Relationship between Fundamental Movement Skills and Physical Activity in Preschool-aged Children: A Systematic Review

Preschool-aged children are in a critical period of developing fundamental movement skills (FMS). FMS have a close link with physical activity (PA). This study aimed to systematically review the associations between FMS and PA in preschool-aged children. Searching Cochrane Library, Web of Science, PubMed, ScienceDirect, and EBSCO (including SPORTDiscus, ERIC, and Academic Search Premier) was utilized to conduct a systematic review of the available literature. Studies were included if they examined associations between FMS and PA among typically developing children aged 3–6 years, published between January 2000 and April 2020. A total of 26 studies met the inclusion criteria, including 24 cross-sectional studies and two cohort studies. There was a strong level of evidence to support low to moderate associations between moderate to vigorous physical activity and components of FMS, specifically, the total FMS (r = 0.11–0.48, R2 = 16%–19%) and object control skill (r = 0.16–0.46, β = 0.28–0.49, R2 = 10.4%–16.9%). Similar associations were also found between the total physical activity and components of FMS, specifically, the total FMS (r = 0.10–0.45, R2 = 16%), locomotor skills (r = 0.14–0.46, R2 = 21.3%), and objective control skills (r = 0.16–0.44, β = 0.47, R2 = 19.2%). There was strong evidence that there is no significant association between light physical activity and FMS, specifically, total FMS and locomotor skills. The associations, including “stability skills–PA” and “locomotor skills–moderate to vigorous PA”, were uncertain due to insufficient evidence. Our findings provide strong evidence of associations between specific FMS components and a specific PA intensity. Future studies should consider using a longitudinal study design in order to explore the causal relationship between specific-intensity PA and the FMS subdomain.


Introduction
Engaging in sufficient physical activity (PA) is linked with a healthy body weight [1,2], cardiorespiratory and muscular fitness [2,3], various cognitive effects [4], and the development of fundamental movement skills (FMS) [5,6] in preschool-aged children. Previous studies have shown that individuals who achieve moderate intensity PA gain more substantial health benefits [7], especially preschool-aged children [8]. The World Health Organization (WHO) recommend that children aged three to six years old participate in at least 60 min of moderate-vigorous physical activity (MVPA) every day [8,9]. However, current evidence suggests that less than half of preschoolers in the USA [10]

Materials and Methods
The process of completing and reporting this review adhered to the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) statement guidelines [37].

Search Strategy
A literature search of all electronically archived literature published was conducted in six electronic databases, namely, Cochrane Library, Web of Science, PubMed, ScienceDirect, EBSCO (including SPORTDiscus, ERIC, and Academic Search Premier). Search terms were based on the combination of four parts: (1) preschool* OR kindergarten* OR "early child*" OR "young child*"; (2) ("motor and skill* OR abilit* OR competence OR performance OR proficiency") OR ("movement and skill* OR abilit* OR competence OR performance OR proficiency") OR (locomotor OR "object control" OR "manipulat*" OR stability); (3) "physical activit*"; (4) associate* OR relat* OR predict* OR indict* OR corelat* OR effect*. Full-text peer-reviewed articles in English that were published from January 2000 to April 2020 were collected. Supplementary Materials Table S1 shows the search strategies used in the five databases.

Screen and Selection
After the removal of duplicates, the search results were reviewed and screened by title and abstract with reference to pre-determined inclusion and exclusion criteria, followed by the retrieval of the full texts for evaluation by two authors (FX and JTH). A third author (YL) was consulted to resolve inconsistencies. Studies were selected according to the eligibility criteria.
The inclusion criteria were as follows: the study population had to (1) include typically developing preschool children aged three to six years; (2) objectively assess and report at least one component of fundamental movement skill using a product-oriented or/process-oriented instrument, or both; (3) quantify at least one certain intensity of PA or total PA by a subjective or objective instrument; (4) report associations between FMS and PA and carry out a statistical analysis to report associations including correlations or regression; and, (5) be observational in design.
The exclusion criteria were as follows: (1) the study population was identified as having a pathological condition or disability that affects movement, such as developmental coordination disorder (DCD) or autism spectrum disorder (ASD); (2) participants were infants or were beyond the targeted age range; (3) the study was experimental or review in design; (4) the assessment of motor skills only included fine motor skills or physical fitness; and, (5) the study only reported PA outcomes in terms of PA types (i.e., outdoor PA, recreational PA).

Assessment of Methodological Quality
The methodological quality of cross-sectional studies was assessed by the Agency for Healthcare Research and Quality (AHRQ) scale [38]. Each study was evaluated by eleven items (I-XI, see Table 2) with two responses (yes = 1, no and unclear = 0). If quality assessment scores were from 8 to 11, then the study was considered as high methodological quality. Scores from 4 to 7 were considered as moderate quality, and scores under 3 were considered to be low quality. The methodological quality of cohort studies was assessed using the Newcastle-Ottawa Scale (NOS) [39]. The judgment was based on star scores, with a score of 7 or more stars indicating high quality, 4-6 stars moderate quality, and 0-3 stars low quality. Two authors (FX and JTH) independently assessed the quality of all included studies, and the third author (STC) resolved their disagreements if they existed.

Data Extraction and Synthesis
This systematic review did not conduct a meta-analysis due to the considerable heterogeneity across the included studies [40]. We used a normal data extraction form to collect the following information: first author, publication year, country, participant details, assessment and measurements of PA and FMS, and study findings. The data were extracted by two authors (FX, STC), and inconsistent data were resolved by the third author (YJC). The degree of association was interpreted while using Cohen's [41] convention as low (r/β = 0.10-0.29), moderate (r/β = 0.30-0.49), or high (r/β ≥ 0.50), and the proportion of shared variance (R 2 ) was interpreted as small (R 2 = 1%-8%), medium (R 2 = 9%-24%), or large (R 2 ≥ 25%).

Methodological Quality of Studies
Agreement rate for the assessment of methodological quality was 88.7% for the 282 items (24 studies × 11 items and 2 studies × 9) between two authors. Disagreements were resolved by discussions until the consensus was reached with the third author. Tables 2 and 3 show the methodological quality assessment per quality item and per study. In cross-sectional studies, a total of 14 studies [5,17,21,22,24,25,28,30,34,44,46,47,50,53] were categorized to be of high quality, and 10 studies [26,27,31,33,45,48,49,51,52,54] were classified to be of moderate quality. There were no included cross-sectional studies that were classified as low quality. In addition, two cohort [32,43] studies were categorized to be of high quality.  Notes: I: Define the source of information (survey, record review); II: List inclusion and exclusion criteria for exposed and unexposed subjects (cases and controls) or refer to previous publications; III: Indicate time period used for identifying patients; IV: Indicate whether or not subjects were consecutive if not population-based; V: Indicate if evaluators of subjective components of study were masked to other aspects of the status of the participants; VI: Describe any assessments undertaken for quality assurance purposes (e.g., test/retest of primary outcome measurements); VII: Explain any patient exclusions from analysis; VIII: Describe how confounding was assessed and/or controlled; IX: If applicable, explain how missing data were handled in the analysis; X: Summarize patient response rates and completeness of data collection; XI: Clarify what follow-up, if any, was expected and the percentage of patients for which incomplete data or follow-up was obtained. Notes: 1a: Representativeness of the exposed cohort; 1b: Selection of the non-exposed cohort; 1c: Ascertainment of exposure; 1d: Demonstration that outcome of interest was not present at start of study; 2a: Cohorts comparable on basis of age; 2b: Cohorts comparable on other factor(s); 3a: Assessment of outcome; 3b: Follow-up was long enough for outcomes to occur; 3c: Adequacy of follow up of cohorts. *: one star could be awarded for the item if the study met the specific quality criterion.

Gender-Specific Aspects
Only three studies [27,34,49] found that there was a gender-specific relationship between PA and FMS, and the correlations in boys were more positive and stronger than those among girls (r = 0.24-0.55) [27,34,49]. Cliff et al. [34] revealed that only OCS specifically accounted for MVPA in boys (r = 0.48, R 2 = 0.169), while girls' LMS was negatively related to MVPA (r= −0.50, R 2 = 0.192). Total FMS were moderately correlated with MVPA in boys (r = 0.48), whereas there was a moderate negative correlation between total FMS and MVPA in girls (r = −0.46) [34]. Laukkanen et al. [49] found that the correlation between total FMS and TPA was higher in boys (r = 0.448) than in girls (r = 0.138). Temple et al. [27] observed moderate correlations between LMS, OCS, and SS with TPA in boys (r = 0.33-0.41), while these associations did not exist in girls.

Intensity-Specific PA and Total FMS
When compared with single or subdomain FMS, total FMS reflected the overall level of preschoolers' motor skills. A total of five studies [21,24,28,33,48] examined the relationship between total FMS and LPA. Only one [33] of the studies demonstrated a negative association between total FMS and LPA (r = −0.23); however, non-significant correlations were reported in four studies [21,24,25,28]. Overall, summary coding showed a strong level of evidence supporting the lack of a significant relationship between total FMS and LPA (see Table 5).  [54] ascertained that the total FMS explained 19% of the variance in MVPA. A total of 12 studies [5,21,22,[24][25][26]44,46,47,49,50,53] examined the bivariate relationship regarding the TPA and total FMS. Eight [21,[24][25][26]44,46,49,53] of nine studies demonstrated a low to moderate correlation between total FMS and TPA (r = 0.10-0.448). In addition, the regression results of Iivonen et al. [22] showed that the total FMS explained 16% of the variance in MVPA and TPA, respectively. Overall, summary coding presented a strong level of evidence supporting low to moderate positive associations among MVPA, TPA, and total FMS (see Table 5).

Intensity-Specific PA and Subdomain Skills
In terms of LPA, only Gu [33] found that the assessment of LMS (r = −0.18, β = −0.19) and OCS by TGMD−2 identified weak negative correlations with LPA (r = −0.18). Several other studies showed non-significant correlations between LMS [21,28,31,32], OCS [28,31,32], and SS [31,32] and LPA. Overall, summary coding presented a sufficient level of evidence in order to support the presence of a non-significant relationship between LMS and LPA, while the level of current evidence to support non-significant relationships among OCS, SS, and LPA was uncertain (see Table 5).

Discussion
The aim of this study was to systematically review the associations between aspect of FMS and PA in preschool-aged children. This study is the first review to synthesize and analyze the associations among specific types of FMS and specific intensities of PA in young children. A total of 26 observational studies met the eligible criteria, including 24 cross-sectional studies and two cohort studies. Sixteen studies were evaluated to be of high methodological quality and 10 studies were determined to be of medium quality. In summary, there was strong level of evidence to support a low to moderate positive association between FMS (specifically total FMS and OCS) and MVPA in preschool-aged children, while there was an inconsistent association between LMS and MVPA. Sufficient evidence supported a low to moderate association between FMS (specifically, total FMS, LMS, and OCS) and TPA. There was also sufficient evidence to support the lack of significant associations among LMS, total FMS, and LPA, and a similar trend of no significant association between OCS and LPA was demonstrated, although the evidence that was presented in the included studies was relatively insufficient. Finally, this review could not identify a significant or non-significant association between SS and any intensity of PA due to a lack of sufficient evidence. Regarding the gender-specific aspect, the difference in associations between boys and girls requires further evidence to be clarified. Our findings are important, because understanding specific relationships between FMS and PA may be significant for improving specific skills or a specific PA intensity.
Regardless of gender, this review found a strong level of evidence to support low to moderate positive associations between the total TPA and components of FMS, specifically LMS, OCS, and total FMS. This finding is consistent with the conclusion of Logan et al. (r = 0.16 to 0.48; R 2 = 3%-23%, n = 4) [35]. This evidence generally verifies the hypothesis of Stodden et al., who developed a conceptual model showing that the association between variable levels of PA and FMS is weak in the early childhood period [16]. However, we have insufficient evidence to support another hypothesis from Stodden et al., which stated that PA might drive the development of motor skill competence. Although one cohort study ascertained that the total TPA at baseline was a significant predictor of total FMS, we need more longitudinal studies in order to verify this assumption. We tried to speculate the reasons for the low to moderate associations between total FMS and TPA while using the perspective of multidimensional factors. Using the ecological prospective, there are various significant factors related to PA and FMS, such as the immediate environment, structured physical education, parental influence, and climate. Welk [55] categorized determinants of PA into five aspects-personal, biological, psychological, social, and environmental-where motor skills were only a factor at the biological level of PA. Niemistö et al. [56] also identified that individual (age, body mass index, temperament), family (parents' education level), and environmental (access to sports facilities) factors were associated with motor skills development in children. Another probable explanation is that the key role of mediating variables (e.g., perceived motor competence, health-related physical fitness, and obesity) might decline/promote or interact with the relationship between FMS and PA [16,57,58]. Only two [30,47] of the included studies explored this mediating effect among preschoolers, but the results were not significant. One relevant explanation for this is that young children tend to exaggerate their perceived competence level relative to their actual motor competence, because they are unable to make social comparisons, differentiate between actual and real self-concept, and take the perspective of others [59]. However, although perceived competence in young children was considered to be of limited accuracy, inflated perceived competence could improve motor skills to some extent, because they were sure that they would be skillful. It is necessary to determine the roles of perceived motor competence and other mediating factors in future studies in order to clarify the mechanism of the relationship and broaden the scope of motor skills. In general, positive but weak associations in early childhood certainly support an emerging developmental relationship.
Regarding the gender-specific aspect, the differences in associations between components of FMS and specific intensities of PA in boys and girl were uncertain due to limited evidence. Evidence in cross-sectional studies suggested that boys are more physically active than girls [10,11,60] and have greater OCS [5,21,27,61], whereas girls' LMS were usually higher than those of boys [5,34,61]. Therefore, it is necessary to add evidence of gender differences from the perspective of specific gender interventions. The included studies comparing this association provided inconsistent conclusions. Some studies indicated moderate associations between LMS, OCS, and TPA [34], OCS and MVPA for boys [34], while other studies suggested a non-significant [27,49] or negative association between FMS and PA among girls [30]. On the one hand, FMS probably plays a more prominent role in the prerequisites for PA participation in boys [62]; in other words, boys are keener to participate in skill-oriented activities. On the other hand, the stronger relationship is partly due to the preferential type of OCS assessment, which, for boys, was ball sports (e.g., soccer, basketball, cricket, baseball) [63] in most of the current studies. However, for girls, the results showed no association of FMS with PA due to the lack of balance, rhythm, and preferred skills in assessments. Negative associations in girls were unusual, and this might have been caused by sample limitations (e.g., sample size, age, demographics) in the investigations [34]. Most of the current studies did not compare gender differences in the relationship and this observation might be due to the use of a single type of global movement skills score rather than using categories for the movement skills scores (for LMS, OCS, or SS) [24]. The lack of gender difference might be related to the age and the stage of cohort, as preschool-aged children are in early childhood and their FMS have not yet matured. Relevant evidence needs to be supplemented by more research in the future, given the importance of gender-targeted interventions.
There was a strong level of evidence to support a low to moderate positive association between FMS (specifically total FMS and OCS) and MVPA in preschool-aged children, while there was a inconsistent association between LMS and MVPA. The WHO proposed PA guidelines for children under five years old for the first time, which recommend that children should spend at least 60 min participating in MVPA per day [8]. From the perspective of PA promotion, it would be worthwhile to explore the respective contributions of OCS and LMS to MVPA. The study of Hall et al. [25] suggested that children who achieved ≥60 min of MVPA had significantly better OCS scores. Similar to previous studies in middle childhood [13,64], it appears that OCS, rather than LMS, is associated with high levels of MVPA in early childhood [17,25,30]. There is sufficient evidence regarding why MVPA is more closely related to OCS, but not LMS. It has been suggested that the ball skills component of OCS is fundamental to participation in various games and sports that involve OCS (i.e., soccer or baseball) [13]. Barnett et al. [13,57] found that the performance of OCS in childhood, as well as in adolescence, was significantly associated with time spent in MVPA in adolescence. The similar explanation in this review demonstrates that the importance of OCS manifests in preschool-aged children. The evidence provided in this review suggested suggests that there are non-significant or even negative correlations between LPA and components of FMS, specifically total FMS and LMS. The evidence supporting the lack of an association between OCS and LPA is insufficient. It is possible that this uncertain association was caused by researchers not reporting result data or not including the variable of LPA. In terms of TPA, it seems that LMS and OCS are equally positively related to TPA.
The use of different types of PA (subjective vs. objective) and FMS assessment tools (product-oriented vs. process-oriented) might have a crucial impact on the comparability of study results. Therefore, these factors should be interpreted and discussed carefully. Accelerometers are as the "gold standard" for objective measurements, and they have been proven to be reliable and valid when used in young children [65]. The most included studies measured PA by accelerometers. However, the different sampling frequencies and cut points used in studies [24,30] affected the classification of PA intensity, which may have led to the underestimate of multidirectional and sporadic PA and was not conducive to comparisons among study results. Two studies [26,51] used pedometers, which is, they measured PA by the number of steps and they could not classify PA. Nevertheless, when compared with objective measurements, subjective instruments (e.g., parent-proxy questionnaires [27,48], and diaries) can be used to measure different PA types, such as recreational activities and skill-based activities, but their results tended to be overestimated or even inaccurate. Crane et al. [30] suggested that asking parents to complete a detailed physical activity log by observation to accompany the accelerometer data would go some way to unpacking this relationship. In fact, it is largely insufficient if studies just measure the amount of different intensities of PA completed in discrete time periods. Some intervention studies found that structural or organized PA could significantly improve FMS competence. Hence, studies examining this association should make a comparison between organized and non-organized PA and they should even access skill-related PA.
The relationship between FMS competence and PA behaviors might be influenced by the orientation of the assessment tool. Webster et al. [5] determined that total FMS and LMS scores assessed by TGMD-3 (process-oriented) were associated with more vigorous physical activity time. In contrast, there was no significant association between PA and FMS measured by MABC (product-oriented). The performance of process-and product-oriented assessments of FMS differs across skills and age groups (r = 0.26-0.88) [66], which complicates the comparison of results. In process-oriented assessments, the comparison of a child with an "expert" performer often results in ceiling effects and floor effects, which reduces the validity. In product-oriented assessments, product measurements do not observe the developmental movement process related to the movement product [16]. Notably, process-oriented assessments seem to get more attention and have greater use by researchers. The series of TGMD tools developed by Ulrich [67,68] is one of the most widely used assessments of FMS, but it does not contain SS, which is regarded as an indispensable component of FMS. The CMSP and PE Metrics also do not involve SS. Of course, there are some assessments measuring not only LMS and OCS, but also SS, such as Motorische Basiskompetenzen (MOBAK) [69], PGMQ [70]. In addition to measuring locomotor and object control skills with TGMD-2, Nilsen et al. [32] also measured dynamic and static balance using the PGMQ. Breaking the limitations of single assessments and combining scores from different assessments might improve scientific evaluation. These observations highlight the necessity of adopting a multidimensional method to understand FMS in preschoolers and combining both product-and process-oriented assessments to deepen the understanding of the delicate differences in the relationship between FMS and PA.
Our systematic review has several critical strengths. Firstly, this review performed a more in-depth analysis of associations between different types of FMS and specific intensities of PA in preschool-aged children than previous reviews. These specific associations are critical for providing a reference for the experiment design of interventions. Secondly, the methodological quality of all included studies was medium to high, which gave rise to reliable evidence. Thirdly, the strength of the evidence was classified in this review to provide more dependable evidence. This study also has some limitations. Firstly, we did not take mediators of the association between FMS and PA into account, although a few studies did analyze the effect of mediators in the association. Secondly, a meta-analysis was not conducted due to the lack of a uniform effect size and adequate studies. In addition, few studies took the influences of other variables, such as demographics, social-economic status, body mass index, and race into account. Therefore, we solely examined the relationship between FMS and PA.

Conclusions and Implication
This systematic review provides strong evidence of the rapidly increasing amount of literature examining associations between FMS and PA. The review found a strong level of evidence to support a low to moderate positive association between FMS (specifically total FMS and OCS) and MVPA in preschool-aged children, whereas there was an inconsistent association between LMS and MVPA. There was sufficient evidence that TPA is significantly and positively associated with components of FMS (specifically total FMS, LMS, and OCS) to a low to moderate degree. There was also sufficient evidence to support the lack of significant associations among LMS, total FMS, and LPA; meanwhile, a similar trend of no significant association was demonstrated between OCS and LPA, although the evidence of included studies was relatively insufficient. However, due to lack of sufficient evidence, this review could not identify a significant or non-significant association between SS and any intensity of PA. In addition, the differences in associations between boys and girl were uncertain due to lack of sufficient evidence. The general findings of this review should be interpreted with caution due to the use of different populations from various cultural backgrounds, the diversity and complexity of the assessment, and the varying levels of methodological quality.
Evidence from cohort studies examining cause and effect relationships was insufficient, but these findings were valuable for examining the intervention study design, especially regarding the assessment of different components of FMS and the specific intensity of PA based on gender. In future cross-sectional studies, it is necessary to add studies examining the association between LMS and MVPA and the association between SS and the specific intensity of PA. Researchers need to utilize a more uniform or mixed assessment of FMS that include not only LMS and OCS but also SS in order to provide more accurate and comprehensive findings in the future. Evidence suggests that accelerometers may underestimate the intensity of PA in manipulative skill performance, and we recommend the use of more scientific sampling frequencies and accelerometer cut off points in future studies.

Conflicts of Interest:
The authors declare no conflict of interest.