Using the CASEN 2017 survey, our results on the SDs of malnutrition in Chilean children under five converge with previous findings and expand the literature’s understanding of this critical public health topic. Our discussion will follow the dimensions proposed in the methodology: demographic, social, and personal child's features.
Regarding demographic variables (gender, age, ethnicity, migrant status, and geographic location), there is evidence that malnutrition in children under five differs in sex by country  . In Chile, measures in children under six indicate that males are more likely to be severely obese, while females are more likely to be overweight and of normal weight . Similarly, male children have a slightly higher percentage when undernutrition categories are grouped ("undernourished" and "underweight"). Our results are in the same direction. Boys have a lower probability of exhibiting normal nutrition than girls; consistently, male children present higher probabilities of under- and overnutrition. It should be noted that this scenario contrasts with adulthood when women have higher levels of overnutrition in Chile . This interplay should search for sociocultural answers such as type of food, physical activity patterns, and gender stereotypes   .
Children younger than 24 months are more likely to suffer from undernutrition in Chile. This condition has been observed elsewhere  and is mainly due to children who stop being breastfed becoming exposed to water, food, and environmental contamination, which occur between the first and second years of life. Consequently, a protective factor in avoiding malnutrition—under and over—is when the child is breastfed for longer than the first six months  . Overnutrition arises mainly between three and five years of age in Chile   and increases as age advances .
Children belonging to native ethnicities are more likely to suffer from both under- and overnutrition. In this vein, previous works have evidenced that native ethnicity becomes a significant SD of undernutrition in children under five in Latin America  . Likewise, research on indigenous Chilean children between six and eight years old has observed a higher prevalence of obesity . However, rather than the ethnicity itself, these findings are explained by the environmental conditions and greater vulnerability of these children's communities   .
To compare migrant status, respondents’ nationality was considered, as we were interested in children's backgrounds (every child born in Chile is of Chilean nationality, which is highly likely in the group investigated). Children from Chilean backgrounds have a significantly higher risk of overnutrition and a lower risk of undernutrition than children of other nationalities. Pavez-Soto et al.  highlighted the challenge of studying migrant children's nutritional conditions, stating that the migrant population in Chile comes mainly from Latin American countries with similar sociocultural conditions. Thus, it would be expected to find high levels of overnutrition, as it is a characteristic seen in this region . Although there are indeed high rates of overnutrition in Chilean children as a whole , our results show this condition attenuates in children from families of nationalities other than Chilean. Similarly, recent works have discovered that migrant children in Chile present a higher biopsychosocial risk, lower access to social programs, and a higher proportion of multidimensional poverty , all factors associated with a higher probability of undernutrition     , which would be compatible with our results.
Regarding children's geographic location, previous studies have found that several regions in other countries display a higher prevalence of malnutrition, either by deficit   or by excess  . In this study, children living in the north zone of Chile (Great and Little North) are less likely to be in the overnutrition category than those living in the Central Zone, while those in the Great North Zone are more likely to be in the undernutrition category than the Central Zone. Likewise, compared to the Central Zone, children living in the South and Austral Zones have a higher probability of overnutrition, and the Austral Zone also has a higher probability of undernutrition. Consequently, our results coincide with national statistics that show overnutrition increases from Chile in a north to south direction , which seems to be related to climate and food patterns linked to geographical characteristics. For example, there is some evidence that the South and Austral Zones' higher levels of overnutrition are due to the low consumption of dairy products, fruits and vegetables, and a high level of sedentarism. Although this information concerns early schoolchildren  , it appears to be applicable to our findings. Similarly, South and Austral Zones exhibit a higher prevalence in native ethnic families, who are especially vulnerable to malnutrition  .
Following geographic location, children residing in urban areas are less likely to belong to the over- and undernutrition categories than children living in rural areas. International evidence consistently reports higher chances of undernutrition in less urban communities  , including Chile . On the other hand, although overnutrition is greatly related to urbanization  , recent Chilean reports with children aged six to eight indicate a higher probability of both normal nutrition and overweightness in urban schools, but a higher probability of obesity in rural schools . Combined, these antecedents indicate that the most severe cases of malnutrition (by deficit and excess) would be in more rural conditions. Thus, since geographic location (zones of the country and rural/urban differences) is one SD in the malnutrition of children under five in Chile, our findings suggest nutritional interventions should be localized and specific as previously indicated .
Second, socio-economic variables are examined. Our findings align with previous evidence and help the literature to better understand what occurs within the country regarding household income. When Quintile V is used as the basis for the analyses (the highest income group), all other quintiles exhibit higher probabilities of under- and overnutrition, steadily increasing toward the poorest quintile, Quintile I. In this regard, robust previous evidence proved that a lower household income is a predictor for undernutrition     . Likewise, it was recently reported that a lower family income is associated with a higher likelihood of overnutrition in Chile . However, there is a change in the trend for Quintile I (the lowest income group), which shows fewer overnutrition odds than Quintiles II and III and a less severe undernutrition likelihood than Quintile II. Furthermore, previous research on Latin America has observed that the probability of overnutrition is more remarkable in families of higher incomes compared to those who are less advantaged  , which specifically occurs between Quintiles II and III compared to Quintile I in Chile. However, contrary to previous evidence, and despite having a higher household income, Quintile II has a higher probability of undernutrition compared to Quintile I. While we do not have a definitive answer for this finding, to some extent it might occur because of targeting by social programs. Access to social programs decreases children's malnutrition   , but they are sharply targeted in Chile at the least advantaged groups . Therefore, this reasoning would encourage the continuation of social programs because of their outcomes, but at the same time, it warns that program targeting might lead to the neglect of other vulnerable groups.
In this same vein, considering household educational level, research consistently observed that a higher educational level is protective against children’s undernutrition   . However, it would not be related to overnutrition. The educational level of either the parents or the household in general—as SD in overnutrition—shows less consistency in previous studies (see, for example,   ). This evidence points out that a higher household educational level carries a higher household income, ensuring the necessary caloric intake. Nevertheless, a higher income is not necessarily related to good food habits, nor does it guarantee a less sedentary lifestyle   .
Our results ascertain that children of families with public health insurance do not have a particular risk of undernutrition compared to children from families with private insurance. Accordingly, previous findings show that health insurance is a protective factor against undernutrition in children under five years old . On the other hand, our results identify a relationship in terms of belonging to the public health insurance and overnutrition groups. We must note that the lowest-income families, which match the profile of the public health system users (called FONASA), are the most vulnerable to both types of malnutrition. Chilean state action has successfully focused on preventing undernutrition with considerable positive outcomes . Nonetheless, the recent nutritional transition has led to a higher prevalence of overnutrition, which has rarely been addressed and so requires more attention  .
Thirdly, we discuss the children's personal characteristics dimension. On the one hand, under- and overnutrition goes hand-in-hand with an increased risk of becoming ill and worsening comorbidities    . Our results indicate that children reporting more significant health problems and health care attendance are more likely to have either under or overnutrition, corroborating the notion that these variables are related. On the other hand, the National Complementary Feeding Program (PNAC) is aimed mainly at families with public health insurance. The PNAC has a basic, single, and identical scheme for all children under six years old (only differentiating by age) but considers a reinforcement for children at undernourished risk. Because children whose families have public health insurance are those with the lowest income, this is likely one factor that underlies these results. Similarly, there is some evidence with other beneficiary groups of complementary feeding programs in Chile (older adults) that have recognized a considerable number of users who do not withdraw or consume these products . Therefore, this issue should be evaluated in the case of PNAC as well. Additionally, it has to be highlighted that the PNAC's technical standards indicate that it is for all children "regardless of their territorial location, nationality, socio-economic status, and social security or migratory status" . This criterion does not expressly include some of the findings in this work—nor in previous research   —pinpointing that variables such as geographic location, ethnicity, and socio-economic status, among others, are SDs of malnutrition risk in children under five.
Finally, we have found that children under five who spend most of their time at home are less likely to suffer from under- and overnutrition. At home, children are primarily cared for by their mother or grandmother in developing countries . Evidence shows that a protective factor for children's nutrition is staying at home under their mother's care or a substitute caregiver for longer . Likewise, our results show that children attending kindergarten are more likely to be undernourished or overnourished. However, it should be noted that there is some evidence with schoolchildren that determines that malnutrition is not directly related to educational facilities themselves but depends mainly on children and families' behavior . Thus, the answer to the increased likelihood of malnutrition of children under five years who attend kindergarten should seek an explanation by incorporating families' social and economic factors and whether there is adequate coordination in food intake between the home and schools.
Among the limitations, this study does not consider interactions between the control variables and their sample weighting. Results discern the likelihood of increasing or decreasing according to the increase in one unit of each variable, taking the other variables as constants into the model. However, as our results are consistent with previous ones, there is convergence validity to our findings.