The African continent as a whole is known to be the only one in the world in which the prevalence of malnutrition continues to be high, with little evidence of any meaningful or consistent improvement in nutritional status1. Projected estimates for 2015 have indicated that 33% of the world's underweight children will live in the African continentReference de Onis and Blossner2. In addition, there is ample evidence that nutrient deficiencies as well as protein–energy malnutrition are important in shaping growth and health status in the developing worldReference Allen3.
As a result of the high prevalence of undernutritionReference Labadarios and van Middelkoop4 and micronutrient deficiencies in South AfricaReference Department of Health5 among emerging populations in the nutritional transition, primary and secondary nutritional interventions are essentialReference Labadarios, Steyn, Mgijima and Daldla6 and scientifically conducted dietary surveys, especially on children, should be carried out regularly, particularly with the changing society, altered food habits, nutritional problems and demands of professional practices in this countryReference MacKeown, Cleaton-Jones and Perdrau7. In the urban context, Africans have recently experienced numerous changes in environmental, socio-economic and dietary factors, and in other respects, thereby undergoing considerable Westernisation of lifestyle compared with the pastReference Walker and Walker8.
However, in developing countries, such as those in Africa, there is a paucity of information available on change in dietary intake over time. The data available have been mainly cross-sectional studies constraining the ability to examine developmental trends over time. This limitation is particularly concerning when we examine the diets of children, since food and nutrient intake may change with physical, social and psychological maturationReference Lytle, Himes, Feldman, Zive, Dwyer and Hoelscher9.
Dietary habits and preferences are believed to form in childhood and become habituated over timeReference Birch10–Reference Kelder, Perry, Klepp and Lytle12. With the global economic development and urbanisation, there has been a great change in the nutritional status of adolescents worldwide during the past two decadesReference Samuelson13. Some improvements in nutrient intake among youth do occur over time, but, by and large, diets tend to become less nutrient-dense as children move into early adolescenceReference Lytle, Himes, Feldman, Zive, Dwyer and Hoelscher9. However, little attention has been paid to adolescent nutrition in developing countriesReference Shi, Lien, Kumar, Dalen and Holmboe-Ottesen14, but nutrition among youth has become of increasing interest as evidence accumulates about the role of diet during this time and the risk of subsequent adult diseases15.
In South Africa in particular, longitudinal nutrition information depicting dietary change over time among pre-adolescent children is completely lacking, with only one study reporting the energy, macro- and micronutrient intakes of a true longitudinal group of children from the Birth-to-Twenty (Bt20) study (previously referred to as the Birth-to-Ten (BTT) study from 1990 to 2000 and the Bt20 from 2001 onwards)Reference MacKeown, Cleaton-Jones and Norris16.
The Bt20 study commenced in 1990 and is a longitudinal observational study, planned to continue to the year 2010, designed to follow factors associated with the survival and health of urban children from all the South African communities living in the Johannesburg/Soweto area of Gauteng Province. This is the largest and longest running study on children's development in Africa and is also the first and only longitudinal study on the nutrient intakes of South African children, and has thus provided unique information. In addition, the period of this study (1990–2003) coincided with the dramatic socio-political changes taking place in South Africa and provided an ideal opportunity to investigate dietary change in urban South African children over this time.
This study reports on the energy, macro- and micronutrient intakes of a true longitudinal group of 143 urban black South African adolescents in 2000 and 2003, at 10 and 13 years old, respectively, participating in the Bt20 study.
Prior to the study, ethical approval was obtained from the University of the Witwatersrand Committee (medical) for Research on Human Subjects. Details of the study design have been reported elsewhereReference Fonn, de Beer, Kgamphe, McIntyre, Cameron and Padayachee17–Reference MacKeown, Cleaton-Jones, Edwards and Turgeon-O'Brien20, so only a brief outline is provided.
The sample size of each community was originally based on a representative proportional sample of the total South African population according to the 1994 population censusReference Steenkamp21. Hence there was a large difference in numbers between the communities for this study. The black community predominates in South Africa, now comprising 79% of the total population according to the 2001 census22. This fact, together with available numbers of children, resulted in adequate numbers for the longitudinal investigation only from the black community. The number of children in the other groups had dwindled to too few. Nutrition interceptions of the Bt20 study took place in 1995, 1997, 1999, 2000 and 2003 when the children were 5, 7. 9, 10 and 13 years old, respectively. A total of 1096 urban black children provided nutrition information for 1995, 550 for 1997, 304 for 1999, 365 for 2000, and 143 for 2003. A total of 143 urban black children made up the true longitudinal cohort of the Bt20 study having nutrition information at all five interceptions, but only the nutrient intakes of these 143 children at 10 and 13 years old as they enter puberty will be reported. These children's nutrient intakes at the other interceptions (1995–2000) have been reported elsewhereReference MacKeown, Cleaton-Jones and Norris16.
Dietary intake was assessed using the same semi-quantitative food-frequency questionnaire at all interceptions, an appropriate method for large-scale studies such as the Bt20 study, in culturally diverse populationsReference Hankin and Wilkens23. A quantitative food-frequency questionnaire proved a reproducible and valid instrument for assessing the dietary intakes of adults and children older than 15 years among the black population in the North West Province, South AfricaReference MacIntyre, Venter and Vorster24, Reference MacIntyre, Venter and Vorster25, and is the present approach used in many South African nutritional studies. Parents or guardians or the children themselves were asked by trained multilingual interviewers to indicate how frequently listed food items were consumed according to the number of days per week. The foods listed in the questionnaire were based on many years of experience in diet assessment among South African children; the questionnaire was designed to obtain a picture of the diet as a whole rather than to rank or categorise selected nutrients. The assistants were initially trained in interviewing and coding, and were monitored throughout the study period, and these same assistants (interviewers and coders) conducted the fieldwork and coded the data for all interceptions. Food items were coded onto computer coding forms using the South African Medical Research Council (MRC) food composition tables and codesReference Langenhoven, Kruger, Gouws and Faber26. For certain food items listed, the amounts consumed were recorded in household measures. For other items, standard portion sizes were used based on the use of the National Research Programme for Nutritional Intervention (NRPNI) food quantities manualReference Langenhoven, Conradie, Wolmarans and Faber27. The daily amounts consumed were calculated from the intake frequency, and our nutrition programme was applied to determine the mean daily nutrient intake for each individual.
The data were analysed using SAS28. The nutrient intake for each individual at each interception was merged to create one large data set from which individual children's results could be extracted. A specific computer program written by the University of the Witwatersrand Computer and Networking Centre was used to merge the data by child case number to ensure identification of the true longitudinal group. A child seen at all interceptions would keep the same case number with the year ‘95’, ‘97’, ‘99’, ‘00’ or ‘03’ as the prefix. Descriptive statistics were used to determine the daily mean intakes of energy, macro- and micronutrients of the children at each interception, and the 95% confidence intervals of these nutrients were calculated. The mean intakes were compared with the 1989 recommended dietary allowances (RDAs)29, and the percentage of children who fell below the RDA was calculated. The 1989 RDAs were used for consistency as they had been used as the standard in all the previous interceptions. The percentage change in mean daily nutrient intake between interceptions and the 95% confidence intervals of the percentage change were also calculated.
The mean intake of energy, all six macronutrients and most micronutrients (17/19) increased from 2000 to 2003, with protein and fibre among the macronutrients increasing only slightly. Of the 19 micronutrients investigated, eight (namely calcium, iron, zinc, vitamin A, riboflavin, nicotinic acid, pantothenic acid and biotin) fell below the RDA at both interceptions, while seven (namely vitamin B12, magnesium, potassium, thiamin, folic acid, vitamin E and manganese) met the RDA in 2000 and 2003. Vitamin B6 and ascorbic acid fell below the RDA in 2000 only, and phosphorus and copper fell below the RDA in 2003 only (Table 1).
RDA – recommended dietary allowance; RE – retinol equivalents.
The percentage of energy from fat increased from 25% in 2000 to 29% in 2003, but the percentage contribution from carbohydrate (63% in 2000; 60% in 2003) and protein (12% in 2000; 11% in 2003) decreased between these interceptions (data not shown).
This, however, concealed the fact that a large percentage of the children fell below the recommended intakes even though the daily amounts appeared adequate (Table 2). The percentage of children falling below the RDA for energy decreased from 73% in 2000 to 59% in 2003, but there was an increase in the percentage of children falling below the RDA for 12 of the 19 micronutrients, with the percentage for phosphorus and magnesium increasing by >30% and that for ascorbic acid decreasing by >30% over this period (Table 2). More than 70% of the children fell below the RDA for calcium, iron, zinc, copper, vitamin A, riboflavin, vitamin B6, pantothenic acid and biotin at both interceptions (Table 2).
RDA – recommended dietary allowance.
Table 3 shows these data as quartiles according to the percentage of children with intakes below the RDA. The number of nutrients decreased within the lower and upper quartiles from 2000 to 2003, but increased within the middle two quartiles. In 2000, < 25% of the children fell below the RDA for four nutrients, namely protein, vitamin B12, magnesium and folic acid, and >75% fell below for 10 nutrients. In 2003, < 25% of the children fell below the RDA for folic acid only, but >75% fell below the RDA for seven nutrients, namely calcium, zinc, copper, vitamin A, vitamin B6, pantothenic acid and biotin. More than 75% of the children fell below the RDA for these seven micronutrients in both 2000 and 2003 (Table 3).
RDA – recommended dietary allowance.
Figure 1 shows the distribution of all nutrients (limits, quartiles and medians) according to the percentage of children falling below the RDA. There was a narrowing of the distribution of the percentage of children falling below the RDA from 2000 to 2003, with the median percentage decreasing slightly between these interceptions; 75% in 2000 to 65% in 2003.
In addition to reporting the mean daily intake of nutrients compared with the RDA and the percentage of individuals who fell below the RDA, the percentage change in mean daily intake between interceptions was also investigated. Table 4 shows the percentage change in mean daily intake and 95% confidence intervals of these percentages of nutrients between the 2000 and 2003 interceptions, and Fig. 2 shows the distribution of all these nutrients (limits, quartiles and medians) as a percentage of change in mean daily intake between interceptions. Most nutrients showed a positive percentage change in intake between 2000 and 2003. The overall percentage change from 1995 to 2003 for these same 143 children was also positive in comparison with the percentage change from 1995 to 2000, prior to adolescence, which was definitely negative (Fig. 2).
Most countries are now affected by demographic transition and changing epidemiology of disease; the nutrition of children is increasingly recognised as crucial for present and future health as there is increasing evidence that childhood nutrition also influences adult health. A key factor is the recognition that nutritional interventions at different stages of life are necessary if childhood nutrition is to improveReference Tomkins30. South Africa has recognised this, and the present study reports unique descriptive information on how the dietary intake of a true longitudinal group of urban black South African children has changed from 2000 to 2003 through two interceptions, but the particular strength of the study is that it is the first, and only, South African nutritional study to describe dietary changes among a true longitudinal cohort. The nutrient intake of this same longitudinal group of urban black South African children did not show an improvement in nutrient intakes from 1995 to 2000Reference MacKeown, Cleaton-Jones and Norris16 before adolescence, but the 2003 interception did show an improvement in mean daily intakes of both macro- and micronutrients from 2000 to 2003 as these children entered adolescence. However, the 2003 interception still showed a large percentage of children falling below the RDA for many nutrients, the percentage increasing from 2000 to 2003 for most nutrients.
One must always keep in mind that dietary intake is only an estimate as there is no absolute ‘gold’ standard, and that dietary intake cannot be assessed without error and probably never will be. However, with the underlying principle of the food-frequency questionnaire approach being that long-term diet is the conceptually important issue rather than intake on a few specific days, its ease of application and cost effectiveness in large-scale studies, such as the Bt20, made this, the semi-quantitative food-frequency questionnaire, the preferred method of choiceReference Kono, Tokui, Yoshimura, Uchioka and Takewaka31.
In culturally diverse populations, such as in South Africa, however, it may be impossible to develop a food list that is reasonably short while, at the same time, includes all of the essential core foods eaten by a study groupReference Sempos32. Foods important to an individual's diet may be omitted if they are not mentioned in the questionnaireReference Guthrie33. The food list in the Bt20 food-frequency questionnaire was based on many years of experience in dietary assessment among South African children, and analysis was based on group, not individual, intake. It is therefore reasonably certain that the questionnaire covered most of the foods consumed by South African children. The fact that group intake was analysed reduced both inter- and intra- individual variation in nutrient intake to some extent. As time passes though, core foods for the cohort may change as may nutrient composition of the foods already present in the questionnaire. It is thus possible that the same questionnaire used repeatedly in this longitudinal cohort may have become obsolete and needed revisionReference Sempos32. While this is a potential weakness in the methodology of the current study, it is also a strength because it did enable consistency in assessment throughout the study period.
Another possible limitation of the study is the predetermined portion size used in the questionnaire for many of the foods. GuthrieReference Guthrie33 found that young adults were not able to describe their portion sizes accurately. In addition, there is substantial within-person variation in portion sizeReference Hunter, Sampson, Stampfer, Colditz, Rosner and Willett34, which varies less than frequencies in most foodsReference Sanet, Humble and Skipper35, Reference Pickle and Hartman36. With most of the variation in a population food intake explained by frequency of use, individual portion size data have been seen as relatively unimportantReference Willett37.
When children and adolescents are the target population in dietary surveys, many different respondent and observer considerations surfaceReference Livingstone and Robson38. The cognitive abilities required to self-report food intake include an adequately developed concept of time, a good memory and attention span, and a knowledge of the names of foods. From the age of 8 years there is a rapid increase in the ability of children to self-report food intakeReference Livingstone and Robson38 and, although it has been found that parents can be reliable reporters of their children's food intake in the home environment, the suitability of parents to be the only informants of their children's intake out of the home is inevitably limitedReference Eck, Klesges and Hanson39. Children aged 10 years and older eat much of their food away from home, therefore they are probably the most accurate reporters of their intake. Undoubtedly, blending the method with respondent capabilities is of paramount importance, but to date the limits of accuracy of the various survey instruments when applied to children and adolescents have not been definedReference Livingstone and Robson38. The food-frequency questionnaire used in the present study was simple and it was almost certain that the children of 10 years and older were able to report how frequently they consumed the listed foods. However, in most cases, parents or guardians were present.
With regard to the population sample, it is a shame that only 143 subjects could be followed longitudinally, but this reflects the difficulty of longitudinal studies. Families moving away from the area or no longer wishing to participate are just two of the many reasons for subject loss. It must also be kept in mind that the original sample size of each community was based on a representative proportional sample of the total South African population according to the 1994 population censusReference Steenkamp21. Hence there was a large difference in numbers between the communities for this study. The black community predominates in South Africa, now comprising 79% of the total population according to the 2001 census22. This fact, together with available numbers of children, resulted in adequate numbers for the longitudinal investigation only from the black community. The number of children in the other groups had dwindled to too few. In addition, there is no indication of whether these 143 children were similar to those lost to follow-up.
Comparisons with other South African studies
There is a general agreement that diet does change over timeReference British Nutrition Foundation40, but in South Africa no longitudinal nutritional studies on children have been conducted to assess dietary change and, unfortunately, the available dietary studies on the nutritional status of pre- and primary school children are mostly of a small scale and localisedReference Labadarios, Steyn, Maunder, MacIntyre, Geriche and Swart41 and little attention has been paid to adolescent nutritionReference Shi, Lien, Kumar, Dalen and Holmboe-Ottesen14. Therefore, comparisons can only be made with cross-sectional studies, which have provided extremely valuable information on the nutritional status of population groups at a particular point in time but have been unable to determine dietary change over time among the same individuals. These cross-sectional studies have been conducted in rural and urban areas; among infants, pre-school and primary school children, adults and the elderlyReference Labadarios and van Middelkoop4Reference MacKeown, Cleaton-Jones and Perdrau7Reference MacKeown, Cleaton-Jones, Edwards and Turgeon-O'Brien20Reference Labadarios, Steyn, Maunder, MacIntyre, Geriche and Swart41–Reference MacIntyre, Kruger, Venter and Vorster53, with most reporting low energy intakes and inadequate micronutrient nutrient intakes.
The South African Vitamin A Consultative Group (SAVACG)Reference Labadarios and van Middelkoop4 found that one in three children had a marginal vitamin A status ( < 20 g dl− 1). The prevalence of vitamin A deficiency was highest in the rural areas. Additionally, one in five children were found to be anaemic. Faberet al. Reference Faber, Jogessar and Benade50 also found that the poor quality of the diet of rural pre-school children was reflected in a poor vitamin A and iron status. The present study was conducted on an older age group than these above-mentioned studies but found 98% of the children at 10 years old and 78% at 13 years old with intakes below the RDA for vitamin A and 71 and 74% below the RDA for iron at 10 and 13 years old, respectively. These percentages were much higher than those reported in the other two studies.
The National Food Consumption Survey (NFCS) of children aged 1–9 years in South AfricaReference Labadarios, Steyn, Maunder, MacIntyre, Geriche and Swart41 showed that the majority of children consumed a diet deficient in energy and of insufficient nutrient density to meet their micronutrient recommendations. For South African children as a whole, the dietary intake of energy, calcium, iron, zinc, selenium, vitamin A, vitamin D, ascorbic acid, vitamin E, riboflavin, nicotinic acid and vitamin B6 was below 67% of the RDA. A large percentage of children (25–82%) of all the age groups had intakes of < 50% of the RDA for vitamin E, ascorbic acid, riboflavin, nicotinic acid, vitamin B6 and folic acid. The mean calcium intake was less than half of that recommended in almost 95% of the children. At the national level, 81–94% of the children had an intake much less than half of the recommended calcium intake. The mean intake of iron and zinc was consistently low in all age groups. At the national level, 41–83 and 52–69% of the children had intakes of iron and zinc, respectively, of less than half of the recommended intake. The present study also found the intakes of calcium, iron, zinc, vitamin A, riboflavin and nicotinic acid to be below the recommended intakes at both interceptions, with >70% of the children falling below the RDA for these nutrients. However, pantothenic acid and biotin were additional micronutrients that fell below the RDA, while folic acid met the RDA at both interceptions for the Bt20 children.
Possibly the only South African study that has tracked nutrient intake over time among the same age group, community and living in the same area is the comparison of energy, macro- and micronutrient intakes of 5-year-old urban black children in 1984 and 1995 living in the Johannesburg/Soweto areaReference MacKeown, Cleaton-Jones, Edwards and Turgeon-O'Brien20. It was found that the intake of energy and most macronutrients was higher in 1995 than 1984, except for vitamin A, ascorbic acid, copper and iron, and only biotin fell below 67% of the RDA. However, a large percentage of children fell below the RDA for many of the nutrients even though the intakes appeared adequate. It was also reported that the children consumed a typical prudent diet in 1984, but by 1995 the diet was more Westernised. Comparing rural and urban women in the Transition and Health during Urbanisation of South Africans (THUSA) studyReference MacIntyre, Kruger, Venter and Vorster53, a definite shift in nutrient intake was observed: the percentage of energy from carbohydrates decreased, but the percentage from fat and protein increased. The present study showed that urban black children living in the Johannesburg/Soweto area consumed a characteristic prudent diet low in fat ( ± 30%) and high in carbohydrate ( ± 60%), but within an urban environment there had been an increase in the percentage of energy from fat, and a decrease in the percentage from carbohydrate and protein from 2000 to 2003. Thus there is evidence not only of a nutrition transition from rural to urban but also within an urban environment itself over time.
The only true longitudinal comparative study that tracks nutrient intake over time among the same urban black South African children is an earlier investigation of these same Bt20 children reported on in this study prior to adolescence from 1995 to 2000 (5–10 years)Reference MacKeown, Cleaton-Jones and Norris16. The nutrient intake of these children appeared to deteriorate over four interceptions from 1995 to 2000, with the lowest intake for most nutrients recorded in 2000. Calcium, iron, zinc and biotin were the most common nutrients to fall below the RDA at most of the interceptions, with also the largest percentage of children falling below the RDA for these nutrients. Most nutrients showed a negative percentage change in intake from 1995 to 2000. At the 2000 and 2003 interceptions, calcium, iron, zinc and biotin remained the common nutrients that did not meet the RDA, with >70% of the children falling below the RDA for these same nutrients. However, in contrast to the overall percentage change in intake for most nutrients from 1995 to 2000 being negative, the percentage change in intake for these children from 2000 to 2003 was positive for most nutrients.
Although the comparative studies mentioned above have not included children of the same age groups as the present study, the findings emphasise the continuous trend in inadequate micronutrient intake among all age groups in South Africa.
The results of the present study should, however, be interpreted with caution as we only report on the nutrient intake. The actual food items consumed, socio-economic factors, anthropometrical and growth data were not included. Inclusion of this information would give a clearer understanding of the nutritional status of this group of children and would show if there had been an improvement or a deterioration in nutrient intake as they entered puberty. This has been planned for a future publication. To emphasise the importance of this future research, it was stated by Lytleet al. Reference Lytle, Himes, Feldman, Zive, Dwyer and Hoelscher9 that more research is needed on the effects of socio-economic status on the nutritional intake of youth, and further research examining differences in food choice behaviour is very important and needed, especially for intervention studies. Since individuals eat food, not nutrients, we need to learn more about how food choices vary over time.
The finding that moderate stunting coexists with overweight and obesity suggests that patterns of under- and overnutrition in South African children are changing and might indicate the early stages of a complex nutritional transitionReference Jinabhai, Taylor and Sullivan54. This necessitates regular clinical and epidemiological monitoring of nutritional status to examine possible future trendsReference Jinabhai, Taylor and Sullivan55, and to state whether a nutrition transition is taking place in South Africa longitudinal monitoring of nutritional status is essential.
BT20 is funded by the Welcome Trust (UK), the Human Science Research Council of South Africa, the Medical Research Council of South Africa, the University of the Witwatersrand, Johannesburg and the Anglo-American Chairman's Fund. For information on Bt20, visit www.wits.ac.za/birthto20