Global malnutrition among children is primarily targeted at children aged less than five. WHO reported that an estimated 149 million and 45 million children aged under 5 years were stunted and wasted, respectively, whereas 38 million were overweight or obese globally(1). In contrast, school-age children are scarcely monitored globally despite the significant impact of malnutrition on their health, cognitive function, academic performance and future economic productivity(Reference Asmare, Taddele and Berihun2,Reference Nugent, Levin and Hale3) .
In Malaysia, it is known that malnutrition among children remains a health concern despite steady economic growth in the past decades. The Malaysian National Health and Morbidity Survey (NHMS) 2019 reported that the prevalences of stunting (height-for-age z-score (HAZ) < –2 sd), thinness (BMI-for-age z-score (BAZ) < –2 sd) and underweight (weight-for-age z-score (WAZ) < –2 sd), among children aged 5–17 years, were 12·7, 10·0 and 15·4 %, respectively(4). Besides, 15·0 and 14·8 % of the same group of children were found to be overweight (2 sd < BAZ ≤ 3 sd) and obese (BAZ > 3 sd), respectively(4), thus highlighting the phenomenon of double burden.
Throughout childhood, micronutrients play significant roles in immune function, energy production, learning and cognitive functions(Reference Nyaradi, Li and Hickling5). It is estimated that micronutrient deficiencies affect at least 340 million children aged under five globally(6). Vitamin A deficiency (VAD) is a significant public health problem affecting about one-third of children aged less than five in 1995–2015, with Southeast Asia recording the highest prevalence at 49·9 %(7). VAD affects normal haematopoiesis, Fe metabolism and immune function(Reference Canete, Cano and Munoz-Chapuli8) due to inadequate vitamin A intake or reduced availability of pro-vitamin A, namely carotenoids(Reference Schweigert, Klingner and Hurtienne9).
South-East Asian Nutrition Survey (SEANUTS) Malaysia reported a prevalence of 4·4 % VAD among children aged 6 months to 12 years old, with higher prevalence recorded in rural areas (6·4 %) compared with urban areas (3·8 %)(Reference Poh, Ng and Siti Haslinda10). In the same survey, it was found that the overall prevalences of iron deficiency (ID) and anaemia were 4·4 and 6·6 %, respectively. In addition, based on several studies conducted among Orang Asli (OA) (indigenous) schoolchildren, the prevalence of anaemia ranged from 26·2 % to 68·4 %, while VAD prevalence was at 27·4 %, and 36·7–54·9 % of them were found to be Fe deficient(Reference Ngui, Lim and Chong Kin11–Reference Al-Mekhlafi, Surin and Sallam14). OA, which transliterates as original people in the Malay language, is the indigenous minority peoples of Peninsular Malaysia, which account for 0·7 % of the population of Peninsular Malaysia(15).
Despite nationwide anthropometric data being highly accessible as indicators of nutritional status in Malaysia, there are limited coherent and updated data on other nutrition and health indicators among children, mainly primary school-age children from rural areas. Thus, this study aims to collate data on the anthropometry and blood biochemical status of primary schoolchildren aged between 7 and 11 years old in the rural areas of Malaysia. The selected micronutrients (vitamin A, α-carotene, β-carotene and vitamin E), haemoglobin (Hb) status, ferritin status (indicator for ID), inflammation status and their associations with the nutritional status of the children were discussed in the present study. Potential socio-economics- and blood biochemical-related factors associated with stunting, the most prevalent malnutrition problem among this population, were examined.
Method
Study areas and subjects
This cross-sectional study was carried out between April 2017 and October 2017. Ten national primary schools from five different states located in rural areas of Malaysia were randomly selected based on suggestions and lists from the Ministry of Education (MOE) and Department of Orang Asli Development (JAKOA), taking into consideration the following criteria: (i) approval by the Ministry of Health (MOH) and MOE; (ii) schools with a population of at least fifty children; (iii) socio-economic status (SES) is generally poor; (iv) accessible by road transportation (for rapid transportation of samples to the laboratory for preservation and storage); (v) Malaysian children aged 7–11 years old and (vi) healthy at the point of the study period.
Among the selected schools, five schools are OA schools consisting of OA majority students (mainly of Semai and Temiar subgroups) and located in the vicinity of indigenous villages. At the same time, another five are Non-Orang Asli (NOA) schools consisting of NOA students (mainly of Malay, Kadazandusun and Iban ethnicities). The locations of the selected schools are shown in Fig. 1. Absent or ill children during the period of study were excluded.
Locations of the selected schools in Peninsular, Sabah and Sarawak of Malaysia. In total, ten national primary schools from five different states located in rural areas of Malaysia were selected. Among the selected schools, five schools are Orang Asli (OA) schools consisting majority of OA students (mainly of Semai and Temiar subgroups) and located in the vicinity of indigenous villages. Another five are Non-Orang Asli (NOA) schools consisting of NOA students (mainly of Malay, Kadazandusun and Iban ethnicities).
The sample size was calculated based on the prevalence of stunting in rural areas at 12·7 %, as reported by NHMS 2019(4) according to the following formula(Reference Leedy16):
$${n \ge {(z/m)^{2}} \times p(1-p)}$$
where n is the minimum sample size, z is the standard score (1·96), m is the rate of sampling error (5 %) and p is the estimated prevalence of the variable in the population. At a significance level of 5 % and a confidence level of 95 %, a minimum sample size of 218 participants was required for this study.
Ethical approval, consent and socio-demographic data collection
The study was conducted according to the guidelines laid down in the Declaration of Helsinki. All procedures involving human subjects were approved by the Medical Research and Ethics Committee, MOH, Malaysia (NMRR-16–1905–32 547). Before the commencement of the study, the participants and parents/guardians were given an oral briefing on the objectives and methodology of the study. They were also informed that their child’s participation was voluntary, and therefore they could withdraw from the study at any time. For all the literate guardians, written informed consents were obtained before the study commenced. As for illiterate guardians, verbal consents were obtained, followed by their thumbprint on the informed consent form. All the verbal consents were witnessed and formally recorded. Besides, an assent form was signed by the children themselves. Socio-demographic data, including age, sex and monthly household income of the children, were collected from student database obtained from the respective school’s administration and parents. Monthly household income was categorised into two groups (<RM500 and ≥ RM500). A monthly household income of <RM500 was regarded as being below the Malaysia poverty income threshold(Reference Nicholas17). The trial was registered on ClinicalTrials.gov with identification number NCT03256123 and can be accessed at https://clinicaltrials.gov/.
Anthropometric assessment
Height and weight measurements were performed using SECA Clara 803 and SECA 213 mobile stadiometer. The height measuring device was placed near the wall according to the device manual. During height measurement, students were requested to remove any footwear or hair accessories that may obstruct measurements. Height was recorded to the nearest 0·1cm. Before weight measurement was conducted, students were asked to empty their pockets. Calibrated SECA weighing scale was placed on a level surface. Each student was instructed to stand barefoot with minimal clothing and empty pockets at the centre of the scale. The weight was recorded to the nearest 0·1 kg. To reduce intra-observer error, height and weight were measured twice and the mean value was used for analysis. BMI was computed by dividing the measured weight (kg) by the square of height (m).
Anthropometric indices were computed using anthro R-package provided by WHO(Reference Schumacher, Borghi and Polonsky18). There were three anthropometric indices: (a) HAZ to assess stunting and (b) BAZ to assess thinness, overweight and obesity and (c) WAZ to assess underweight. All three anthropometric indices were expressed as differences from the median in standard deviation units or z-scores. Students were classified as stunted, thin and underweight if HAZ, BAZ and WAZ were less than 2 sd below the WHO reference median(19). On the other hand, BAZ value 1 and 2 sd above WHO reference median were used to determine overweight and obesity status, respectively(19).
Blood biochemical assessments
Blood collection was conducted within a week of the anthropometric measurements. After an overnight fast (10 h), approximately 6 ml of venous blood was collected by trained nurses and medical assistants from MOH. Vacutainer tubes were covered with aluminium foil to prevent direct exposure to sunlight and kept in an icebox surrounded by ice and ice packs. The blood samples were then transported back to the Nutrition Unit Laboratory in Malaysian Palm Oil Board (MPOB), Bangi, Selangor, on the same day of collection for processing. An automated haematology analyser performed a complete blood count on the same day (Sysmex XN-10). The rest of the blood tubes were centrifuged at 3000 rpm for 15 min at 4°C to obtain serum and plasma and were stored at –80°C until further analysis.
Plasma retinol, α-carotene, β-carotene and α-tocopherol (vitamin E) levels were measured using reverse-phase HPLC (Agilent 1260 Infinity) as described by Kand’ár et al. (Reference Kand’ár, Novotná and Drábková20). Briefly, plasma samples were extracted twice with hexane, while retinyl acetate and α-tocopheryl acetate were used as internal standards. The mean recovery of retinyl acetate and α-tocopheryl acetate was 94·45 (sd 3·34) % and 97·73 (sd 3·44) %, respectively. Retinol-binding protein (RBP) was measured by quantitative sandwich enzyme immunoassay (R&D Systems). VAD was defined by both retinol and RBP concentrations at <0·70 µmol/l, while marginal VAD was defined when the concentrations were at 0·70 to <1·05 µmol/l(21,Reference De Pee and Dary22) .
Anaemia was diagnosed when a complete blood count test shows Hb concentration <115 g/l, and further classified by the degree of severity into severe (<80 g/l), moderate (80–109 g/l) and mild (110–114 g/l)(23). The concentration of serum ferritin was measured by two-site sandwich immunoassay via a direct chemiluminescence method using ADVIA Centaur (Siemens). Serum ferritin value of <15 µg/l was considered as ID, while iron-deficiency anaemia (IDA) was defined as ID concurrent with anaemia(Reference Ngui, Lim and Chong Kin11,24) . High-sensitivity C-reactive protein (hs-CRP) was measured using an immunoturbidimetric method. hs-CRP concentration > 5·0 mg/l was considered as having a high inflammatory response(Reference Kundu, Rai and Shukla25). All these procedures on Hb, serum ferritin and hs-CRP were conducted at an accredited laboratory, Pathology & Clinical Laboratory (M) Sdn. Bhd., Malaysia.
Statistical analysis
Data collected were entered and analysed using the Statistical Package for the Social Sciences (IBM SPSS Statistics) programme for Windows version 22. Before analysis was carried out, data entered were cross-checked on a timely basis to ensure all the data were entered accurately. For descriptive data, count (percentage) was used. Kolmogorov–Smirnov Z test was used to examine the normality of quantitative data. As most of the data did not meet the criterion of a normal distribution, the distribution of continuous data was presented as the median and interquartile range (25th, 75th percentiles; P25, P75). Mann–Whitney U test or Kruskal–Wallis H test (followed by Dunn’s multiple comparison test as appropriate) was used to test for differences between continuous variables with two or more than two groups, respectively. Categorical variables are reported as percentages and compared by the χ 2 test. Univariate and multivariate logistic regressions were performed to examine the potential factors associated with stunting. All variables that were significantly associated with stunting in univariate model were included in a logistic multivariate analysis using forward elimination model, and presented as the crude OR (COR) and adjusted OR with 95 % CI. Model fitness was assessed by the Hosmer–Lemeshow goodness of fit test. A value of P < 0·05 was considered statistically significant.
Results
Socio-demographic characteristics
Table 1 demonstrates the socio-demographic characteristics of the overall population. A total of 776 children (379 boys and 397 girls) aged 7–11 years were included in the final analysis. Overall, there were 458 (59 %) children aged <10 years and 318 (41 %) children aged ≥ 10 years with a median age of 9 years (P25:8, P75:10). No significant difference was observed for age between sexes. The highest number of children was recruited from the states of Pahang (36·6 %), followed by Perak (21·8 %), Sarawak (18·4 %), Johor (12·4 %) and Sabah (10·8 %). More than half of the population (63·8 %) were studying in OA schools, and the majority (93 %) were from <RM 500 monthly household income families.
Socio-demographic characteristics of the overall population
(Numbers and percentages, median and percentiles, n=776)
Anthropometric characteristics
Overall, it was found that 53·4 % of children have malnutrition problems, with 34·9, 4·8 and 30·5 % of the children were stunted, thin and underweight (only included children aged <10 years, n=551), respectively, while 15·5 % were found to be overweight/obese (Table 2). Among stunted children, 29·9 % were severely stunted children, while for those who were underweight, 26·2 % were severely underweight. In addition, 29·7 % of children were found to be severely thin among those who had thinness problems. Similar anthropometric characteristics were observed between sex and age group (Table 2).
Anthropometric characteristics of the children stratified by school category, age group and sex (Median values and percentiles, numbers and percentages, n=776)||
WAZ, weight-for-age z-score; HAZ, height-for-age z-score; BAZ, BMI-for-age z-score.
Significant difference based on Mann–Whitney U test: *P < 0·05; **P < 0·01; ***P < 0·001.
Significant difference based on χ 2 test: †P < 0·05; ††P < 0·01; †††P < 0·001.
‡ Only included children aged ≤ 10 years old (n=551).
§ Children with any form of malnutrition (stunting, thinness, overweight and underweight).
|| Data presented as median (25th percentile, 75th percentile) or n (%).
¶ Hodges–Lehmann estimation and its associated 95 % CI.
The previous analyses based on age and sex among the population showed no significant difference, hence we further analysed the population based on school category. Children from OA schools were found to have significantly lower values of anthropometric measurements, including height, weight, BMI, HAZ, BAZ and WAZ compared with children from NOA schools (P < 0·001). In addition to that, the prevalences of stunting and underweight among children in OA schools (47·3 and 40·0 %) were more than three times higher than in NOA schools (11·8 and 12·9 %) (P < 0·001). On the other hand, NOA schoolchildren had significantly higher prevalences of overweight and obesity (11·8 and 12·2 %) than their counterparts (6·8 and 3·8 %).
Biochemical characteristics
Generally, the children aged ≥ 10 years exhibited significantly higher values for all the blood parameters than those aged <10 years old except α-tocopherol, serum ferritin and hs-CRP, where they exhibited higher but not significantly different (P ≥ 0·05) (Table 3). By sex, a higher value of α-tocopherol was observed among girls (8·01 µmol/l (P25:6·89, P75:9·25)) compared with boys (7·67 µmol/l (P25:6·62, P75:8·91)) (P = 0·03). On the other hand, ferritin level in girls (34·0 µg/l (P25:21·0, P75:54·0) was significantly lower compared with boys (53·0 µg/l (P25:28·0, P75:76·0)) among children aged ≥ 10 years (P < 0·001) (data not shown).
Biochemical characteristics of the children stratified by school category, age group and sex (Median values and percentiles, numbers and percentages, n=776)‡
RBP, retinol-binding protein; hs-CRP, high-sensitivity C-reactive protein.
Significant difference based on Mann–Whitney U test: *P < 0·05; **P < 0·01; ***P < 0·001.
Significant difference based on χ 2 test: †P < 0·05; ††P < 0·01; †††P < 0·001.
‡ Data presented as median (25th percentile, 75th percentile) or n (%).
§ Hodges–Lehmann estimation and its associated 95 % CI.
Overall, the prevalence of VAD based on retinol and RBP levels was 20·6 and 39·8 %, respectively (Table 2). More than half of the children suffered from a marginal deficiency of retinol (56·8 %), and nearly half of them were found to have a marginal deficiency of RBP levels (47·8 %). The overall prevalence of anaemia was 14·9 %. Nearly half of the anaemic children exhibited mild anaemia (48·3 %) and moderate anaemia (49·1 %), while only 2·6 % had severe anaemia. A higher prevalence of VAD, based on both retinol and RBP level <0·7 µmol/L, was observed among children aged <10 years (23·4 and 43·7 %) when compared with their older counterparts (16·7 and 34·3 %) (Table 3). Also, the prevalence of anaemia among children aged <10 years (19·2 %) was about two times higher than their older counterparts (8·8 %) (P < 0·001) (Table 3).
Table 3 shows the biochemical characteristics of children from both OA and NOA schools. OA schoolchildren were observed to have significantly lower concentrations of retinol (0·83 µmol/l (P25:0·69, P75:0·97)), RBP (0·70 µmol/l (P25:0·58, P75:0·82)), α-carotene (0·06 µmol/l (P25:0·04, P75:0·08)), Hb (124·0 g/l (P25:117·0, P75:129·0)) and ferritin (35·0 µg/l (P25:19·5, P75:62·5)). In contrast, the concentrations of α-tocopherol (8·03 µmol/l (P25:6·94, P75:9·25)), β-carotene (0·30 µmol/l (P25:0·20, P75:0·43)) and hs-CRP (0·2 mg/l (P25:0·2, P75:1·2)) were found to be significantly higher among the OA schoolchildren. Moreover, VAD based on both retinol (27·8 %) and RBP (51·1 %) parameters was more prevalent among OA schoolchildren. Higher occurrences of anaemia (20·1 %), ID (17·9 %), IDA (9·1 %) and high inflammation (hs-CRP levels > 5·0 mg/l) (11·5 %) were noted in the same group of children.
Biochemical measurements in relation to nutritional status
Table 4 illustrates the median levels of biochemical measurements in relation to the nutritional status determined by anthropometric indicators. In general, significantly lower levels of retinol, RBP, α-carotene, Hb and ferritin were perceived among stunted and underweight children. On the other hand, overweight children were found to have higher levels of retinol (0·96 µmol/l (P25:0·79, P75:1·21)), RBP (0·87 µmol/l (P25:0·71, P75:1·05)), Hb (133·5 g/l (P25:125·0, P75:138·0)), ferritin (53·0 µg/l (P25:32·5, P75:84·8)) and hs-CRP (1·2 mg/l (P25:0·2, P75:4·2)) compared with both thin and normal children.
Biochemical measurements in relation to the nutritional status of the children
(Median values and percentiles, numbers and percentages, n=776)
RBP, retinol-binding protein; hs-CRP, high-sensitivity C-reactive protein; P25, 25th percentile; P75, 75th percentile; HAZ, height-for-age z-score; BAZ, BMI-for-age z-score; WAZ, weight-for-age z-score.
Significant difference based on Mann–Whitney U test: *P < 0·05; **P < 0·01; ***P < 0·001.
Significant difference based on Kruskal–Wallis H test: †P < 0·05; ††P < 0·01; †††P < 0·001.
Significant difference based on χ2 test: †P < 0·05; ††P < 0·01; †††P < 0·001.
Different alphabets in the same column within each variable indicate significant statistical differences (P < 0·05, Dunn’s multiple comparison test).
‡ Only included children aged ≤ 10 years old (n=551).
Potential factors associated with stunting
Given that stunting was one of the most prevalent malnutrition problems, the associations between stunting with other variables were further analysed and shown in Table 5. Univariate logistic regression revealed that OA schoolchildren (COR: 6·1 (95 % CI 4·1, 9·0), P = <0·001), monthly household income <RM500 (COR: 2·5 (95 % CI 1·2, 5·0), P = 0·009), deficiency in retinol level (COR: 2·9 (95 % CI 1·8, 4·7), P = <0·001), marginal deficiency in retinol level (COR: 2·0 (95 % CI 1·3, 3·0), P = 0·001), deficiency in RBP level (COR: 2·9 (95 % CI 1·7, 5·1), P = <0·001), marginal deficiency in RBP level (COR: 2·0 (95 % CI 1·1, 3·4), P = 0·014), anaemia (COR: 2·2 (95 % CI 1·5, 3·3), P = <0·001), ID (COR: 2·4 (95 % CI 1·6, 3·7), P = <0·001) and IDA (COR: 3·6 (95 % CI 1·9, 6·6), P = <0·001) were significantly associated with stunting. The final multivariate analysis indicated that only OA schoolchildren (adjusted OR: 6·1 (95 % CI 4·1, 9·0), P = <0·001) and IDA (adjusted OR: 3·6 (95 % CI 1·9, 6·6), P = <0·001) retained as significant predictors for stunting among the children (goodness of fit: χ 2 = 0·683 (df = 2); P = 0·71).
Univariate and multivariate logistic regression models for determination of factors associated with stunting
(Odds ratios and 95 % confidence intervals, numbers and percentages, n=776)
RBP, retinol-binding protein; hs-CRP, high-sensitivity C-reactive protein.
Significant difference: *P < 0·05; **P < 0·01; ***P < 0·001.
Discussion
This study depicts a high level of malnutrition (53·4 %) in primary schoolchildren living in rural areas of Malaysia. Our results simultaneously demonstrate the coexistence of underweight, stunting, thinness, overweight and obese children in rural areas of Malaysia, which is a common and significant problem in low- and middle-income countries and is described as a double burden of malnutrition(Reference Winichagoon and Margetts26). Based on the recent NHMS 2019, the prevalence of stunting (17·1 %) and underweight (14·2 %) reported among children 5–17 years living in rural areas was about three times lower than our current findings. On the other hand, NHMS 2019 reported a higher prevalence of overweight (13·7 %), obesity (13·2 %) and thinness (11·2 %) compared with the prevalence reported in this study(4). Another earlier study by Poh et al. (Reference Poh, Ng and Siti Haslinda10) also demonstrated a lower prevalence of stunting (7·3 %) and higher prevalence of overweight (12·6 %), obesity (13·0 %) and thinness (9·3 %) among rural children aged 7–12 years as compared with our present study outcomes. The differences could be due to the different sampling protocols, population type, geographical areas, SES and age category (inclusion of pre-school age and adolescent children in the previous surveys).
Our study demonstrated a higher prevalence of overall malnutrition and undernutrition (stunting and underweight) among OA schoolchildren, while a lower prevalence of overnutrition (overweight and obesity) compared with NOA schoolchildren, with no difference observed between sex and age groups. Comparable findings were reported by previous local studies(Reference Zulkifli, Anuar and Athiya27,Reference Mas-Harithulfadhli-Agus, Hamid and Rohana28) . Nevertheless, the high prevalence of undernutrition among the OA children in this study reflects the persistence of poor nutrition among OA children, as they are significantly associated with socio-economic disadvantage, which leads to food insecurity and compromised growth and development(Reference Mas-Harithulfadhli-Agus, Hamid and Rohana28).
Based on SEANUTS Malaysia, VAD is considered a mild public health problem among Malaysian children as only a tiny percentage of children (4·4 %) were observed to have VAD with a higher prevalence in rural areas (6·4 %)(Reference Poh, Ng and Siti Haslinda10). However, the present study revealed a higher prevalence of VAD among rural schoolchildren at 20·6 % based on retinol concentration and 39·8 % based on RBP. Based on WHO definition (29), VAD among our children population is considered a severe public health problem, whereby more than ≥ 20 % of them were diagnosed with retinol concentration <0·7 µmol/l. The notable variations in the VAD prevalence reported could be due to the differences in population, ethnicity and geographical areas covered, as SEANUTS Malaysia does not include OA population.
The prevalence of VAD based on retinol level among OA schoolchildren in the present study was similar to findings reported by previous local studies(Reference Al-Mekhlafi, Surin and Sallam14,Reference Al-Mekhlafi, Azlin and Aini30) . In addition, Ngah et al. (Reference Ngah, Moktar and Isa31) reported a high occurrence of ocular manifestation of VAD at 64·3 % among 213 OA children aged under 15 years old in an OA settlement at Pos Piah, Perak. However, retinol concentration, which is more commonly used in vitamin A status assessment, was not being measured in the study mentioned above. In the present study, compared with NOA schoolchildren, OA schoolchildren exhibited a higher prevalence of VAD. Poor dietary intake/food availability, dietary, cultural behaviour and relatively high intestinal parasitic infections in OA schoolchildren due to poor hygiene practices and inadequate sanitation may contribute to this difference in prevalence(Reference Al-Mekhlafi, Surin and Sallam14,Reference Al-Mekhlafi, Azlin and Aini30,Reference Elyana, Al-Mekhlafi and Ithoi32) .
We also found that retinol and RBP levels increased with age, while VAD decreased with age. These findings were consistent with those found in previous studies(Reference Poh, Ng and Siti Haslinda10,Reference Yang, Chen and Guo33–Reference Drott, Meurling and Gebre-Medhin36) . It is possibly due to physical growth, lower dietary diversification and higher risk of infections among the younger age group(Reference Yang, Chen and Guo33). Another possible reason could be the use of a single cut-off point of VAD status for children, which might overestimate VAD status among younger children who possibly have lower physiological vitamin A levels(Reference Chen, Liu and Mao35).
The present study demonstrated that the prevalence of anaemia and ID was at 14·9 and 12·8%, respectively, which is in line with the classification of anaemia as a mild public health problem (5·0–19·9 %) based on WHO cut-off(23). Higher rates were reported in other local studies(Reference Ngui, Lim and Chong Kin11). Meanwhile, Poh et al. (Reference Poh, Ng and Siti Haslinda10) reported a lower prevalence of anaemia (5·1%) and ID (2·2%) among Malaysian rural children aged 7–12 years old. However, among the currently studied OA schoolchildren, the prevalence of anaemia, ID and IDA was relatively lower than the range of prevalence reported in previous local studies conducted among OA children(Reference Al-Mekhlafi, Surin and Atiya12,Reference Muslim, Lim and Mohd Sofian13,Reference Aini, Al-Mekhlafi and Azlin37–Reference Yee, Asma’Ali and Yusof39) . This study also observed that the OA schoolchildren exhibited significantly higher anaemia, ID and IDA levels with lower Hb and ferritin concentrations than NOA schoolchildren. Based on findings from previous studies, recurrent infections, in particular soil-transmitted helminths, low dietary intake of Fe and poor SES among OA schoolchildren could be the contributing factors(Reference Ngui, Lim and Chong Kin11,Reference Al-Mekhlafi, Surin and Atiya12) .
In addition, we found that older children exhibited higher Hb levels and a lower prevalence of anaemia, which is following previous findings(Reference Agho, Dibley and D’Este40–Reference Gonzales, Tapia and Vásquez-Velásquez42). The current observation could be partly explained by the significantly lower levels of vitamin A in children aged <10 years, as vitamin A plays a vital role in modulating Fe metabolism and erythropoiesis(Reference Semba and Bloem43). By age group, our study showed that both α- and β-carotene levels were higher among children aged ≥ 10 years. This scenario is consistent with findings reported by Gregory et al. (Reference Gregory, Lowe and Bates44), where β-carotene levels increased with age in both sexes among British children. On the contrary, α- and β-carotene levels were inversely related to age among US children and adolescents aged 6–16 years old(Reference Ford, Gillespie and Ballew45). The reason for these differences in findings is unclear, but we speculate that it could be due to the diversity of plant-based foods consumption in different geographical locations.
In the present study, we observed that OA schoolchildren exhibited higher levels of α-tocopherol and β-carotene and lower α-carotene levels than NOA schoolchildren. The cause of this outcome is uncertain as diet consumption data are not collected in this study. Nevertheless, the higher level of β-carotene among OA schoolchildren may indicate higher consumption of carotenoid-containing foods as the proportion of β-carotene in plant sources and their theoretical conversion efficacy to vitamin A (retinol) are higher than α-carotene(Reference Nagarajan, Ramanan and Raghunandan46).
In the comparative analysis, both retinol and RBP levels were also found to be significantly lower among stunted and underweight children while higher among overweight/obese children. These findings are in agreement with previous studies that link VAD with a higher risk of stunting and underweight among children, which may reflect the impact of VAD for a prolonged period on growth retardation(Reference Ssentongo, Ba and Ssentongo47,Reference Gowele, Kinabo and Jumbe48) . In addition, similar to our findings, previous studies have shown that RBP levels were higher among overweight and obese groups, which indicates RBP levels are associated with a degree of adiposity(Reference Rhie, Choi and Eun49,Reference Reinehr, Stoffel-Wagner and Roth50) . We also observed that Hb levels were significantly lower among stunted and underweight children than overweight/obese children. Previous studies have also shown an association between anaemia with undernourished children(Reference Aini, Al-Mekhlafi and Azlin37,Reference Rahman, Mushfiquee and Masud51,Reference Hoang, Orellana and Le52) . A meta-analysis conducted among pre-school-age children also demonstrated that anaemia was associated with stunting and being underweight in most studies(Reference Engle-Stone, Aaron and Huang53).
A reduced level of ferritin was also observed among stunted and underweight children. Both ferritin and hs-CRP levels were also found to be elevated among overweight/obese children. Numerous studies have reported elevated inflammatory markers among overweight and obese people(Reference Sal, Yenicesu and Celik54–Reference Sharma, McKenna and Lepage56). Besides, the present study findings are similar to findings reported by Khan et al. (Reference Khan, Khan and Ayub57), where ferritin and hs-CRP were found positively associated with BMI. These findings support the evidence on elevated ferritin levels in the obesity-related inflammatory process as serum ferritin is an acute-phase reactant protein that is similar to hs-CRP, which increases in response to inflammation due to increased cytokine levels(Reference Khan, Khan and Ayub57,Reference Shattnawi, Alomari and Al-Sheyab58) .
Further findings showed that stunting problem was associated with OA school and IDA. Stunting is known to occur concurrently with ID and anaemia, which is associated with poor SES(Reference Mohammed, Larijani and Esmaillzadeh59). It is reported that individuals with poor SES are more likely to have low-quality diets which are less adherent to dietary guidelines, thus leading to their poor health status(Reference Alkerwi, Vernier and Sauvageot60). In this study, majority (93 %) of the current population have low monthly household income (<RM500) which could be one of the contributing factors for the occurrence of IDA among the children, hence increasing the odds of being stunted. IDA can be due to the low intake of Fe from animal food sources, especially among populations consuming mainly plant-based diets(Reference Skolmowska and Głąbska61). It was reported that OA children have low dietary diversity and meat consumption, which could be the causative factor for IDA resulting in higher risk of stunting among them(Reference Zalilah and Tham62).
Most of the previous studies that discussed malnutrition among OA children targeted pre-school-age children, mainly from single education institutions or villages(Reference Yee, Asma’Ali and Yusof39,Reference Wong, Zalilah and Chua63–Reference Murtaza, Gan and Sulaiman65) . To our knowledge, this is the first study to report malnutrition levels based on both anthropometric assessments and biochemical assessments among children from multiple OA schools and their comparisons with children from NOA schools in Malaysia. The present study provides valuable findings on the nutritional status of Malaysian children from rural areas.
A limitation of this study is the unbalanced sample size between the OA school group and NOA school group, whereby the number of OA schoolchildren included in the data analysis was two times the number of NOA schoolchildren. Nevertheless, non-parametric tests and χ 2 tests were used as these tests do not require equal sample size. It is also acknowledged that the selection of schools with generally poor SES and exclusion of schools not accessible by road transportation may create sampling bias, but it is inevitable as most of the rural OA schools have generally poor SES and accessibility by road transportation is crucial to process blood samples timely to ensure the good quality of the samples. Additionally, since our study involved a vulnerable group, the difficulty of collecting blood samples from the children led to insufficient blood volume collected. Therefore, we had to omit them from the data analysis. However, the sample size was still sufficient despite dropouts due to incomplete samples or data collected. In addition to what we have measured (retinol, RBP and ferritin levels) which could be impacted by inflammatory status, both α-1-acid glycoprotein and CRP have also been shown to correct ferritin levels in response to inflammation(Reference Thurnham, Northrop-Clewes and Knowles66,Reference Namaste, Aaron and Varadhan67) . However, the limitation of the current study is that α-1-acid glycoprotein was not measured and therefore adjusted values of the ferritin for inflammation could not be determined otherwise in this aspect. Malnutrition determinants such as dietary pattern, access to food sources, physical activity and infection status are also essential factors in the study. Nevertheless, due to limited resources, these data were not collected in the study, and further research is warranted to enhance the understandings of the current population.
In conclusion, malnutrition is still highly prevalent among primary schoolchildren living in rural areas of Malaysia. The present study revealed that the prevalences of malnutrition, VAD, anaemia, ID, IDA and elevated inflammation were higher among OA schoolchildren than NOA schoolchildren. In addition, higher occurrences of VAD and anaemia were found among children aged below 10 years. Besides, children with undernutrition exhibited lower retinol, RBP, α-carotene, ferritin and Hb. On the other hand, overweight/obesity status was linked to elevated hs-CRP levels. It was also found that children from OA schools and IDA were associated with stunting. The findings of this study provided valuable information for public health authorities to re-evaluate the existing measures in addressing malnutrition among rural primary schoolchildren, especially those from OA schools and younger age groups.
Acknowledgements
We gratefully acknowledge the Ministry of Education (MOE), Ministry of Health (MOH) and Department of Orang Asli Development (JAKOA), head of Orang Asli villages (Tok Batin), school principals and teachers for giving the permission to collect data and samples from the children. We would also like to express our appreciation to the doctors, medical assistances, nurses, MPOB staffs and internship students for their assistance and support throughout the field trips. Most importantly, we thank all the children and their parents for their voluntary participation in this study.
The study was funded by Malaysian Palm Oil Board (RMK-11 Grant- PD219/16). This research was supported (in part) by the Intramural Research Program of the NIH, National Institute of Allergy and Infectious Diseases (NIAID).
K. T. T., R. L., Y. A. L. L., S. C. L. and R. N. performed the conception and design of the study. P. Y. T. and S. N. M. J. conducted the research and data collection. P. Y. T. performed the statistical analysis, data interpretation, drafting and editing of the manuscript. Y. A. L. L., K-T. T., R. L., S. C. L., K. R. S. and R. N. reviewed and edited the manuscript. All authors critically reviewed and approved the final manuscript.
The authors declare that they have no known conflicting interests.
