The Relationship between Blood Pressure and Anthropometric Indicators in Rural South African Children: Ellisras Longitudinal Study

Background: Underweight, overweight and high blood pressure (BP) are known cardiovascular risk factors in children. The objective of the study was to investigate the relationship between the neck circumference (NC) and mid upper arm circumference (MUAC) with BP in Ellisras rural children age 5-12 years. Methods: All children (n=1029 boys, n=932 girls) had BP and anthropometry measured. The receiver operating characteristics curve was used to discriminate children with high BP. Linear regression was used to assess the relationship between BP and anthropometric indicators. Results: NC and MUAC significantly discriminated children with high BP. The regression analysis showed a significant association for Systolic blood pressure (SBP) with NC (β=0.764, 95% CI 0.475 to 1.052) and MUAC (β=1.286, 95% Cl 0.990 to 1.581) for unadjusted and adjusted for age and gender NC (β=0.628 95% Cl 0.303 to 0.953) and MUAC (β=1.351 95% CI 1.004 to 1.697) also showed a significant association. Conclusion: The prevalence of underweight was high in these children. NC and MUAC can predict children with high BP. There was a positive significant association between BP and anthropometric indicators in this study population.


Introduction
Obesity is one of the major challenges in the world and is a risk factor for cardiovascular disease (CVD). Together with underweight and high blood pressure (BP), they are directly associated with the prevalence of morbidity and mortality from CVD [1]. It is estimated that one-third of the cases of arterial hypertension in adults are related to obesity. Obesity and underweight are likely to predict CVD risk in the long term if they manifest themselves amongst children at an early stage [1][2][3].
Several well-known anthropometric indicators that are in use to diagnose underweight and obesity are waist circumference (WC), hip circumference (HC), body mass index (BMI), neck circumference (NC), mid upper arm circumference (MUAC) and skinfolds [4][5][6]. Both longitudinal and cross sectional studies have shown that these anthropometrics indicators are associated with blood pressure in adults [7][8][9]. Recently, NC and MUAC were reported to be having a significant association with BP in children in developed countries [4,5]. However, the association between NC, MUAC and BP has not been studied extensively in children from rural areas of Africa.
Therefore, the objectives of the present cross-sectional study were to investigate 1) the prevalence of malnutrition (underweight and overweight), 2) assess the susceptibility of NC and MUAC to discriminate children with elevated BP, and finally, 3) the relationship between the anthropometric (NC and MUAC) indicators and BP of Ellisras rural children aged 5-12 years.

Geographical area
Ellisras, now known as Lephalale, is a deep rural area situated within the north-western area of Limpopo Province, South Africa. The population is ~50 000 people residing in 42 settlements. These villages are ~70 km from the town of Ellisras (230 40S 270 44W) adjacent to the Botswana border. The Iscor coal mine and Matimba electricity power station are the major sources of employment for many residents in Ellisras, whereas the remaining workforce is involved in subsistence farming and cattle rearing, while a minority is in education and the civil services.

Sample
The Ellisras Longitudinal Study (ELS) initially followed a cluster sampling method. In brief, we estimated the prevalence size from the literature, selected a level of significance (P<0.05) and then determined the power of the test which produced a significant difference at a given significant level. The study was undertaken at 22 schools (10 preschool and 12 primary schools) randomly selected from 68 schools within the Ellisras area. Birth records were obtained from the principals of each school. Only those records that were verified against health clinic records were used to determine the age of each potential participant. Each of the 22 selected schools was assigned a grade with the expectation that most of the children in a particular age category (3-10 years) would be found in that grade.
In May 1999, medical students from VU University, Amsterdam included the BP parameter in the ongoing anthropometric measurements of the ELS participants, and this was performed on 1029 boys and 932 girls (n=1961), aged 5-12 years. The Ethics Committee of the then University of the North, now known as University of Limpopo, granted ethical approval prior to the survey, and the parents or guardians provided informed consent.

Anthropometry
All children underwent a series of anthropometric measurements according to the International Society for the Advancement of Kinathropometry (ISAK) [10]. Weight was measured on an electronic scale to the nearest 0.1 kg, and a Martin anthropometer was used to measure height to the nearest 0.1 cm. Neck circumference and mid upper arm circumference were measured in centimeters with a flexible steel tape.

Blood pressure
Using an electronic Micronta monitoring kit, at least three BP readings of systolic blood pressure (SBP) and diastolic blood pressure (DBP) were taken at an interval of five minutes apart after the child had been seated for 5 minutes or longer [11,12]. The bladder of the device contains an electronic infrasonic transducer that monitors the BP and pulse rate, displaying these concurrently on the screen. This versatile instrument has been designed for research and clinical purposes. In a pilot study, conducted before the survey, a high correlation (r=0.93) was found between the readings taken with the automated device and those taken with a conventional mercury sphygmomanometer.

Quality control
All training of anthropometric measurements was done in accordance with the standard procedures of the ISAK [10]. Reliability and validity of anthropometric measurements were reported elsewhere [13,14]. In brief, the absolute and relative values for intra-and intertester technical error of measurements (% TEM) for stature, ranged from 0.04-4.16 cm (0.2-5.01%), for body mass from 0.01-0.02 kg and girth measurements range from 0.0-3.4 cm (0-4%).

Statistical analysis
Descriptive statistics were presented for NC, MUAC, BMI, BP, height and weight in the Ellisras rural children aged 6 to 12 years. The independent t-test was applied to test the significance level (P<0.05) between sexes. All children were classified as overweight and obese according to Cole et al. [15] cutoff points. The international cutoff points for underweight (grade one, two, and three) by sex for exact ages defined to pass through BMI of 16, 17, and 18 kg/m 2 were used [16,17]. Hypertension was defined as the occurrence of SBP and DBP levels greater than or equal to the 95th percentile of height and sexadjusted reference levels [10]. We assessed the susceptibility for NC, MUAC and BMI to discriminate children with elevated BP. For this purpose, we produced sex-specific receiver operating characteristics (ROC) curves and used the corresponding area under curves (AUC) to determine the ability of each anthropometric indicator in identifying children with elevated BP. The ROC curve is a plot of true-positive rate (sensitivity) against the false-positive rate (1-specificity). A good test has ROC skewed to the upper left corner with AUC of 1, whereas an AUC of 0.5 means that the test performs no better than chance [18,19]. Sensitivity and specificity of NC and MUAC have been calculated at all possible cutoff points to find optimal cutoff value. The optimal sensitivity and specificity were the values yielding maximum sums from the ROC curves (clinical significance of cutoff was checked with the Youden index). Cutoff values and the corresponding AUCs as well as positive predictive value (PPV) and negative predictive value (NPPV) of NC and MUAC for underweight were computed along age and gender. The Linear regression models were used to assess the relationship between BP and anthropometric indicators (NC and MUAC) adjusted for age and sex. All the statistical analyses were performed using the Statistical Package for the Social Sciences (SPSS). The statistical significance was set at P<0.05. Table 1 showed descriptive statistics for weight, height, NC, MUAC, BMI and BP by gender and age group for Ellisras rural children aged 5 to 12 years. Boys (25.6 to 26.2 cm) showed a significantly (P< 0.05) higher mean NC than girls (25.0 to 25.8 cm) from age 9 to 11 years, while girls (17.9 cm) exhibited a significantly higher mean MUAC than boys (17.4 cm) at age 11 years. Boys and girls did not show significant differences with their mean BP throughout the age range.     Tables 4-5 showed the AUC for each age group including the optimal NC and MUAC cutoff points and the corresponings sensetivities and specificities for classifying children into low BMI groups in boys. The predictive values for each cutoff point were also shown. For example, PPV for a 12 year old girl with NC<26.2 cm and MUAC<18.6 cm indicate that she is more likely to be underweight (    Table 6: Linear regression coefficients, P-value and 95% confidence intervals for the association between Neck circumference (NC), mid upper arm circumference (MUAC) and blood pressures in Ellisras

Discussion
The relationship between anthropometric indicators and BP is important in evaluating the health of children and adolescents, since cardiovascular abnormalities are becoming more common these days. In this present study, we found that the prevalence of underweight (severe, moderate and mild) was high while overweight and obese by BMI was low amongst these children. NC and MUAC significantly discriminated boys with both high SBP and DBP, while in girls NC and MUAC only identified high SBP. However, BMI discriminated boys with only high SBP.
In this study, we measured NC and MUAC to describe underweight defined by BMI. We recommend that each country must determine their own cutoff points for NC, MUAC and BMI. While the use of BMI as a surrogate for fat distribution among children raises debates, NC and MUAC are increasingly recognized as a useful indexes refletive of both fat distribution and risk of metabolic syndrome. The primary contribution of this present study was that both NC and MUAC can be substituted for one another as an additional evaluation tool next to BMI in detecting underweight in the Ellisras rural children.
The results showed a strong positive significant association between BP and anthropometric indices (NC and MUAC) in this sample though the association of DBP and NC disappeared after adjustments for age and gender. Similar findings do exist in literature that supports our results [5]. However, it seems that the prevalence of overweight, obesity and high BP of the cited studies were high compared with the current study, which showed high prevalence of underweight (Table  2). These differences may be due to the fact that those studies were carried out in developed countries in contrast to our study which is exclusively based on under-resourced rural settings. According to WHO [20], the prevalence of overweight or obesity increases as a country develops. It is possible that nowadays the aforementioned changes and developments in South Africa have resulted in increased prevalence of overweight and obesity [21].
Few studies have reported the association of high BP and several anthropometric indicators. However, it is still controversial whether absolute (general adiposity) or relative body fat (central adiposity) is a better predictor for hypertension [22]. Most studies are supporting central adiposity as the good predictor for high BP [23,24]. NC and waist circumference (WC) as a marker of central adiposity appeared to have the accuracy of discriminating children with high BP in America [5,25]. Moreover, compared with waist circumference, NC seems to be an ideal measure as it does not change during the day [26]. The present study agree with the previous studies as NC discriminated children with high BP and also added that MUAC can also be used to identify children with high BP.
For the association of anthropometrics and BP, Nafiu et al. [5] found a positive association between BP and NC in American children aged 6-18 years. NC showed an independent association with various cardiometabolic risk markers including BP [27] and had strong correlation with BMI [28]. Mazicioglu et al. [4] reported that MUAC is strongly associated with both SBP and DBP in adolescents aged 11-17 years from Turkey. Bassaroe et al. [29] also reported that MAUC had a significant relationship with systolic BP but not diastolic BP in Sardinian children aged 11-14 years. Therefore present study concurs with previous studies on the association of NC and MUAC with BP.

Study Limitations
In our study we did not consider the socioeconomic status of families of the participants in the analysis. Insulin resistence as metabolic condition characterized by an impaired ability of plasma insulin at normal concentrations to promote peripheral glucose clearance, suppress glucoseneogenesis in the liver and inhibits very low density lipoprotein output. This condition together with an increase in BMI values worsen the endothelial parameters, independent of malnutrition of children were not investigated in our study [30][31][32]. Furthermore, even though the data of the study was collected in year 1999, it is possible that prevalence of overweight and obesity has increased, at present time, due to developments and changes in South Africa [21].
The anthropometric and BP measurements were taken directly, and hence recall or estimation bias will not prevail in our study. In addition, we measured BP during early childhood, demonstrating that proper monitoring should be started from children's early days from a viewpoint of screening for vulnerable individuals [33].

Conclusion
The prevalence of underweight (severe, moderate and mild) was high, while overweight was low amongst Ellisras rural children. NC and MUAC can predict both high SBP and DBP in boys while in girls NC and MUAC can only predict high SBP. However, BMI can only predict high SBP for boys only. There was a positive significant association between BP and anthropometric indicators in this study population, though the association of DBP and NC disappeared after adjustments for age and gender. Further research is recommended to establish the NC and MUAC cutoff points for overweight, especially in low-resourced settings as these indices are inexpensive to use and can be used in larger populations.