Food intake, nutritional anthropometry and blood chemical parameters in 3 selected Dutch schoolchildren populations

The major health problems in populations of economically developed countries at the present time are of a chronic nature with, as their main clinical characteristic, the frequently occurring premature coronary heart disease. When food intake data are to be evaluated, it would be incorrect in this situation to devote attention chiefly to the intake of essential nutrients. Instead, investigations into the food intake in these countries should currently be concerned in the first place with observed nutrient properties suspected of causing the development of elevated blood lipid levels in the majority of the individuals under study. This consideration was the most important point at issue during the performance, and presentation in this thesis, of a regional comparative study into the food intake and nutritional health status of 3 selected Dutch schoolchildren populations.

The introduction starts with a short description of some recently conducted in vitro experiments investigating the role of lipoproteins in atherogenesis. Next, reviews are presented which cover fields of research related to the evaluation of associations between certain features of the life style and the incidence or mortality of atherosclerotic complications. One of these reviews was concerned with the socio-economic status, a very widely used index of life style, and its association with CHD death and CHD risk factors. It was concluded that the observed differences in CHD death and the relationship demonstrated to exist between socio-economic status, CHD death and CHD risk factors, strongly suggest environmental influences in the etiology of atherosclerosis. Diet, of course, is also one of the features of life style. The habitual diet of populations and, particularly, of individuals is difficult to characterize. The various dietary survey methods were the subject of an extensive review in an effort to assess the value of results to be expected from applying them in the collection of food intake data. It was concluded that, with due attention to survey period and number of participants, most dietary measurement techniques can be applied accurately enough to characterize the average diet of groups. If it is essential to assess accurately the individual's habitual diet, carefully conducted long lasting surveys will be needed. This conclusion was shown to have severe impact upon the objective of correlating nutrient composition of the individual's habitual diet to observed serum total cholesterol levels. The other argument used to demonstrate that zero correlation is the inevitable consequence of relating diet to serum total cholesterol within a population, has to do with the inaccuracy of characterizing individuals with respect to their serum total cholesterol level on the basis of single determinations. The variation in observed serum total cholesterol level was the subject of a further review. It was concluded that a single determination is insufficient to provide an accurate description of the individual's position in the observed distribution of serum total cholesterol determinations.Results from inter-population studies and from intervention trials were shown to be consistent with the saturated fat - serum cholesterol - CHD hypothesis. The conclusions reached in the foregoing reviews were reconciled with this hypothesis which may be shortly formulated as follows: a confluence of socio-economic and socio-cultural factors in industrialized countries, often indicated briefly with affluency, has led to the fact that the majority of the population now consumes an abundant diet. This diet, excessive in energy in relation to expenditure and rich in animal protein, saturated fat, cholesterol, sugar and salt, leads to a high prevalence of obesity and hyperlipidaemia in the population. Sustained hypercholesterolaemia markedly increases the probability of premature atherosclerotic disease in the population. In addition, a high prevalence of obesity often has the concomitant consequence of hypertension, hypertriglyceridaemia, etc. Thus, diet is related to the CHD epidemic through at least two identifiable changes of etio-pathological events. This justifies the designation of the current habitual diet as the decisive factor in the pathogenesis of the epidemic appearance of premature atherosclerotic disease.Of course, it is not the intention in this hypothesis to ignore the genetic contribution. It was formulated in the last review of the introduction that inheritance clearly primarily determines the question of whether or not a certain physiological status might be obtained, but that the environment moderates the extent to which the genetic make-up can express its potential. Although the simultaneous investigation of parents and children does not permit the disentangling of hereditary and environmental contributions, the practical consequences of observed familial resemblance in coronary risk factors would be the prediction of increased risk from elevated risk factor levels even in childhood and, at the same time, the justification of efforts aimed at reducing this risk. It has also been recognized that since the education towards the modern, affluent life style occurs during childhood, it is justified to include children into public recommendations intended to reduce the generally elevated risk factor levels in industrialized countries through modification of the diet in the whole community.

The objective of the present study was to investigate whether differences in food intake data would lead to differences in nutritional health status, as measured by anthropometry and by blood lipids. Three comparative epidemiological surveys were carried out in the period 1974-1976 in schoolchildren between the ages of 3.5-13.5 from towns in different provinces of The Netherlands. Based on demographic data from the 1971 census the towns Heerenveen, Roermond and Harderwijk were chosen. In each town, schools with predominantly authochthonous children were selected by mutual arrangement with the local schoolphysician.

Anthropometric measures and blood lipids were assessed in the participating schoolchildren and in parents from a sub-sample of these children. Food intake data were obtained from Grades 1-3 primary schoolchildren. Relationships between nutrient intake data and body fatness and between nutrient intake data and serum total cholesterol could be studied in these children. In another sub-sample, parent- child relationships concerning anthropometric data and blood lipids could be analysed.

About 95% of the primary schoolchildren in Heerenveen and Harderwift participated in the survey. The participation rate was about 10% lower in Roermond. The participation rates in the nursery schools ranged between about 75 and 95 %. The participation rate among Heerenveen nursery schoolchildren was relatively low. The contacts with the parents of the nursery schoolchildren were intensified during the Roermond and Harderwift survey, resulting in higher participation rates among nursery schoolchildren in these towns.

Demographic data about the education and occupation of the parents, socioeconomic status of the family and birth-places of parents and children was obtained by a general questionnaire filled in by the parents. The socio-economic status of the families tended to be highest in Roermond, intermediate in Heerenveen and lowest in Harderwift. However, the difference between Roermond and Heerenveen families seemed to be larger than the difference between Heerenveen and Harderwijk families. The difference between Roermond and the other towns may be explained by differences in the method of selection of schools. About 15% of the Heerenveen and Harderwift mothers were fully or part-time employed. In Roermond this percentage was twice as high. In Roermond about 2/3 of the parents was authochthonous compared to about 1/3 in Harderwift. In both Roermond and Harderwift about 2/3 of the examined children had been born and raised in the province of Limburg and the region of the North-west Veluwe respectively. During the Heerenveen survey information about the birth-places of parents and children was not obtained.

The anthropometry included measuring of each child's body weight, standing height, knee widths, mid-arm muscle circumference and skinfold thicknesses at 4 sites - biceps, triceps, subscapular and suprailiac. Results of the measurements were adjusted to a joint level on the basis of comparison experiments, by correcting the mean level of measurements obtained in Heerenveen and Harderwijk children to the level of measurements observed for the regular Roermond investigator. Children from Heerenveen compared favourably to their agepeers from Roermond and Harderwift with respect to the anthropometric measurements. Frank obesity, tentatively defined to be present if estimates of body fatness based on skinfold thickness measurements exceeded 25-27%, was observed in 3.2 % of the Heerenveen children. Compared to these Heerenveen children, children from Roermond were found to be smaller and to have less skeletal and muscle mass, but. considerably more body fat. This was reflected by higher mean skinfold thicknesses and a higher prevalence of frank obesity, amounting to 6.0% of all Roermond children. Children from Harderwijk took the intermediate position with respect to body weight and standing height. Skeletal mass in children from Harderwijk was found to be roughly comparable to that in children from Heerenveen, but Harderwift children were found to have even less muscle mass than children from Roermond. At the mean level, children from Harderwift had the thickest subcutaneous fat layer but the prevalence of frank obesity found, 4.6%, was intermediate between that observed in children from Heerenveen and Roermond. A high dependence of the observed body weights in the children upon the examined anthropometric measures was demonstrated by multiple regression analysis. This finding was interpreted as a substantiation for the above-described comparisons. Comparison of the findings for the thickness of the subcutaneous fat layer in these Dutch children with figures reported for schoolchildren populations from other economically developed countries, reveals that these Dutch children generally exhibited lower skinfold thicknesses than their age-peers from the USA, Canada, England and West- Germany.

The determinations in venous blood, obtained by a Vacutainer system while participants were in the fasting state, comprised the assessment of the concentrations of blood haemoglobin, serum total cholesterol and serum triglycerides. The determination of the serum HDL-cholesterol concentration was added to these during the execution of the Roermond and Harderwift projects. Results of serum total cholesterol and triglycerides determinations from the 3 project towns were adjusted to the level of measurement observed in a subsample of the collected sera by the Johns Hopkins University LRC Laboratory, which performed the determinations according to regulations of the international WHO Lipid Standardization Program under surveillance of the CDC Standardization Laboratory at Atlanta, Georgia. Results of the haemoglobin determinations in both boys and girls from all three towns were consistent with the well-known increase in haemoglobin concentration with age in schoolchildren. Girls were found to have a somewhat higher haemoglobin concentration than boys. The prevalence of clinical plus subclinical anaemia (haemoglobin < 12.0 g/100 ml) was highest in children from Roermond and lowest in children from Harderwift. Only a very small number of children had clinical anaemia (Haemoglobin < 11.0 g/100 ml) and it has consequently been concluded that the results of blood haemoglobin determinations have confirmed the assumption that the participating children belonged to the normal, healthy schoolchildren population from The Netherlands.Observed mean serum total cholesterol concentrations in boys and girls from the 3 towns varied between 160 and 190 mg/100 ml. Consistently lower values were observed in children from Heerenveen, compared to those from Roermond and Harderwift. Only in the Heerenveen children was the mean serum total cholesterol concentration positively related to age-categories. Girls had approximately 5 mg/100 ml higher mean concentrations than boys in all 3 towns. The overall prevalence of borderline plus frank cholesterolaemia (cholesterol>, 200 ing/ 100 ml) was found to be 19.5 % in boys and 24.8 % in girls., 200 ing/ 100 ml) was found to be 19.5 % in boys and 24.8 % in girls.Frank hypercholesterolaemia (cholesterol>220 mg/100 ml) was present in 7.2% of all boys and in 9.8% of all girls. By both criteria, children from Heerenveen showed the lowest, and children from Harderwift the highest prevalence. However, the differences observed in prevalence of hypercholesterolaemia did not consistently reach statistical significance. Comparisons of the findings in these Dutch children with reported serum total cholesterol concentrations in children from other Western countries generally revealed that the levels in these Dutch children were among the highest reported. In addition, almost identical figures have been reported in adolescents from New Zealand and the USA.220 mg/100 ml) was present in 7.2% of all boys and in 9.8% of all girls. By both criteria, children from Heerenveen showed the lowest, and children from Harderwift the highest prevalence. However, the differences observed in prevalence of hypercholesterolaemia did not consistently reach statistical significance. Comparisons of the findings in these Dutch children with reported serum total cholesterol concentrations in children from other Western countries generally revealed that the levels in these Dutch children were among the highest reported. In addition, almost identical figures have been reported in adolescents from New Zealand and the USA.The observed mean triglycerides concentrations in boys and girls from the 3 towns varied between 55 and 75 mg/100 ml. Mean triglycerides concentrations in children from Harderwijk were about 10% lower than those in Roermond children. Heerenveen children were found to have only slightly lower mean concentrations than children from Roermond. Girls had approximately 4 mg/I 00 ml. higher mean serum triglycerides concentrations than boys in all 3 towns. No consistent relationship with age-category was observed. The prevalence of frank hypertriglyceridaemia, tentatively evaluated by the cut-off point of 100 mg/ 100 ml. was found to be 7.5 % in all boys and 7.1 % in all girls. Comparison of serum triglycerides concentrations with values reported in schoolchildren populations from other countries seems to indicate that the values found in these Dutch children were relatively low.The mean serum HDL-cholesterol concentrations in children from Roermond and Harderwijk were found to vary between 50 and 60 mg/100 ml. Only minor differences were observed in mean serum HDL-cholesterol concentrations between children from Roermond and Harderwift. In these schoolchildren about 1/3 of the serum total cholesterol was contained in the HDL. Although this percentage was highly variable between individual children, a positive association was found with age-categories in the 4 project-sex combinations.

The examination of parents from a sub-sample of the participating children offered the possibility of comparing the level of measurements in fathers and mothers, both separately and by the combination of spouse pairs, to the levels observed in children. The same analysis was done for the measurements obtained among married couples and among siblings. Strong concordances in measurement levels for standing height, body fatness and blood chemical parameters were observed between married couples and their children. The same concordance was noted for all anthropometric and blood chemical parameters in the analysis of resemblance between siblings. If levels observed in parents were compared to those of their children for fathers and mothers separately and, moreover, for daughters and sons separately, the degree of resemblance became substantially smaller. The concordance of measurement levels between spouses seemed to be virtually absent. It was concluded from these observations that the position which the child takes in the distribution of examined characteristics must be on the polygenic control, and that the degree and nature of resemblance was in agreement with the concept of determination by both parents. In addition, it has been argued that the same environment shared by the first degree relatives must have contributed to the degree in which inheritance could have expressed its potential. It has been recognized that the repeatedly demonstrated familial resemblance in disease and the risk factors justified the inference of risk from elevated levels even in childhood and, at the same time, the justification for attempts to alter these risks in childhood.

In all 3 towns statistically significant inverse associations were observed between the prevalence of obesity in the children and the educational level of the mother. Consistent inverse relationships between hypercholesterolaemia in children and the educational level of the father or mother and the socioeconomic status of the family were observed in Harderwift only. These data seem to indicate that nowadays elevated biological parameters are more prevalent in children from the lower social classes compared to children from the higher social classes. However, these associations could only be studied by making broad categories. More detailed studies are needed in order to corroborate these findings.

Information about the food intake of the children was obtained by a two-day record filled in by the mother. This record was checked for completeness after the recording period, during an interview between the mother and one of the investigators. The percentage of children in which reliable dietary records were obtained ranged from 77 % in Heerenveen to 91 % in Harderwift. Co-operation was refused by the mothers of 8% of the Heerenveen children, 9% of the Roermond children and 3 % of the Harderwift children. It was concluded that the results of the dietary surveys were not profoundly influenced by the refusal rate.

As suggested by previous studies, differences in food intake patterns between the children from the different towns could be observed. The percentage of children who ate cheese, yoghurt, sweet fruit squashes or breakfast cake (including 'Friese koek') was higher in Heerenveen compared to the other towns. The percentage of children who ate apple sauce or mean-PUFA low-fat margarines was lower in Heerenveen compared to the other towns. The percentage of children who ate mean-fat pork or French fried potatoes was higher and the percentage of children who ate peanut butter was lower in Roermond compared to the other towns. The percentage of children who drank low-fat milk or ate (whole) meal bread was higher, and the percentage of children who ate white bread was lower in Harderwift compared to the children in the other towns. These differences in food intake patterns led to some differences in nutrient intake data. However, from the energy percentage point of view these differences were very small.

The energy intake of boys aged 6-10 was higher compared to that of girls of the same age. Consistent differences in energy intake between obese and lean children were not observed. These findings confirmed the results of other studies.Statistically significant relationships between nutrient intake data and serum total cholesterol level could not be found in the present study. Significant relationships between nutrient intake data and serum total cholesterol could be shown when the results of studies carried out in schoolchildren in the period 1946-1951 were included. These findings demonstrated that the relationship between nutrient intake data and serum total cholesterol was concealed when the range in these parameters was relatively small. When this range was increased, significant relationships reappear.Compared to the results of previous studies, the current diet of Dutch children is high in animal proteins, saturated fats, dietary cholesterol and oligosaccharides and low in polysaccharides, vegetable proteins and dietary fibre. Such a diet is known for its atherogenic properties and the results of the present study showed that hypercholesterolaemia, and to a lesser extent obesity, was present in a considerable percentage of children. It is known from prospective studies that the predictive power of CHD risk factors is inversely related to age. From these data it can be inferred that the effectiveness and efficiency of preventive measures could be increased if they are started in childhood.A prudent diet, low in animal proteins, saturated fats, dietary cholesterol and oligo-saccharides and high in polysaccharides, vegetable proteins and dietary fibre, compared to the current diet is recommended for the whole population in the prevention of CHD. Some investigators have also recommended such a diet in the prevention of cancer. It may be expected that if such a diet was used by the whole population, including children , the burden of premature death from chronic diseases could be reduced. To prove this hypothesis by means of intervention trials is one of the major challenges in preventive health of the present time.