple health-related fitness and motor skill variables as correlates of PA levels
101
within a single multi-variable model.
102
Therefore, the primary objective of the current study was to determine which indicators
103
of health-related fitness and motor skill are independently associated with pedometer-determined
104
PA in a multi-variable model adjusted for potential confounders (age, gender, season of the year
105
and school area socioeconomic status). Of note, motor skill was assessed in a dynamic setting
106
using a recently developed and validated obstacle course. The second objective was to determine
107
the strength of the associations between PA and indicators of physical fitness and motor skill in
108
children.
109 110 111
METHODS Participants. 784 students in grades 4 through 6 from 7 different schools in the Ottawa
112
(Canada) region were invited to participate in the Canadian Assessment of Physical Literacy
113
(CAPL) study. The study was approved by the Children’s Hospital of Eastern Ontario Research
114
Ethics Board and both participating school boards. All students assented to study participation
115
and written informed consent was provided by a parent/guardian. Consent and assent were
116
obtained from 679 (53.2% female) participants, a response rate of 86.6%. Of these, 71.3% (n =
117
484) were recruited in rural schools and 28.7% (n = 195) in urban/suburban schools.
6
118
Measures. Data collection was performed during the fall of 2011 and winter/spring 2012.
119
The primary outcome was steps per day measured by a Digi-Walker SW200 pedometer (Deep
120
River, ON) that was worn over the iliac crest for 7 consecutive days. Participants completed a
121
daily log sheet recording their daily step counts as well as the time the pedometer was worn
122
during waking hours. Digi-walker pedometers have high validity and reliability for measuring
123
steps taken (Crouter et al. 2003) and have previously been used for population-based studies of
124
PA levels in Canadian children and youth (Craig et al. 2010).
125
Cardiovascular fitness was measured with the Progressive Aerobic Cardiovascular
126
Endurance Run (PACER) test (Welk and Meredith 2008); the 15m course was used with most
127
participants, and results were converted to be equivalent to the 20m course for analysis with the
128
conversion chart available at http://www.fitnessgram.net/PACER_conversion.pdf. Initially, an
129
examiner explained and then demonstrated performance of the PACER test. Children were
130
instructed to continue running as long as possible, even if other children have stopped. The
131
PACER test was conducted in small groups (1 to 8 children per examiner depending on available
132
gymnasium space) to allow more individual attention. Upper body muscle strength was assessed
133
with a Takei handgrip dynamometer (A5001), with maximal grip strength assessed alternately,
134
twice with each hand (Canadian Society for Exercise Physiology 2003). Trunk flexibility was
135
evaluated with the sit and reach test (Canadian Society for Exercise Physiology 2003). Torso
136
muscular endurance was measured with the plank isometric position test, in which participants
137
were asked to maintain a straight body position (as shown in Figure 1) for as long as possible.
138
The plank test has substantial convergent validity with other measures of strength and substantial
139
intra-rater, inter-rater and test-retest reliability in 8 to 12 year old children (Boyer et al. 2013).
7
140
Body composition measurements were performed following the Canadian Society for
141
Exercise Physiology (2003) protocols. Specifically, height was measured to the nearest 0.1 cm,
142
without footwear, using a portable stadiometer (SECA, Hamburg, Germany). Weight was
143
measured to the nearest 0.1 kg with a digital scale (A&D Medical, Milpitas, CA) with the
144
participants wearing light clothing without footwear. Waist circumference was measured to the
145
nearest 0.5 cm at the mid-point between the top of the iliac crest and the bottom of the last
146
floating rib, and at the end of a normal expiration.
147
Motor skills were assessed with a newly developed obstacle course that was developed to
148
facilitate in-school testing of groups of children in the context of physical education classes
149
(Figure 2). Existing motor skills assessments, such as the Test of Gross Motor Development
150
(Ulrich 1985) or Movement ABC (Henderson et al. 2007), are relatively more time-consuming;
151
thereby undermining their feasibility in this context. These traditional assessment batteries
152
require the separate performance of each individual skill, with initiation of the skill from a static
153
position. In contrast, the obstacle course assessment is a dynamic setting, incorporating both
154
object control and locomotor skills while the child moves through a timed and scored obstacle
155
course. The time needed to complete the obstacle course is measured with a stopwatch, and the
156
score is determined with a systematic observation grid that contains 20 process- and product-
157
oriented motor skill criteria. Pilot data indicate modest convergent validity with the Movement
158
ABC test for the time and score components (Longmuir et al. 2013). Testing also showed
159
substantial inter-rater reliability for the time component, as well as inter-rater reliability and
160
intra-rater reliability for the score (Longmuir et al. 2013). Participants completed the obstacle
161
course 5 times; the first 3 trials were practice trials and the last 2 trials were timed and scored.
162
All health-related fitness and motor skill assessments were conducted by trained research staff.
8
163
Data Treatment. Pedometer data were excluded if participants failed to meet the
164
following established criteria: 1) between 1,000 and 30,000 steps/day (Rowe et al. 2004); 2) at
165
least 10 hours of data/day (Eisenmann et al. 2007) and 3) at least 3 days of valid data meeting the
166
aforementioned threshold values (Tudor-Locke et al. 2005). 535 participants returned their
167
pedometer log sheet (78.8% of the total sample) and application of the thresholds led to the
168
exclusion of 44 participants. Thus, acceptable pedometry data were available from 491
169
participants (72.3% of the total sample), including 214 boys and 277 girls. Average daily step
170
counts (hereafter referred to as step counts) were used as the measure of PA in the analyses. As
171
recommended in the Fitnessgram / Activitygram Reference Guide (Welk and Meredith 2008),
172
the Léger equation (Léger et al. 1988) was used to estimate aerobic power based on PACER test
173
results. For the handgrip test, the sum of the best trial for each hand was computed. Similarly, the
174
best performances were used in the analyses for the sit and reach and the obstacle course. For the
175
obstacle course, the time and score components were considered to be independent because both
176
higher and lower scores could be associated with slower times. For example, a participant who
177
drops the ball or makes a mistake hopping/jumping (i.e., lower score) will take more time to
178
complete the course, as will someone who goes more slowly in order to concentrate on accurate
179
skill performance. Body mass index (BMI) was computed from measured height and weight (e.g.
180
kg/m2) and converted into percentiles based on the Centers for Disease Control (2000) growth
181
charts.
182
As a proxy for socioeconomic status (SES), the median household income of the census
183
tract for each school location was obtained from the 2006 Canadian census (Statistics Canada
184
2006). Four schools were located in census tracts in which the median income was lower than
9
185
the Ottawa-Gatineau region average (low SES schools, n = 251 participants) and three were
186
located in census tracts with above average median income (high SES schools, n = 240).
187
Variables found to be skewed (handgrip, plank time, waist circumference and obstacle
188
course time) were transformed into their natural logarithm in order to achieve quasi-normal
189
distributions. The BMI percentile distribution was highly skewed to the left even after log-
190
transformation; therefore participants were categorized by weight status (overweight: 85th
191
percentile or higher; non-overweight: less than 85th percentile).
192
Statistical analyses. To examine the study participants for potential biases in our sample,
193
t-tests and chi-square tests were used to investigate differences between participants who
194
provided acceptable pedometry data versus those who were given a pedometer, but either did not
195
return their log sheet or had step counts that did not meet inclusion criteria. Differences between
196
children who did not receive a pedometer and those who provided acceptable pedometer data
197
were examined the same way. Second, Pearson correlations were used to assess the relationship
198
among average daily step counts and the health-related fitness and motor skill variables (a
199
Spearman correlation was used for weight status). Holm’s (1979) method was used to adjust for
200
the multiple comparisons of step counts with these 8 variables (weight status, waist
201
circumference, predicted aerobic power, plank, handgrip, trunk flexibility, obstacle course time
202
and obstacle course score). Third, a linear regression analysis with backward variable selection
203
using the General Linear Model (GLM) procedure (IBM SPSS 19) was used to determine which
204
fitness and motor skill variables were significantly associated with step counts, while adjusting
205
for gender, age, season of the year and school area SES (dichotomized as high vs. low).
206
Variables and interaction terms with p values > 0.05 were manually removed from the model.
207
Effect sizes were determined with the proportion of explained variance (R2), using the
10
208
conventions proposed by Cohen (1988) (0.01 = small effect; 0.06 = moderate effect; 0.15 = large
209
effect). Within the final model, the independent effect size of each predictor was determined with
210
the partial eta-squared (η2) statistic with Cohen’s (1988) conventions (0.01 = small effect; 0.06 =
211
medium effect; 0.14 = large effect).
212 213 214
RESULTS: First, differences between the participants who provided acceptable pedometry data and
215
those who did not were examined. Participants who did not provide acceptable pedometry data
216
(n = 141) took more time to complete the obstacle course (17.7 vs. 16.5 seconds; t = 3.75; p