Factors associated with the health and cognition of 6-year-old to 8-year-old children in KwaZulu-Natal, South Africa

Abstract

Objective

To investigate 6-year-old to 8-year-old children’s health, nutritional status and cognitive development in a predominantly rural area of KwaZulu-Natal, South Africa.

Methods

Cohort study of 1383 children investigating the association of demographic variables (area of residence, sex, pre-school education, HIV status, height for age and haemoglobin level) and family variables (socioeconomic status, maternal and paternal level of education), with children’s cognitive performance. The latter was measured using the Grover-Counter Scale of Cognitive Development and subtests of the Kaufman Assessment Battery for Children, second edition (KABC-II). General linear models were used to determine the effect of these predictors.

Results

Area of residence and height-for-age were the statistically significant factors affecting cognitive test scores, regardless of attending pre-school. Paternal level of education was also significantly associated with the cognitive test scores of the children for all three cognitive test results, whereas HIV status, sex and their socioeconomic status were not.

Conclusion

Children with low cognitive scores tended to be stunted (low height-for-age scores), lacked pre-school education and were younger. Area of residence and their parents’ educational level also influenced their cognition.

Introduction

More than 60% of children in sub-Saharan Africa are estimated to be at high risk of failing to achieve their potential in cognitive development, with subsequent negative effects on their educational attainment, work capability and poverty [1]. In this prospective cohort study of children 4–6 years in a resource-poor area of Kwazulu-Natal, South Africa, the Asenze study, we investigated precursors of cognitive competence in children in the years leading up to formal education, which span a transitional period in the children’s lives characterised by ‘key events or processes of change over a life course’ [2].

Factors relevant to a child’s cognitive development in this environment were measured and several were found to be associated with the scores of at least one of the three cognitive tests also administered to the children when they were 4–6 years. These were the area where the children lived (site), sex, whether children received pre-school education, their HIV status, the level of parental education (mother and father), height-for-age and haemoglobin level.

The investigation showed a significant association between the site where they lived, poor nutrition as measured by height-for-age z-scores and cognitive outcomes as measured by the Grover Counter Test and subtests of the KABC-11[3–5], selected as age and culturally appropriate.

The same children were followed up 2 years later when they were attending school and this article reports on the investigation of the association between the factors of significance when the children were 6–8 years old and their cognitive outcomes at this early stage of their schooling.

Methods

The study design was a prospective cohort and undertaken in KwaZulu-Natal, the second most populous of nine provinces in South Africa, which has the largest number of children living in households with per capita income below the poverty line [5]. Although they are adjacent, the five local authority areas, which make up the total study site have different characteristics, extending from peri-urban to deep rural areas and are governed by different local authorities. There are also extreme differences in terrain and proximity to urban areas.

Prior to the start of the study in 2010 a team of four women (all first language Isizulu speakers) received intensive training on the use of the study instruments. The assessment team was led by an assessor with over twenty years of experience in testing children in isiZulu, and the other team members each had several years of experience.

In 2010, 1581 children were enrolled in a door-to-door operation where all the children of 4–6 years were invited to participate, 2 years later 1383 (87.5%) of these children now between the ages of 6–8 years participated in the cognitive and anthropometric testing reported in this paper. The data collected from this study of school-age children included information on their demographic and socio-economic status, their parents’ educational level and information about the child – whether or not the child had attended pre-school, anthropometry, HIV status and cognitive test scores as described below.

Cognitive assessment instruments

In the follow-up phase of the study Atlantis, Grover counter and Hand movement tests were administered by the same four assessors, who had participated in phase 1 of the Asenze study. The Grover Counter Scale (GCS) was developed in South Africa for the assessment of children aged 3–10 years. The administration of the test does not require many verbal instructions or responses and this has the advantage of minimising translation errors in a multilanguage country like South Africa. The construction and standardisation of the scale is described in the manual [3] and significant correlations were found between the GCS scores and scores from the Eye-hand coordination and Performance subscales of the Griffith’s Scale of Mental Development [3].

Subtests from the Kauffman Assessment battery (KABC-II) were also used to assess the children’s cognitive development. The KABC-II was chosen because it has been shown in several countries in Africa and elsewhere to have good reliability because the testing procedure is age-appropriate and relatively culture fair [6]. The Hand Movement subtest was selected from this battery as a memory test on the basis that it was not language-based and would not need adaptation. Both this and the Atlantis subtest had been piloted in isiZulu-speaking communities, and the only adaptation found necessary for the Atlantis subtest was to pronounce the nonsense names in accordance with isiZulu pronunciation rather than that recommended in the manual.

Data collection

The child and caregiver were transported to the study clinic where the child received a clinical examination and participated in the cognitive tests that were undertaken by trained research assistants. The caregiver was also interviewed about the child’s health and development.

Anthropometry

The children’s height and weight were measured. The height was measured using a fixed stadiometer and the weight using a Masskot electronic scale. Z-scores were calculated for the children’s mean height-for-age (measuring stunting), weight-for-age (measuring underweight) and weight-for-height (measuring acute malnutrition) using the WHO anthropometry package [7].

Haemoglobin.

The child’s haemoglobin was measured using a Hemocue [8]. And if Hb measured <11.5 g/dl the child was considered to be anaemic.

HIV prevalence.

All the caregivers were asked whether they and their children had previously tested for HIV and what the results were. Caregivers were offered HIV testing on site for themselves and their children as part of approved study procedures, following the protocol of the KwaZulu-Natal Department of Health.

Variables included in the analyses

This analysis investigated the association of demographic, socio-economic and child variables, on children’s cognitive performance as measured by the Grover counter test and the subtests from Kauffman’s KABC-11 (namely, the Atlantis and Hand Movement tests). General linear models were used to determine the effects of these predictors. Our analysis included three dependent cognitive outcome variables: Atlantis test scores, Grover counter test scores and Hand Movement test scores. The independent variables investigated included site (denoting each of the five geographic areas), sex, education (whether or not the child had attended a pre-school or crèche), child’s HIV status (ChildHIV); positive, negative and unknown, socio-economic status (for SES, an household asset index was developed using quintiles where 1 is the lowest), parents’ educational level (paternal level of education and maternal education level), haemoglobin (ChildHb), height-for-age (Hazscore).

Ethical approval was obtained for all study procedures and interview material from the University of KwaZulu-Natal Biomedical Research Ethics Committee (BF036/07) and the Columbia University Institutional Review Board.

Results

Characteristics of participating children

Due to non-response and dropout of some of the 1581 children who were initially enrolled in the study, anthropometric and cognitive data were collected from 1386 children, giving an approximate 88% follow-up success. The totals for each variable differed as a result of the incompleteness in the required information which were regarded as missing, and excluded and hence did not influenced the result of any analysis. We report in this paper on these 1386 children from the five geographic areas (Sites 1, 2, 3, 4, and 5), 1572 children with five indices of socio-economic status (Lowest 20%, Low middle, Middle, High middle, Top 20%) and 1582 Parents’ education level (None, Primary, attended High School, Completed High School, Tertiary and Unknown). Approximately 50% (n = 698) of the children were male and 65% (n = 883) had received pre-school education, (Table 1). The study recorded 62 HIV positive children, 1278 HIV negative children and 241 children with unknown HIV status. (Table 1).

Table 1.

Descriptive statistics of children’s categorical variables at phase 2

Variable		Frequency	Per cent
Site	1	241	17.4
	2	222	16.0
	3	150	10.8
	4	461	33.3
	5	312	10.8
	Total	1386	100
Sex	Male	698	50.4
	Female	688	49.6
	Total	1386	100
Pre-school Education	None	481	35.3
	Received	883	64.7
	Total	1364	100
ChildHIV	Positive	62	3.9
	Negative	1278	80.8
	Unknown	241	15.2
	Total	1581	100
Socio-economic status	1. Lowest 20%	327	20.8
	2. Low Middle	355	22.6
	3. Middle	250	15.9
	4. High Middle	314	20.0
	5. Top 20%	326	20.7
	Total	1572	100
Maternal Education level	0. None	65	4.1
	1. Grade 1–7 (Primary)	223	14.1
	2. Grade 8–11 (High School)	665	42.0
	3. Grade 12 (Completed)	359	22.7
	4. > Grade 12 (Tertiary)	3	0.2
	5. Unknown	267	16.9
	Total	1582	100
Paternal Education level	0. None	88	5.6
	1. Grade 1–7 (Primary)	170	10.7
	2. Grade 8–11 (High School)	407	25.7
	3. Grade 12 (Matric)	437	27.6
	4. > Grade 12 (College)	2	0.1
	5. Unknown	478	30.2
	Total	1582	100

Descriptive analyses of the cognitive scores

Information for the minimum, maximum, mean and standard deviation for the children’s cognitive scores for the Atlantis, Grover and Hand movement test scores is provided in the table below. (Table 2)

Table 2.

Descriptive statistics of cognitive scores from the study

Cognitive score	n	Minimum	Maximum	Mean	SD
Atlantis test scores	1363	0	98	46.81	16.798
Grover test scores	1368	0	100	44.30	18.614
Hand Movement test scores	1360	0	17	6.73	2.478

Child variables

In this study, there was a wide range in the children’s anthropometry variables with a 14.7% prevalence of stunting and 3% underweight children and 48.3% of the children’s haemoglobin level was below the anaemia cutoff of 11.5 g/dl.

Factors associated with cognitive outcomes in the study

In fitting linear regression models with the cognitive scores (Atlantis, Grover and Hand Movement) as the response variable, there were significant site effects for all the three cognitive tests (Table 3). Naturally, age had a significant effect in all three cognitive tests and children who had attended a pre-school were also found to have significantly better scores in all the cognitive tests. Neither the child’s HIV status nor the level of haemoglobin was significant in any of the cognitive tests. Similarly, sex was also not a significant factor in any of the tests of cognition. The effect of paternal education level was significant in all the cognitive tests. There was a trend in the role of maternal education in the P-value of 0.07 which is significant at a 10% level. The effect of height-for-age was significant in all three of the cognition tests.

Table 3.

Univariate regression for the three cognitive tests (Atlantis, Grover and Hand Movement) for KZN children 6–8 years, F and P-values

Source	Cognitive test
	Atlantis		Grover		Hand Movement
	F	P-value	F	P-value	F	P-value
Corrected model	1195.641	< 0.05	788.814	< 0.05	1191.694	< 0.05
Intercept	1977.355	< 0.05	967.533	< 0.05	1855.229	< 0.05
Site	8.248	< 0.05	7.888	< 0.05	8.211	< 0.05
Sex	0.021	0.886	0.011	0.918	0.033	0.855
Pre-school education	237.942	< 0.05	241.858	< 0.05	236.060	< 0.05
ChildHIV	0.526	0.591	0.529	0.589	0.512	0.599
Socio-economic status	0.510	0.729	0.542	0.705	0.567	0.686
Maternal education	2.140	0.074	2.098	0.079	2.137	0.074
Paternal education	3.794	< 0.05	3.745	< 0.05	3.808	< 0.05
Height-for-age z-score	5.804	< 0.05	5.797	< 0.05	6.427	< 0.05
Age	27254.702	< 0.05	17355.888	< 0.05	27324.846	< 0.05
Haemoglobin level	0.354	0.552	0.242	0.623	0.406	0.524

It is worth noting that there was strong multi-collinearity between the anthropometric scores (weight-for-age, weight-for-height and height-for-age). Hence, this model included only height-for-age.

The results of the multivariate analysis confirmed the previous results for the three cognitive tests for the association of children’s cognitive outcomes with pre-school education, and stunting, low height-for-age (Table 4). However in this model, the site was no longer associated with children’s cognition and maternal level of education appeared to be more relevant than paternal education.

Table 4.

F and P-values for the multivariate test based on the three cognitive scores for Atlantis, Grover and Hand Movement for KZN children 6–8 years

Effect	F	P-value
Intercept	0.229	0.876
Site	0.829	0.621
Gender	0.970	0.406
Pre-school education	3.010	< 0.05
ChildHIV	0.603	0.728
Age (in months)	44.455	< 0.05
Height-for-age z-score	12.689	< 0.05
Haemoglobin level	1.329	0.263
Maternal Education	2.240	< 0.05
Paternal Education	0.970	0.475
Socio-economic status	0.645	0.805

Discussion

The results from our study, have drawn attention to some of the influences on health and cognitive development in a resource-poor environment when the children are in the early grades of school. Two factors reported here which are significant indicators of performance on the cognitive tests are preschool experience and height-for-age. These are therefore the prime targets for interventions designed to improve children’s health and cognitive functioning.

Our study presents evidence that a pre-school education, however minimal, is strongly associated with higher scores on cognitive tests in this environment in the data collected. This implies that pre-school education has a strong effect on the brain development and the educational outcomes of children living in rural communities, as suggested in a series of papers on child development in developing countries [1]. Another paper in the Lancet series discusses strategies to prevent loss of development potential in young children and proposes that the most effective strategies are comprehensive ones involving health and nutrition as well as stimulation. [9]. In South Africa, as in many other countries, child health and early learning programmes are administered by different government departments, making cross-cutting programmes difficult, but not impossible, as the recent implementation of school health programmes has shown [10].

An extensive review of pre-school research findings in the United States and internationally examined the policy implications from single studies and multiple meta-analyses [11]. Barnett concludes that there is immediate benefit for cognitive development across all types of programmes, that the effect size tends to be greatest in previously disadvantaged groups, and that there is substantial evidence of lasting effects through school (less grade retention, fewer special education placements, and more likely to complete high school). The quality of the pre-school programme is a factor in producing benefits. While our study has shown that any preschool educational experience has an effect, interventions are required not only to increase access to programmes, but also to improve the quality of what is on offer. Further, from an economic point of view, provision of pre-school education for children in such disadvantaged areas would reverse the negative effects of the current underdevelopment conditions and improve cognitive capability [12].

Stunting (14.7%) remains a concern in this study and for many South African children and children living in developing countries [13, 14]. Thus good nutrition, which encompasses food security, freedom from hunger, a diverse diet and sufficient essential micronutrients and fatty acids has to be encouraged [15]. The linear model reported above has shown that height- for-age is positively associated with the cognitive scores of children and was statistically significant in the study. Currently the high rate of anaemia reported in adults and women of reproductive age has been described as the most common nutritional problem in the world [16]. Although the children’s haemoglobin level had no detectable effect on their cognitive scores in this study, almost half of the children who participated in this study were anaemic. This result calls attention to the need to improve iron deficiency which is the major cause of anaemia in children and other important micronutrients which cause anaemia as earlier reported in the literature [17, 18]. Nutrition interventions to increase dietary diversity and improve micronutrient intake would offer hope for haemoglobin levels to reach the recommended thresholds. Parasite infections such as soil transmitted infections which were not measured are endemic in the study area and may also contribute to anaemia [5]. These results have identified good nutrition as an important target area to improve educational outcomes. Although school feeding schemes are in place [19, 20], further remedial actions are required which could bring improvement to the health, nutrition and cognition of children, as part of national efforts to ensure the achievement of the full potential and development of children. The school health teams visit schools and need to identify children at risk [21], particularly focusing on the younger children, whose poor nutritional status has not been identified previously.

The effect of site (area of residence), was significant, and impacted on child cognitive development in complex ways [22]. Although, the study sites were adjacent to one another, they were governed by different local authorities; they differed in terms of distance to urban centres and also in topography. Some local authorities were more efficient than others in providing services, particularly health services and crime prevention. Some sites were not suitable for subsistence farming because the land was steep and the soil poor, while others were more level and close to a river so soil was rich in nutrients and water was available for irrigation. The sites closest to urban centres provided better opportunities for employment and consequently somewhat better household income. These microgeographic differences would appear to have influenced child rearing, health and cognitive outcomes: the site where cognitive scores were lowest had very steep terrain and high crime relative to other sites.

Cost-effective intervention programmes need to be flexible in their approach, prioritizing where the need is greatest.

Paternal level of education was significantly associated with the three cognitive scores of the children when they were 6–8 years of age and maternal education in the multivariate model. In a Canadian study an intervention using children’s cognitive stimulation which focused on less educated mothers proved to be effective in achieving better cognitive performance in their children [23], emphasising the important parental role.

The effects of sex, child’s HIV status, socio-economic status and haemoglobin level did not have a statistically significant association with any of the cognitive test scores. Within these ages, the study provided evidence of the benefits of the provision of antiretroviral therapy [24]. It was pleasing to find from this study that the effect of HIV was not a major barrier to child development as HIV positive children have access to free antiretroviral therapy at state clinics.

Parents or caregivers of every child in the study area when aged between 4 and 6 years were invited to participate and 2 years later nearly ninety per cent of the invited children participated and agreed to anthropometry and cognitive assessments. The research assistants measuring the cognitive scores had many years of experience of working in the area and in using the study instruments. However, the caregiver provided the information on pre-school education, maternal and paternal education and socio-economic status, and these variables were not verified independently.

This paper reports on the effects of some of the many facets of poverty on the health and cognitive development of children in the five sites that make up the study area. This poverty and lack of resources experienced by the families is typical of many communities in South Africa and indeed in Africa [25]. Despite the presence of limited child support grants in South Africa which should alleviate the worst of the deprivations, and the roll-out of antiretroviral therapy for these HIV-endemic communities, the constraints on child health and early learning discussed here which impact on development are considerable. They require attention at all levels of government, but particularly the local level where service delivery is actually managed. Remedial action required for children in the 6–8 year age group includes:

• Attention to improved food security through food gardens where feasible;

• Health programmes at family level based on the World Health Organization’s Community Integrated Management of Childhood Illnesses;

• Protection of children from infectious diseases;

• Parenting programmes to increase age appropriate stimulation of children’s cognitive abilities in the home;

• High quality pre-school education.