Introduction

Physical activity is widely accepted to be an important feature in the promotion of health and well-being [1]. In view of the mounting evidence of health-related benefits of regular physical activity among children [2–4], concerns have been raised about decreasing physical activity and fitness levels in children and adolescents [5, 6]. New research reveals that regular physical activity not only contributes to improved physical health [7], but also has a beneficial effect on children’s cognitive functioning, such as executive function [8], attention [9], and academic performance [10], all of which are important conditions for gains in academic performance. Yet, physical education has been neglected in many low- and middle-income countries (LMICs), while more time is being allocated to academic subjects [11]. In South Africa, some teachers and parents believe that participation in physical activity might interfere with learners’ academic success [12]. Educators often teach several subjects and do not feel confident enough to systematically instruct in sports and exercise as part of a school program. Indeed, they do not feel able to offer attractive and didactically well-conducted physical education classes. Furthermore, some of them lack the motivation to be physically active. Hence, physical education is not being taught in most public schools. As a result, a considerable number of children do not engage in the recommended daily 60 min of moderate-to-vigorous intensity physical activity [13]. According to Walter [14], this might be one of the key contributors to the increasing level of physical inactivity among South African school children.

While low physical activity levels constitute an important issue from a public health perspective [15], children growing up and living in low socioeconomic environments are faced with a multitude of challenges that may jeopardize their health and well-being [16]. These challenges include insufficient hygiene, lack of clean water, and inadequate sanitation, factors that all favor helminth and intestinal protozoa infection [17]. This, in turn, can cause abdominal pain, diarrhea, growth retardation, anemia, cognitive impairment, poor academic performance and reduced physical fitness [18, 19]. Of note, the effects of mass deworming on children’s cognition and school performance remain ambiguous [20]. Another potentially harmful effect of living in deprived environments is the commonly limited access to health care in the absence of universal health coverage. Consequently, increased risk of illness causes more frequent school absenteeism, and hence, reduced academic exposure [21]. Poor nutrition and repeated infection causes stunting, which in turn has been found to be associated with motor development problems and poor cognitive development resulting in low intelligence quotient [22]. Taken together, these factors can obstruct children’s ability to process information, concentrate, and focus on academic work [21].

While researchers have pointed toward the potential of physical activity and physical education to increase cognitive performance and academic performance [23–25], there is a paucity of studies on the effects of school-based physical activity intervention programs on children’s schooling outcomes in disadvantaged areas from LMICs. To fill this gap, the aim of the present study was to examine the impact of a 20-week school-based physical activity intervention on selective attention and academic performance on children attending schools in disadvantaged areas in Port Elizabeth, South Africa.

Materials and methods

Study area, school selection, and randomization

The DASH (Disease, Activity, and Schoolchildren’s Health) study was carried out in quintile three schools, located in disadvantaged communities of Port Elizabeth, South Africa. All quintile three primary schools in the Port Elizabeth district (n = 103) were invited to participate in the study. South African public schools are classified into five groups, with quintile five standing for the least poor and quintile one standing for the poorest. The quintiles are determined through the national poverty table, prepared by the Treasury. Areas are being ranked on the basis of income levels, dependency ratios, and literacy rates in the area. The quintile ranking of a school determines the no-fee status of the school and also the amount of money that a school receives, with the poorest schools receiving the greatest per-learner allocation [27].

From the 103 quintile three schools, 25 schools expressed an interest in the form of a written response. Those 25 schools were invited to an information sharing meeting, and 15 schools were represented, whereas 10 schools declined to attend the meeting. The final eight schools were selected based on (i) large grade four classes (n >100); (ii) geographical location; (iii) representation of the various target communities (black and colored inhabitants); and (iv) commitment to support the project activities. The seven schools that were excluded had all numbers <100 in grade four. Due to financial constraints, logistics, and limited manpower, we were able to implement the physical activity intervention program only in three schools. From the final eight schools, the DASH core team therefore randomly selected three intervention and five control schools, on the basis of a computer-generated random number list. A computer generated random number list was also used to allocate the intervention and control schools to one of five intervention/control conditions (see Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Intervention measures at the eight primary schools in Port Elizabeth, South Africa carried out between February 2015 and May 2016.

https://doi.org/10.1371/journal.pone.0206908.t001

Participants

As shown in Fig 1, 1009 grade four primary school children were assessed at baseline in February 2015. Data sets with complete records in the main outcome variables concentration performance (CP), error percentage (E%), and end of the year results (EoYR) were available for 663 learners at post-intervention in May 2016, after including only children aged 8–13 years. The mid-follow-up is not included in the present paper.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Participant flow chart.

Notes: EoYR: end of the year results, average of the four subjects home language, additional language, mathematics, and life skills.

https://doi.org/10.1371/journal.pone.0206908.g001

Procedures

An overview of the study design is provided in Fig 2. Recruitment started in September 2014, and the practical intervention study took place between February 2015 and May 2016. The study length was restricted to 15 months due to financial reasons.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Overview of study design and the 20-week physical activity intervention, Port Elizabeth, South Africa carried out between February 2015 and May 2016.

https://doi.org/10.1371/journal.pone.0206908.g002

At the beginning of the study, project information sessions with principals, teachers, parents, and school governing bodies were held. Workshops with life orientation teachers and class teachers were organized to discuss the materials and information provided during the intervention.

Statistical analysis

Data were double-entered and validated using EpiData version 3.1 (EpiData Association; Odense, Denmark), and merged into a single data file. Statistical analyses were performed with STATA version 13.0 (STATA; College Station, United States of America) and with SPSS version 23 (IBM Corporation; Armonk, United States of America) for Windows.

Sample size calculation is described in the study protocol [26]. In brief, sample size was based on the prevalence of soil-transmitted helminth infections at baseline, taking into account clustering within schools and classes as well as loss to follow-up.

Descriptive statistics are provided as frequencies (n) and percentages (%) for categorical variables, and means (M) and 95% confidence interval (CI) for metric variables, separately for the total sample, and learners of the intervention group (IG) and control group (CG).

To determine whether a 20-week school-based physical activity intervention had an effect on selective attention (E% and CP) and academic performance (EoYR), three separate mixed linear regression models were employed with random intercepts for school classes, in order to adjust for cluster effects. This was facilitated using the multilevel mixed effects linear regression procedure (covariance structure = independent) in STATA version 13.0 (STATA; College Station, United States of America). In all three regression analyses, indicators of selective attention and academic performance at post-intervention were used as outcome/dependent variables. Before testing the effect of the intervention condition (IG versus CG), all regression analyses were controlled for baseline levels of selective attention and academic performance and potential confounders (gender, stunting, anemia, intestinal protozoa and soil-transmitted helminth infections, age, BMI, SES, self-reported physical activity, grip strength, and VO₂ max).

To interpret the findings, the unstandardized Beta coefficients and the 95% confidence intervals (CIs) are displayed. Statistical significance level was set at p<0.05 across all analyses. Some missing data occurred in the covariates in less than 5% of all possible data entries. Detailed missing data inspection revealed no systematic missing data pattern or significant correlations. Hence, it was decided not to impute missing data and the mixed linear regression models were based on data of children with complete data records.

Results

Table 2 summarizes the descriptive statistics of selective attention and academic performance (at baseline and post-intervention) and of the potential confounders (at baseline), separately for the total sample, IG, and CG. A first inspection revealed that the descriptive measures indicate few noticeable differences between the IC and CG at baseline. The groups seemed to differ in levels of self-perceived physical activity (IG: M = 4.03; 95% CI: 3.73 to 4.33] versus CG: M = 2.97; 95% CI: 2.74 to 3.21), concentration (CP) (IG: M = 51.33; 95% CI: 47.68 to 54.98 versus CG: M = 58.34; 95% CI: 55.74 to 60.94) and academic performance (EoYR) (IG: M = 3.90; 95% CI: 3.75 to 4.06 versus CG: M = 4.93; 95% CI: 4.80 to 5.06).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Descriptive statistics of all variables at baseline in February 2015 and outcome/dependent variables also at post-intervention in May 2016, Port Elizabeth, South Africa.

https://doi.org/10.1371/journal.pone.0206908.t002

Effect of the physical activity intervention on concentration performance

The results of the mixed linear regression models are provided in Table 3, after adjustment for clustering effects of school classes. In Model 1, concentration performance (CP) at post-intervention was used as outcome/dependent variable. After adjusting for confounding variables, this model shows that the physical activity intervention had no statistically significant effect on children’s CP (B = 2.93, 95% CI:-5.01 to 10.86, p = 0.469). Model 1 shows that baseline CP was statistically significant and positively associated with children’s post-intervention CP scores (p<0.001). Children scored 0.8 points higher at post-intervention per additional CP point at baseline. Moreover, Model 1 reveals a significant influence of some cofounding variables. At post-intervention, girls had a 7.05 point higher CP score than boys, and younger children scored 4.65 points higher per additional year of age. Furthermore, the CP increased by 1.27 per score of self-reported physical activity. Better grip strength and higher VO₂ max were independently and positively associated with higher CP at post-intervention. The mean CP score increased by 1.16 points per kg grip strength and by 0.77 points per ml·kg^-1·min^-1 VO₂ max.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Effect of intervention condition on the three outcome variables at post-intervention (May 2016), after adjustment for clustering effects of school classes, controlling for baseline levels (February 2015) of selective attention and academic performance, and potential confounders, Port Elizabeth, South Africa.

https://doi.org/10.1371/journal.pone.0206908.t003

Discussion

The most important finding of the present study is that a 20-week physical activity intervention had a positive effect on children’s EoYR. Indeed, the academic performance remained stable in children in the IG, whereas a decrease by half a grade was observed in the CG. Yet, no effects of the physical activity intervention were found on selective attention.

Our results are in line with previous studies showing maintenance and/or a smaller decline of academic performance in children participating in a physical activity intervention, compared to those experiencing no change in physical activity levels [39, 40]. Our findings thus indicate that physical activity promotion may be a strategy to maintain academic performance with increasing demands. Other studies found enhancing effects within this domain [41]. For example, Hollar et al. [42] reported improved reading and mathematics skills after a 1-year physical activity intervention, in combination with a nutrition intervention in elementary school children from low-income families. Similarly, Chaya et al. [43] found that a 3-month physical activity intervention, including yoga and physical education lessons, had a positive impact on cognitive performance (arithmetic, coding, and vocabulary) in socioeconomically disadvantaged Indian school children. Szabo-Reed et al. [44] reported that moderate to vigorous physical activity lessons were significantly associated with more on-task time behavior.

In the present study, a possible explanation for the decreased academic performance in the CG could be the transition of language of instruction, also referred to as the “fourth-grade slump” [45]. The baseline results refer to the final grades of grade three learners and the post-intervention results refer to the final grades from grade four. In South Africa, most schools offer mother-tongue instruction for the first three grades of school, while the transition to English as the language of instruction occurs in grade four [46]. Children undergo a shift from “learning to read” to “reading to learn”, in conjunction with a change in the language of instruction. Cummins [47] argues that only after the mastery of the first language a child will have the necessary skills to transition to a second language. Thus, our findings are in line with previous studies [45, 48], which show that academic performance is decreasing in South African school children progressing from grade three to grade four. Hence, one interpretation of the present study’s results could be that increasing in-school physical activity levels in third grade schooling holds promise to counteract a negative tendency of academic performance as transitioning to English language instruction.

Our findings further suggest that self-perceived physical activity, cardiorespiratory fitness, and grip strength were independently associated with selective attention. The first finding accorded well with previous studies showing that higher physical fitness levels are associated with better cognitive performance. For instance, our results are similar to a study conducted by London et al. [49], in which overall physical fitness predicted academic performance. In a recent review, Donnelly et al. [41] stated that increased cardiovascular fitness and physical activity has been associated with improved cognitive function, brain structure and function, and academic performance. Although cardiorespiratory fitness and physical activity are not identical concepts [50], physical activity can be considered as a proxy measure for physical fitness, particularly as fitness is a potential outcome of regular physical activity participation [51]. Hence, our findings corresponds with earlier studies showing that a positive relationship exists between physical activity and cognitive performance and academic performance [40]. However, as attention remained unchanged, this cognitive domain cannot explain why a decline in academic performance was observed in the CG only. Although children with higher physical fitness levels and higher self-reported physical activity seem to have an advantage in paying attention compared to their less fit peers, this does not necessarily guarantee the maintenance of academic performance. Evidence from longitudinal studies suggests that physical activity has an indirect effect on academic performance through a pathway of executive function (i.e., top-down control of behavior, especially in non-routine situations) [41]. The direct effect of physical activity on this higher-order cognitive function is well documented [52] and has been attributed to morphological changes (i.e., angiogenesis, neurogenesis, and synaptogenesis) in brain regions that are important for learning [53]. Another pathway that has been discussed as a potential mechanism underlying executive function benefits is the exercise-induced psychological stimulation that contributes to the improvement of self-control ability, which may impact on academic performance [54]. Although speculative, it seems that changes in higher-order cognitive functions rather than attention may have contributed to the maintenance of academic performance in the IG. This suggests that physical activity, and therefore physical fitness, may positively affect important brain areas that stimulate cognition and as a result give fit children an advantage compared with their less fit peers [39].

In contrast, Spitzer et al. [55] and Adsiz et al. [56] found that a physical activity program has the potential to enhance selective attention. Hence, study results seem to vary considerably, most likely due to differences in the mode of assessment of selective attention, the setting in which the study took place, or the nature and intensity of the physical activity intervention. In the present study, the intervention might not have resulted in a positive effect because the intervention period was relatively short (total of 20 weeks) and because of the context in which the intervention took place (e.g., large class sizes, heterogeneous student population in terms of age and academic performance). These factors have complicated the implementation of the intervention and may have hindered the detection of subtle effects.

To our knowledge no study has examined the effect of a physical activity program on socioeconomically underprivileged children’s selective attention, only acute effects have been reported [57]. Given these findings, new research is needed to deepen the understanding of whether and how a physical activity might be associated with SES.

Our findings further show that older children have significantly lower scores for selective attention and academic performance. This may be explained by the fact that disadvantaged communities do not have the financial means to promote children with special needs or learning disabilities [58]. Children suffering from attention deficit hyperactivity disorder (ADHD), fetal alcohol syndrome, reading difficulties, or neglect, might not get the required academic support and, subsequently, they might not be able to keep up with their peers. Learners with inadequate grades are retained up to three years until they become too old and automatically progress to the next grade [58, 59]. Moreover, girls seemed to achieve better academic results compared to boys and achieve higher scores for selective attention. This is in line with a meta-analysis by Voyer and Voyer [60], in which a consistent female advantage with regards to school grades was found for all subject content areas.

Our study has several limitations. First, academic performance was operationalized by the average end of the year grade (achieved at the end of grade three and four), which corresponds to the summary of four subjects (i.e., mathematics, home language, additional language, and life skills). While the objectivity of school grades can be questioned as a reliable outcome in empirical research (e.g., due to attributions or stereotypes of the teachers and/or different standards between classes/schools) [61], this measure has a high ecological validity because sufficiently high grades are needed for academic promotion, and the present study showed that selective attention and the academic performance scores were moderately correlated (r >0.30).

Second, allocation to the intervention and control condition was done school-wise. We are aware that random allocation at class level is the ‘gold’ standard, but this was difficult to achieve in the present study. For instance, one component of the physical activity intervention was the creation of a “physical activity friendly” school environment. Thus, changes in the infrastructure were performed, which cannot be isolated for learners from specific classes. Nevertheless, we considered school class as random factor in our multivariate regression analyses to account for the variation in academic performance and intervention implementation between schools and classes.

Third, we used an indirect measurement of VO₂ max to assess aerobic fitness and children might not have performed to their best abilities due to lack of motivation. However, this standardized test was chosen because it is well-suited for a resource-constrained setting due to its ease of application [30].

Fourth, physical activity was assessed with a single self-reported item about children’s physical activity levels. While it can be questioned whether children are able to accurately respond to this item, previous data from the present study have shown that children with higher physical activity levels indeed report higher health-related quality of life [62], have lower blood pressure scores, and are less likely to be overweight or obese [63].

Fifth, the physical activity intervention module was well perceived by the children and teachers alike. Yet, the level of the teacher’s compliance and adherence toward a high intervention quality varied considerably. For instance, some of the teachers in the intervention schools did not have a high motivation to be physically active. Furthermore, the length of the intervention was relatively short. Hence, it may be that a longer intervention period is needed to positively impact selective attention among primary school children.

Sixth, as acknowledged previously, the present study took place in disadvantaged communities (quintile three schools). Consequently, variation in SES was limited, which might have resulted in an underestimation of SES as a predictor of selective attention and academic performance.

Seventh, on a descriptive level, baseline differences were apparent between the IG and CG in academic performance, although these schools were similar in size and student population. Such baseline differences pose a challenge with respect to data analyses, as the impact of the intervention could be interpreted as a regression to the mean. In other words, since learners of the CG had higher scores at baseline, there was also more scope for a decrease compared to peers from the IG.

These shortcomings should be addressed in future research by either using study designs that allow a class-wise group assignment or by controlling more systematically for academic performance when schools are selected to ensure that no between-school differences exist in children’s EoYR prior to the beginning of the intervention program.

Conclusion

Participation in a 20-week physical activity intervention implemented in disadvantaged schools in Port Elizabeth, South Africa, was positively associated with children’s academic performance. Our findings suggest that such a physical activity intervention has the potential to counteract decrease in academic performance. Hence, reintegration of physical education into the curriculum might have beneficial effects for children’s academic performance. Yet, this conclusion needs to be interpreted with caution because the intervention period was relatively short, compliance was uneven, and there were differences in academic performance between the IG and CG at baseline. Future research is needed in disadvantaged schools with a physical activity intervention being carried out over an extended period of time and allocation procedures used that minimize the risk of baseline group differences.

Supporting information

Acknowledgments

This study has been funded by the Swiss National Science Foundation (SNSF; project no. IZLSZ3 149015) and the National Research Foundation (NRF; project no. 87397) in South Africa. The funders had no role in the study design, data collection and analysis, preparation of the manuscript, or decision to publish. We are thankful to the children and parents or guardians for their willingness to participate in this study. Special thanks are addressed to the education authorities of the schools, including the indispensable support and collaboration of the teaching staff, community volunteers, and the respective school health services. Thanks are also due to Mrs. Leyli Zondie and Mrs. Lindsey Beyleveld and her team of the Department of Medical Laboratory Sciences at the Nelson Mandela University. Furthermore, we are grateful to all the participating Master- and Honors-students for their dedicated contribution to the long and intensive data collection.

All authors have no financial or no other relationships that could appear to have influenced the submitted work.