1. Introduction
Worldwide, childhood malnutrition remains of concern, including micronutrient deficiencies. The global nutrition report indicates that 24% of children under-five are stunted and 8% wasted [
1]. According to the World Health Organization, iron deficiency is the most common and widespread nutrition disorder. Children and women in developing countries are particularly affected [
2].
The UNICEF conceptual framework of malnutrition illustrates the multisectoral nature of nutrition problems [
3]. According to this framework, immediate determinants of nutritional status are inadequate dietary intake and diseases that are determined by underlying factors, namely access to food, care, quality health services and healthy environment in turn, influenced by basic causes acting at the societal level.
Over the past decades, attempts have been made to raise awareness about the multisectoral aspect of malnutrition and of the importance of involving sectors other than health. In fact, as demonstrated by the UNICEF framework, malnutrition is determined by a range of factors relying on the implementation of interventions by a multitude of sectors [
3]. This multisectoral approach has resurged recently, given the rising of more favorable conditions for initiating projects [
4,
5]. Likewise, country assessments of nutrition problems are increasingly involving more and different sectors and a growing body of evidence is available about what works or not for improving nutritional status of populations. At the country level, governments and development partners are looking to determine where the problem lie and its causes to develop appropriate multisectoral strategies and programs to ensure optimal impact on the nutritional status of vulnerable populations [
5].
To our knowledge, the relationship between young child nutritional status and its multiple determinants in that same study has rarely been assessed. In Gabon, using the UNICEF’s conceptual framework of malnutrition, immediate and underling determinants of young child nutrition status have been investigated [
6]. Results have shown that feeding practices and access to improved water and sanitation were the best predictors of length-for-age Z-score (LAZ) explaining 14% of its variance. Recently, in Bangladesh and Vietnam, as part of a development program evaluation, the association between young children’s nutritional status and some of its determinants has been assessed [
7]. The results have shown that household food security or having attended at least four prenatal visits, better caregiver nutrition knowledge and hygiene practices were positively associated to the height-for-age (HAZ) among children under five in both countries.
In Cambodia, as well as in other Asian countries, malnutrition is of concern. Stunting and wasting respectively affect 32% and 10% of Cambodian children 0–59 months old [
8]. Moreover 55% are anemic whereas the prevalence of inadequate zinc intake ranged between from 15% to 25% [
8,
9]. An analysis of the 2005 Cambodia Demographic and Health Survey (DHS) has shown that the consumption of animal-source foods was associated to a reduction of the risk of being stunted and underweight among children 12–59 months old [
10]. Results of our research conducted in Soth Nikum district of Siem Reap province have also shown that the quality of the 6–23 months old child diet is limited, while only a few appeared to be healthy. In particular, data have revealed that energy, iron, and zinc requirements were not fulfilled, while the degree of satisfaction of protein requirements exceeds almost double the needs [
11]. These findings are in accordance with others, which revealed inadequate nutrient intakes (especially iron and zinc) among Cambodian children [
12,
13,
14]. Yet, although these studies shed some light on the situation on immediate determinants of young child nutritional status, it is necessary to extend the investigation to understand the causes of poor dietary intake and health status, as well as the relationship between immediate and underlying determinants and child nutritional status [
3]. Such analysis will help shape future investments to improve the nutrition situation among young children.
Therefore, to get a better understanding on which determinants are mostly associated with child undernutrition in Cambodia, this research aims to investigate immediate and underlying determinants of young child nutritional status among children 6–23 months old living in the Soth Nikum, Siem Reap. The UNICEF conceptual framework of malnutrition is used for this purpose [
3]. This paper complements our previous publication which has shown that nutritional status, dietary intake and health status of young children living in Soth Nikum district were not optimal [
11].
3. Results
Overall, with the exception of the proportion of children born with low birth weight, there was no significant difference among sociodemographic characteristics among all groups (
Table 2).
In our population, the prevalence of stunting and wasting was 19.1% and 8.4% while 76.1% and 37.5% of children had low levels of hemoglobin and ferritin, respectively.
Mean LAZ and WLZ were lower among the children aged 18–23 months old but the hemoglobin value was significantly higher in this group as compared to children 6–11 months and 12–17 months old (
Table 3). Moreover, the mean LAZ was lower among children born with a low birth weight, while the mean WLZ was lesser among those whose mothers had a BMI below 18.5. Children aged from 12 to 17 months, male, and those born with a low birth weight had the lowest mean ferritin values.
Results of bivariate analyses indicate no association between anthropometric Z-scores and indicators on iron status (
Table 4). LAZ was positively correlated to the degree of satisfaction of energy requirements (DSER) but negatively associated with the degree of satisfaction of protein requirements (DSPR) and the presence of intestinal parasites. There was a negative association between WLZ and DSPR, as well as with the health status, as measured by the presence of signs of three illnesses in the past two weeks. Hemoglobin level was positively correlated with the ferritin value, DSER, and DSPR, but negatively correlated with breastfeeding.
The DSER was positively correlated to the DSPR, the degree of satisfaction of iron (DSIR) and of zinc requirements (DSZR) as well as with breastfeeding status (
Table 4). Receiving the minimum acceptable diet was positively correlated with breastfeeding status, the DSPR, DSIR, DSZR, and the degree of satisfaction of vitamin C requirements (DSCR), but negatively with household food security. Access to a healthy environment was not associated with the child’s health status, but it was negatively correlated with access to foods.
Results of the linear regression models indicate that predictors of LAZ were the DSPR and the DSZR as well as being breastfed and having access to improved water source and sanitation (
Table 5). The DSPR and being breastfed were negatively associated with LAZ while the DSZR and having access to a healthy environment were positively associated with LAZ. Being a male and older were both negatively associated with LAZ. These determinants together with child characteristics explained 17.0% of the variance associated with LAZ.
With the exception of DSCR, the degree of satisfaction of energy and nutrient requirements as well as the health status were all predictors of WLZ but the DSPR and health status was negatively associated while there was a positive association between DSER, DSIR, DSZR, and WLZ (
Table 5). All determinants explained 15.5% of the variance associated with WLZ. The DSZR explained around 6% of the variance associated with LAZ and WLZ.
Health status (having presented signs of different illnesses in the past two weeks) was the only predictor of ferritin. Together with being a male and born with a low birth weight, it explained 4.1% of its variance. There was no determinant associated with the hemoglobin level with the exception of being aged 18–23 months (
Table 5).
4. Discussion
Undernutrition among young children remains of public concern in low and middle-income countries such as Cambodia. An understanding of determinants of this situation is essential to make successful investments to ensure optimal child growth and development.
Results of this research show that predictors of the length-for-age z-score in young children are the degree of satisfaction of proteins and zinc requirements, the access to a healthy environment and the breastfeeding status. Health status, the degree of satisfaction of energy, protein, zinc, and iron requirement were the predictors of weight-for-length index. While no firm conclusion can be drawn on the cause-effect relationship between young child nutritional status and its determinants, increasing the degree of satisfaction of energy, iron and zinc requirements and reducing illnesses may help improve WLZ among young children. This result was expected since WLZ reflects short-term conditions. In this context, data on food intake and presence of signs of illnesses were collected at the same time that the WLZ measurement was performed. The presence of illnesses was associated with a reduced WLZ. This can be explained by the reduced appetite caused by the illness, which can affect the child’s weight.
A higher degree of satisfaction of zinc requirements and an improved access to a healthy environment were both positively associated to LAZ. Zinc plays an essential role in many biological processes such as cell growth, and a deficiency in this micronutrient may restrict childhood growth and decrease resistance to infections, which contribute significantly to morbidity and mortality in young children [
34,
35,
36]. It is thus likely that a better satisfaction of zinc requirements would contribute to optimal linear growth and weight gain, and thus, to improved LAZ and WLZ.
With regard to the positive impact of access to healthy environment on LAZ, in spite of the fact that 60% and 80% of households living in rural areas have access to improved water source during the dry and rainy seasons, respectively, still, the access to improved sanitation appears to be limited in this setting. In fact, around 50% of households have no access to improved sanitation facility in rural area [
8]. Therefore, in our study area, even though young children may have access to improved water sources, they may not have access to improved sanitation. However, young children are not necessarily using household theses facilities butthey can be used for child’s stools disposal thus reducing risks of fecal-oral diseases. Our results are in line with a recent Cochrane review which has identified a statistically significant effect of WASH interventions on HAZ [
37]. In Gabon, Blaney et al. have shown that access to improved water and sanitation explained 10% of the variance associated with LAZ among children 0–23 months [
6]. The impact of access to improved water sources and sanitation on child nutritional status may be through the reduction of the risk of diarrhea [
38,
39], environmental enteropathy [
40], and parasite infections [
41].
Our results showed that the DSPR was negatively associated with LAZ and WLZ. As observed in our study, as well as by Reinbott et al. [
42], children were mainly eating rice, which is a source of low quality protein. In fact, as pointed earlier, the DSPR appeared to be above the needs while it was not the case for energy and other nutrients [
11]. This situation has been reported by Millward [
43] who highlights that most diets and especially cereal-based diets provide more than adequate amounts of protein requirements. In addition, as far as protein quality is concerned, it is unclear on how well linear growth can occur in the best circumstances for cereal-based diets with minimal animal source foods given the potential dietary limitations of micronutrients as well as the suboptimal protein quality. Millward [
43] also underlines that deficiencies in some nutrients such as protein, but also of zinc could inhibit growth. As such, there is evidence that protein deficiencies can occur in the diet, especially for populations consuming diets based on starchy roots or cereals with little or no animal-source foods. For instance, in India, Swaminatha et al. [
44] have shown that the largely cereal-based diets among populations, which exhibit high prevalence rates of stunting, appear adequate in terms of protein intakes.
In the Soth Nikum area, a limited diet diversity was most likely a limiting factor of the child nutritional status. An analysis of DHS data from 14 countries including Cambodia, has shown that a higher dietary diversity was strongly associated with a higher HAZ score [
45]. Yet, we did not find any correlation between the number of food groups consumed by the child and LAZ and WLZ. However, with the exception of vitamin C, the degree of satisfaction of nutrient requirements were all strongly (
p < 0.01) and positively associated with the number of food groups consumed in the past 24 h (results not shown).
Being breastfed at the time of the study was negatively associated with LAZ. It is possible that there was an overreliance on breastfeeding among our population, which may have competed with the provision of complementary foods. It is also possible that there was a delay in the introduction of complementary foods among children, which may have led to difficulties to introduce them. In fact, 10.0% of children 6–23 months old part of our study were not yet given soft, semi-solid, or solid foods, while only 15.2% of these had 1–2 meals the previous day. Similar observations were reported from an analysis of data from 19 countries by Caulfield et al. [
46]. Data from nine of the 19 indicate that older still breastfed children were lighter and/or shorter than no longer breastfed children. This situation deserves further research to ensure that quality complementary foods are provided timely and in sufficient quantity along with the continuation of breastfeeding.
In accordance with the findings of this study, results of previous research in Cambodia have shown that the prevalence of stunting increased with age and higher among male children [
8,
45].
The positive association between ferritin and a poor health status may be explained by high ferritin level in presence of acute or chronic infection. It has been reported that the synthesis of ferritin is stimulated by infection. To control partially for that situation, it was suggested that, in addition to ferritin, an independent indicator of the acute phase response, such as CRP to be measured [
20,
47]. CRP was thus assessed in this study and children with elevated levels were excluded (5% of children) from the data analysis. However, we did not control for chronic or sub-clinical infection. As such, we could have added AGP (alpha1-acid-glycoprotein) as it indicates chronic infection and it may better reflect the changes in the concentration of ferritin during infections [
48].
An association between low level of ferritin and being born with a low birth weight has been reported previously [
49,
50]. Authors explained this finding by the fact that ferritin stores of low birth weight children may have been depleted by rapid growth in early infancy, which may have increased the risk of iron deficiency. With regards to hemoglobin, the absence of relationships between the determinants under study may be explained by the fact that levels may also be influenced by other factors such as vitamin B12 intake, genetic abnormalities such as thalassemia and other hemoglobinopathies and by infections. In the Soth Nikum operational district, as mentioned previously, CRP was assessed to check for infections among children and those with elevated CRP were excluded from the data analysis. Given the small proportion of children that were excluded (5%), their removal has not likely induced any bias. Moreover, feces were collected to assess the presence of intestinal parasites and in our regression models, this variable was considered. However, we did not assess the presence of hemoglobinopathies, which may lead to lower hemoglobin level in our group. Nevertheless, available data suggest that hemoglobin E (HbE) and β-thalassemia, is also likely and even common among Cambodian children [
8,
51]. Even though many (32%) have normal hemoglobin, around 25% of them are affected by either heterozygote (24%) or homozygote HbE (3%). An additional 23% have other forms of hemoglobinopathy [
8].
The absence of a relationship between the access to the minimum diet and young child nutritional status may be due to the fact that the current situation may not reflect past practices. Food practices among young children evolve with their age and change significantly in the first two years. As expected, benefiting from the minimum acceptable diet was positively associated with the satisfaction of energy and nutrient requirements. Lastly, the lack of association between household food security and child nutritional status has also been reported elsewhere [
6,
52,
53]. This may be attributed to the fact that the measurement is at the household level and may not reflect access to food for young children who may be privileged by the family with regard to intra-household food distribution, or even benefit from special foods.
In addition to the aforementioned limitations, although significant efforts were dedicated to assess food and nutrient intakes, and in spite of the fact that 24 h recall were included to minimize memory recall bias, estimating young child dietary intake remains a challenge when small amounts of food are consumed. Unlike WLZ, LAZ is the result of past insults related to food intake and illnesses. Therefore, the current situation may not reflect past practices, which may also be a limitation of this analysis, which was cross-sectional and did not capture past practices and events.