How are academic achievement and inhibitory control associated with physical fitness, soil-transmitted helminth infections, food insecurity and stunting among South African primary schoolchildren?

The present paper is based on cross-sectional baseline data of the KaziAfya cluster randomized controlled trial (registration date: August 9, 2018, with ISRCTN; https://​doi.​org/​10.​1186/​ISRCTN29534081) on the effectiveness of increased physical activity and multi-micronutrient supplementation. The baseline data assessment took place from January to April 2019. Based on an a-priori power analysis [25], we intended to recruit 1320 South African primary schoolchildren from grades 1–4 (aged approx. 6–12 years).

Four quintile 3 schools located in peri-urban areas in Port Elizabeth (South Africa) were recruited. South African public schools are classified into five groups, with quintile 1 standing for the poorest and quintile 5 for the least poor school environment. School authorities were first contacted before contacting schools through the school principals who were informed about the objectives, procedures and potential risks/benefits of the study.

Written informed consent was obtained from the parents/guardians of the children before the start of the baseline data assessment. After having signed written informed consent, the parents/guardians were interviewed regarding the families’ SES and dietary intake. Additionally, all children participating in the study provided oral assent before the start of the study.

Schools were eligible for the study if they were public quintile 3 schools located in marginalized areas, if facilities were available for the implementation of physical education lessons and if they did not engage in any other research project, clinical trial, or were part of governmental nutrition interventions. Children had to meet the following inclusion criteria: (i) attend grade 1–4; (ii) not be older than 12 years; (iii) have written informed consent from their parents/guardians; (iv) not participate in other research projects, clinical trials, or food/nutritional programmes; and (vi) not suffer from clinical conditions that prevent participation in physical activity, as determined by qualified medical personnel.

Ethical approval was obtained from the ‘Ethikkommission Nordwest- und Zentralschweiz’ in Switzerland (EKNZ; reference number: Req-2018-00608). The intervention study was registered in the ISRCTN registry (https://​www.​isrctn.​com/​ISRCTN29534081). Approval has also been obtained in South Africa from the research ethics committee (REC-H) of the Nelson Mandela University in Port Elizabeth (reference number: H18-HEA-HMS-006) and the Eastern Cape Department of Education of the Eastern Cape Province. Children excluded after diagnosis of a relevant medical conditions and/or children suffering from malnourishment (as diagnosed by a nurse, following national guidelines) were referred to the nearest local clinic.

Data collection took place over several days at the schools between January and April 2019. Data assessment procedures used here were based on a set of standardised, validated and quality-controlled standard operating procedures [25]. Hand filled data sheets were double-entered and validated using EpiData (version 3.1, EpiData Association; Odense, Denmark) and merged into a single database.

Normal distribution of the collected data was examined by applying the Kolmogorov-Smirnov and Shapiro-Wilk tests. Descriptive statistics are reported as M and SD or n and percentage (%), for the total sample, and separately for boys and girls. Differences between boys and girls in the study variables were tested with univariate analyses of variance (ANOVAs) or χ²-tests. Kolmogorov-Smirnov, Shapiro-Wilk tests, descriptive statistics, ANOVAs and χ²-tests were performed with SPSS Version 26 (IBM Corporation, Armonk NY, USA) for Mac.

A series of different mixed linear regression models with random intercepts for school classes were performed to examine the extent to which children’s age, SES, body mass index (BMI), food insecurity, stunting, STH infection status (not infected versus a single or multiple infection), grip strength, and CRF explained inhibitory control and academic achievement. After detection of considerable sex differences (see results section for more details), separate analyses were carried out for boys and girls. Mixed linear regression models were calculated with the Mplus software (version 7, Muthen & Muthen, 1998–2020) with a robust maximum likelihood estimator (MLR). These main analyses were based on the full sample, hereby handling missing data via full information maximum likelihood (FIML). To examine the prerequisites of FIML, we performed Little’s missing completely at random (MCAR) test, using SPSS. With regard to inhibitory control, the analyses were repeated, after exclusion of children who did not perform higher than chance (≤50%) in the Flanker task, to reduce the potential bias introduced by a failure to understand and follow task instructions. To interpret the findings of the mixed linear regression analyses, the following statistical coefficients were displayed: (a) Estimate (standardized Beta-weight), (b) standard error (S.E.) of the estimate, and (c) p-value. For all statistical analyses, the level of significance was set at p < 0.05.

From the 1369 children with informed consent, 65 dropped out before the baseline data assessment took place, most of them due to relocation or because they left or changed school. Of the remaining 1304 children, 27 were not considered in the data analysis because no information on sex was available. Accordingly, the final sample of the present paper consisted of 1277 children (664 boys, 613 girls), with a mean age of 8.3 years (SD = 1.4). Sample characteristics and descriptive statistics of all independent and dependent variables are presented in Table 1. With the exception of reaction time for congruent and incongruent stimuli, Kolmogorov-Smirnov and Shapiro-Wilk tests indicated that none of the dependent variables were normally distributed (p > 0.05). Hence, we used MLR in the mixed linear regression analyses to handle non-normal distribution of dependent variables. Table 1 also shows that the number of missing values differed considerable across study variables. Thus, whereas 1277 children had valid data for age, only 754 had valid data on SES. Little’s MCAR tests showed that data were missing completely at random, χ²(df = 1406) = 1434.5, p = 0.293, so FIML could be applied to impute missing data.

Table 1 shows that boys and girls differed in several study variables. More specifically, boys were taller, and had lower academic results. However, boys performed better in some indices of the Flanker task, achieving faster reaction times. No significant differences were found with regard to SES, food insecurity, STH infections, stunted growth, grip strength and CRF.

Table 2 shows that among boys, few factors were associated with their academic achievement. Higher SES was related to better end of the year results and grades in life skills and mathematics. Moreover, higher CRF was associated with better grades in mathematics, whereas stunted growth was related to poorer grades in the life skills subject.

A different picture emerged for girls (Table 2). Consistently across all academic domains, higher CRF was associated with higher grades. Furthermore, stunted growth was linked to poorer grades in language and mathematics.

Table 3 provides an overview of factors that were associated with information processing and inhibitory control of attention among boys. Across all Flanker task performance indicators, age was the most important explanatory factor. Thus, higher age was associated with better information processing (higher accuracy and lower reaction time on congruent trials) and inhibitory control of attention (higher accuracy and lower reaction time on incongruent trials). Additionally, among boys, children with higher grip strength performed better in all Flanker task indices, also after excluding boys who did not reach accuracy levels higher than chance.

The results for girls are displayed in Table 4. Similar to boys, higher age was associated with better performances across all Flanker task indices. Contrary to boys, grip strength was not associated with the Flanker task outcomes in a consistent way. Importantly, our analyses show that higher CRF was associated with better information processing and inhibitory control of attention as reflected by higher accuracy/lower reaction time on congruent and incongruent trials, respectively. Finally, girls who were infected with STH had lower accuracy scores in response to congruent stimuli than non-infected girls. However, no significant differences were found in the other Flanker task domains.

As mentioned above, the present data revealed significant sex differences with regard to academic achievement and inhibitory control. While girls had higher grades across all school subjects, boys achieved a faster reaction time with consistent accuracy in the Flanker task. Based on the standards of Cohen [43] who defined differences with 1–5.8% of explained variance as small, the sex differences found in the present study were of relatively small magnitude (2.8–3.7% of explained variance for school grades, 1.8–1.9% for reaction time). Thus, while on average girls may be achieving higher grades, the difference on the report card is likely to be not that pronounced. Nevertheless, better school performance in this age range among girls compared to boys has been frequently found in other studies [44, 45]. Potential underlying factors favouring girls are differences in motivation, effort, approaches towards schoolwork and learning styles, parental expectations and encouragement, stereotype threat, activity level, and temperament [46]. However, researchers have also emphasized the subjective side of school grades, as they depend on teachers’ perceptions and evaluations, which may lead to sex-biased treatment and/or self-fulfilling prophecies [47, 48]. While the majority of these explanations are based on samples from Europe and North-America, other factors might account for the observed sex differences in the present South African sample. For example, stunted children are more likely to miss school due to their increased infectious diseases risk [49], and to show lower academic achievements [50]. Our data further showed that in boys, children with higher SES presented with better school performances than their peers, which is in line with prior research [51]. This is plausible as higher SES might be associated with exposure to a better and more challenging learning environment at home, higher interest of parents in the education of their offspring, and more favourable educational styles [52]. By contrast, the association between academic achievement and SES was less strong among girls, maybe because in young children, girls are more intrinsically motivated at school than boys due to their perception of what are appropriate and important activities for their gender [53].

Contrary to previous studies [23, 54, 55], we did not find evidence in the multivariate analyses that STH infections are associated with academic achievement (although a tendency was observed among girls). Researchers have suggested that reduced well-being, higher fatigue, increased levels of pro-inflammatory cytokines or abdominal discomfort may explain how STH infections impact on academic achievement and cognitive function. The general low prevalence of STH infections in the present sample might have been a reason why such a relationship was not found in our study.

With regard to physical fitness, our study indicates that children with higher CRF seemed to perform better at school, which is in line with previous studies carried out in higher- and lower-income countries [56]. However, our study results also suggest that the association seems to be more pronounced in girls than in boys. This finding contrasts with previous research where physical fitness parameters were similarly associated with academic performance in boys and girls [20, 57]. However, the state of research is not entirely consistent, with some studies showing stronger associations in boys [58], whereas others found stronger relationships in girls [59]. Although speculative, it is possible that indirect pathways (e.g. mediated via self-esteem, self-discipline) may be responsible for the fact that cardiorespiratory was more strongly associated with academic performance among girls [60]. However, such indirect pathways need be tested more thoroughly in future studies. It is also possible that other (non-assessed) factors that are particularly relevant for children living in marginalized neighbourhoods are responsible for the observed results pattern. Interestingly, in the present study, no significant association was found between children’s grip strength and academic achievement although previous studies have reported significant associations between children’s muscular strength and their academic performances [56].

The fact that academic achievement is influenced by a multitude of health-related and environmental factors and not solely based on cognitive performance is reflected in the present Flanker task results, with boys performing better than girls. This was not due to a speed-accuracy trade-off, since boys had faster reaction times on both congruent and incongruent trials although accuracy did not differ between sexes. Overall, children’s sex explained less variance in inhibitory control than in academic achievement. It has been reported previously that sex differences are smaller if comparisons are based on more objective and standardized achievement tests [44]. Our findings are also in line with studies showing that boys score better in some (but not all) inhibitory control tasks [61]. Boys seem to excel in tasks which involve inhibitory control of attention (such as the Flanker task), whereas they perform similarly on tasks tapping behavioural inhibition [62]. Nevertheless, we found significant correlations between higher school grades and better cognitive performance in the computerized Flanker task independent of children’s sex. This accords well with the international literature showing that higher inhibitory control, which may reflect a more favourable executive function profile [63], translates into children’s future academic success.

As expected, higher age was associated with better performance in all Flanker task outcomes, independent of children’s sex. This common pattern reflects the functional plasticity of the developing brain [64] and highlights the substantial cognitive development that takes place during this sensitive period of life [65].

A further observation was the association of higher muscle strength with better cognitive performance in boys, whereas higher CRF was linked to better cognitive performance in girls. We are not aware of any comparable findings from the existing literature; however, previous studies have shown that children with better CRF and higher muscular strength perform better on tasks tapping inhibitory control and other aspects of executive function [66, 67]. However, few studies have found a relationship between CRF and reaction times [66, 68], whereas associations for response accuracy were more frequently observed [69‐71]. Multiple physiological and psychological mechanisms have been described to explain how physical fitness contributes to better performance on cognitive tasks demanding information processing and/or inhibitory control. Building on the strength model of self-control, physical exercise targeting improvements in physical fitness demands and trains self-control resources, which in turn can be transferred into the cognitive domain by facilitating self-regulation [72]. In this model, effort is a common resource of both executive function and self-regulation that alters or maintains behaviour under specific situations. Consequently, higher self-regulation abilities with higher physical fitness may translate from one domain to the other. In addition to psychological pathways, alterations of specific cognitive processes have been suggested to account for the association of physical fitness with information processing and/or inhibitory control. A review of neurophysiological evidence from studies employing electroencephalographic measurements has found that high CRF is linked with more effective allocation of attentional resources towards the cognitive task [73]. Moreover, improved monitoring of the stimulus conflict has been found to mediate the association between this aspect of fitness and inhibitory control of attention [74]. In addition, limited evidence also suggests structural brain changes in regions associated with executive function in response to regular physical activity [75], so that the association between high physical fitness (as a result of regular engagement in physical activities) and high cognitive performance might partly be explained by such changes. Nevertheless, as highlighted by Howard et al. [76], researchers should also consider possible negative effects of physical activity on executive functions particularly among children who have little excess energy.

The significant relationship between grip strength and boys’ performances in the Flanker task is interesting. There is growing evidence in general population studies that handgrip strength is a predictor of all-cause and cardiovascular mortality [77] and mental health [78]. Studies also have shown that grip strength is closely associated with cognitive function in older people [79, 80] or patients with psychiatric disorders [81]. By contrast, research on grip strength as a health resource among children is in an early stage, and so far little evidence exists that grip strength is suitable as a predictor of children’s health [82]. To the best of our knowledge, only one study has shown that grip strength is positively associated with children’s selective attention [83], and no studies exist on the relationship between grip strength and performance in the Flanker task.

The present study contains strengths and limitations. Particular strengths were that we took into account the nested nature of the data because academic achievement can strongly vary as a function of school and class. Moreover, the findings are based on a relatively large sample, in which boys/girls and children from grades 1–4 are similarly represented. Furthermore, statistical software was used that is able to handle missing data in a meaningful way and a robust estimator was applied to deal with non-normally distributed data. Because sex differences existed on multiple variables, we performed separate analyses for boys and girls, and the multivariate analyses enabled us to simultaneously control for multiple explanatory factors. Importantly, independent variables were only weakly to moderately correlated with each other, and all variance inflation factors were low (VIF ≤ 1.62). Accordingly, multicollinearity was not an issue in the present study. Limitations are that the findings are based on cross-sectional data which precludes conclusions about cause and effect. Moreover, school performances depend on a multitude of factors, which were not all assessed in the present study (e.g., parental education/literacy, parental support, etc.). It should also be noted that school grades are not a fully objective measure; rather, they are dependent on perceptions and expectations of the teachers or performances of class-mates [84, 85]. While standardised measures of academic achievement are not available for this age group, in the present study, we were at least able to include an objective indicator of cognitive function that is independent from the above influences. Fourth, implementing the Flanker task with young children in marginalized settings poses some special challenges. Our impression was that some of the children (particularly the younger ones) have seen a laptop for the very first time; thus, we had to provide extensive instructions to ensure that all children have well understood the task. Moreover, all children had to perform two practice trials before the assessment started. Despite these measures, not all children reached sufficient accuracy levels. Last, we also acknowledge that generalizability of our findings is limited by the fact that all children were recruited from schools located in marginalized neighbourhoods in one peri-urban setting. We also need to be cautious with generalizing the findings to younger or older school-aged children, as the effects of CRF and grip strength might be distinct.