Nutritional status and effective verbal communication in Peruvian children: A secondary analysis of the 2019 Demographic and Health Survey

Background To estimate the impact of stunting on the development of effective verbal communication (EVC) in children aged 24 to 36 months. Methods We conducted a retrospective, quasi-experimental study design using data from 4452 children between 24 and 36 months of age available in the Early Childhood Development (ECD) section of the Demographic and Family Health Survey (ENDES) 2019 survey. Achievement of EVC was considered as the dependent variable. After propensity score matching analysis, a total 601 children were included in the exposed (children with stunting) and 3848 in the unexposed group (children without stunting). The psmatch2 Stata software command was used to estimate the impact of stunting on EVC with a level of 5% for significance. Results The EVC indicator was achieved by 57.4% of the children between 24 and 36 months of age evaluated, while the prevalence of stunting in this population was 14%. The evaluation of impact showed that the group exposed to stunting was 8% less likely to show EVC compared to the unexposed group (ATT -0.08; 95% CI -0.106 to -0.054; p<0.001). Conclusions The presence of stunting was found to have a significant impact on the propensity to show EVC in Peruvian children between 24 and 36 months of age. Strengthening of strategies for reducing malnutrition in vulnerable areas, as well as those directly affecting EVC is a priority for diminishing gaps in the achievement of this indicator in our country.

General guidance is provided below.
Consult the submission guidelines for detailed instructions.Make sure that all information entered here is included in the Methods section of the manuscript.
The study did not require the approval of an ethics committee because it is an analysis of secondary data that is in the public domain and does not allow the identification of the participants evaluated.Yes -all data are fully available without restriction

Introduction
Child malnutrition produces serious metabolic and structural consequences for the individual and has a high social and economic impact [1], mainly in low-and middle-income countries, where nearly half of the children under five years of age die due to malnutrition [2].The chronicity of this condition affects a large proportion of children under five throughout the world, with an estimated 23% (144 million) of people in this age group presenting stunting in 2019 [3].In 2011, low-and middle-income countries reported the highest proportions of malnutrition with 28% of stunting in children, in comparison to the 7.2% reported in high-income countries [4].Also in 2017, Peru reached a 12.9% of stunting in children under five years of age, which is under the average prevalence of 25% for that outcome in developing countries the same year [5].This becomes relevant because children integral development at an early stage (perceptual, motor, cognitive, language, socio-emotional and self-regulation skills), is highly dependent of an active and healthy life, which cannot be well achieved in the presence of malnutrition [6].
Early childhood development (ECD) is one of the goals of the Sustainable Development Goals in developing countries [7].However, in these countries, nearly 150 million children under five years of age are at risk of failing to reach their potential development due to malnutrition and extreme poverty [8].ECD is essential to achieve adequate learning, occupational productivity, physical and psychological health, and social well-being [9].In Peru, the government, led by the Ministry of Development and Social Inclusion, implemented a program aimed at improving ECD.This program included seven areas of development corresponding to a healthy birth, safe attachment, adequate nutritional status, effective verbal communication, walking alone, regulation of emotions and behaviors and symbolic function [8,10].
Effective verbal communication (EVC) begins to develop at an early age, through crying or other forms of rudimentary communication as an immediate mechanism of interrelationship between children and parents.This communication influences adolescent development in the academic stage and work productivity as well as the development of emotions in adult life [11].To achieve adequate EVC, the nutrition of a child from the prenatal period through the first 24 months of life must be adequate because the risk of stunting during this period in developing countries is high, leading to consequences in cognitive development [11].Chronic malnutrition also creates learning difficulties in the first one thousand days of life and has an impact on EVC in the long term [12,13].Therefore, in order to achieve the proper development of EVC, adequate nutrition and eating habits of the child must be ensured (during the first months of life) and those of the mother must be improved.Moreover, safe attachment of children and their parents is necessary as well as stimulation of motor developmental and gesticulation [14][15][16].
In Peru, there are no population-based studies on the impact of nutritional status on the development of EVC during childhood.However, the National Institute of Statistics and Informatics (INEI in Spanish) collects nutritional information on children under five years of age through the Demographic and Family Health Survey (ENDES in Spanish).During a recent validation of the ECD instrument the ENDES also included information on seven components of ECD including EVC [17].Therefore, the objective of the present study was to estimate the effect of chronic malnutrition in children aged 24 to 36 months on the development of EVC.

Design and study population
The ENDES is a survey representative of the Peruvian population.In a secondary analysis of the ENDES 2019, the sample was characterized by being bi-stage, probabilistic of a balanced, stratified and independent type, at the departmental level and by urban and rural areas.The sample size of ENDES 2019 was 36,760 homes, 9320 of which were in urban areas and 12,660 were from rural areas.
The study had a retrospective, quasi-experimental design and included 4452 children between 24 and 36 months of age.The children were analyzed in the ECD section of the ENDES 2019, which evaluates different aspects of child development including EVC.To estimate the impact of chronic child malnutrition on EVC the sample was divided into exposed (children with chronic malnutrition) and unexposed children (without chronic malnutrition).

Variables
The World Health Organization (WHO) classification was used for the exposure variable "malnutrition" (HW70) [18] and EVC in children aged between 24 to 36 months (R4).The latter variable, measures EVC behaviors at the comprehensive and expressive level for each evolutionary stage of children aged 24 to 36 months according to a psychological instrument (Battelle Development Instrument) previously validated by experts who agreed on the questions to be included by the calculation of an Aiken indicator V greater than 0.7 (85.2%).This instrument was previously used in a pilot test in four Peruvian cities to determine adequate functioning and is based on questions that are performed and answered by the mothers [19].
The study covariates (codes in parentheses) included: age of the child in months (QI478), age of the mother (V012), years of education of the mother (V133), hemoglobin level of children (HW53), sex of the child (B4), total householders (HV009), health insurance coverage (V481), total children ever born (V201) and cesarean delivery (M17).In addition, the following variables: were constructed: "quintile" of the wealth index (V190) categorized into five levels (poorest, poor, middle, rich, richest), "prenatal control" according to prenatal visits (inadequate [less than 8] and suitable [8 or more]) during pregnancy (M14), "altitude" (m.a. s. l.) of the place of residence (HV040), "area" classifying the type of place of residence as urban or rural (V025), "ethnicity" (V131) considering ethnic origin (native and nonnative), "natural region" including the natural region of origin (coast, highlands and jungle), "immediate lactation" considered as whether the infant was breastfed within the first hour after birth or not (V426), and "breastfeeding in months" according to the number of months the child was breastfed (M5).The covariates were selected according to revision of the literature available [20][21][22][23].

Statistical analysis
The model proposed by Roy-Rubin [24,25], was followed, defined as an exposure variable to chronic malnutrition (Di) and as an EVC outcome variable in children between 24 and 36 months (  ).
That is,   (1) is the result variable if the individual i was exposed and   (0) is the result variable if the individual i was not exposed.Thus, the effect of exposure (  ) for an individual i can be calculated as: The analysis focused on determining the average impact of exposure (Average Treatment In this case, the average impact of exposure in exposed subjects (  ) is the expected value of the difference in the result variable in the group of individuals exposed when there is the presence of exposure [  (1)|  = 1] and the expected value of the result variable in the exposure group in the absence of exposure, known as contrafactual [  (0)|  = 1].That is, the estimated impact is the difference in the average of the result variable of the exposure group and the control group.
The estimated impact is composed of the effect of exposure and selection bias.Formally, it is assumed that the selection bias is due only to differences in observable characteristics (conditional independence condition (IC)).What this assumption makes is to ensure that the selection bias is equal to zero, generating an unbiased estimate of the true effect of exposure.This is achieved by finding an individual exposed to malnutrition with the same observable characteristics in the control group.
An individual is an adequate control of an individual exposed to malnutrition, if both individuals are similarly likely to be exposed to malnutrition.This implies that the propensity score matching (PSM) method proposed by Rosenbaum Rubin (1983) can only be calculated in the Common Support Region (CS) to ensure that the exposure and control groups are very similar.
The PSM estimator is the average difference of the result variables of the individuals in the exposure group and the control group in the common support with an appropriate probability of exposure participation.

Propensity and match estimation
A logit model estimated the exposure probabilities for the covariates, i.e. the probability of participation with the observable characteristics of individuals.Having the probability of exposure and with the intention to compare the exposure impact results, a matching procedure was implemented with the kernel estimator that matches each respondent in the exposure group with a weighted average of all respondents in the control group.In addition, a matching procedure was estimated with the nearest neighbor, without replacement, using the estimated exposure probabilities, implementing a 1:1 correspondence structure.To improve the quality of the matching the condition that the maximum absolute difference between the exposure probabilities of matched respondents was not greater than 0.1 was imposed.Finally, an IPWRA matching procedure was implemented using weighted regression coefficients to calculate the impact of exposure, where weights are the estimated inverse probabilities of exposure.

Post-match balancing diagnosis
To ensure the balance of coincidence matching in the probability of exposure between the specified covariate distributions of the two exposure groups, two assumptions were required for correct estimation of the impact of exposure: a) the difference in the average of the exposure variable of the exposure and control group is composed of the effect of chronic malnutrition and the selection bias, and b) any combination of characteristics observed in the exposure group also exists in the control group, ensuring that the exposure and control groups are very similar or are within a common support.
Compliance with these assumptions is assumed for the estimation of the PSM.
Post-matching, the covariate distributions between the exposure groups in the matching sample were evaluated using the standardized absolute mean differences (bias) and variance ratios.After correspondence, standardized differences in the means of the covariates are expected to be close to zero, and variance ratios are expected to be close to one, if there is an adequate balance of covariates.

Estimating the impact of chronic malnutrition on outcomes
The impact of chronic malnutrition on children between 24 and 36 months of age was estimated by comparing the results between those exposed and unexposed in a subsample of individuals with similar characteristics.To do this, the psmatch2 command was used.In the case of the PSM estimate per Kernel and the PSM estimate per nearest neighbor, the teffectsipwra command of Stata was also used for weighted effect estimates with adjusted regression, considering that all impact estimates were evaluated at the 5% significance level.

Sensitivity analysis
A sensitivity analysis was performed on the effect estimates using the limit approach of Rosenbaum.This approach explains that if there were the influence of unmeasured covariates, then a match between the two groups of covariates would not have the same probability of exposure allocation, and their probability ratio (Γ) would be different from one.By increasing the value of Γ to identify the probabilities of exposure allocation (Γ>1), the degree of influence that an unmeasured covariate must have on the impact allocation and the validity of a study can be observed.Sensitivity analyses were performed for different values of Γ>1, in increments of 0.05, to determine the extent to which the Γ value remained at a significance level of 0.05.The inferences are then considered sensitive to bias caused by non-measured covariates if the Γ values are closer to 1 by altering the results (>0.05), and are considered relatively robust if values greater than Γ are required to obtain results that affect inference.
The mhbounds command of Stata was used for this sensitivity analysis [24].

Ethical considerations
The study did not require the approval of an ethics committee because it is an analysis of secondary data that is in the public domain and does not allow the identification of the participants evaluated.

Results
Table 1 shows the characteristics of the 4452 children included, of which 604 (13.6%) were exposed to chronic malnutrition (exposure group) and 3848 (86.4%) were not exposed (control group).
The mean current age of the mother was 30 years, while the mean lactation period of the child was 19 months, and the mean hemoglobin level of the child was 12.2 g/dl.In addition, the highest proportion in the wealth index was the poor (28.3%) and very poor quintiles (26.1%); 27.8% were from rural areas, 42.2% were from the geographical area of the Coast; 82.6% had health insurance; 57.4% of the children 199 evaluated reached the EVC indicator.200 Table 2 shows the distribution and comparison of covariates between the exposed and 203 unexposed groups in matched and unmatched samples, of which four covariates had biases greater 204 than +/-0.75 after matching, while 10 covariates had differences +/-0.25 in lower matching biases.

205
Overall, bias of the model decreased from 31.1 to 2.0 after matching.Among exposed children, the 206 altitude of the conglomerate, hemoglobin levels, live births and ethnicity were positively associated with 207 exposure (0.01 -0.03), while the current age of the mother, months of lactation, higher wealth quintile 208 and residents in the jungle region (0.80-0.98) were less likely to be exposed.As a result of the matching technique, all children found their unexposed counterpart, and the coincidence eliminated three observations.After matching, all covariates showed standardized means close to zero.The propensity score was estimated for each of the observations, and the balance was diagnosed, verifying the quality of the matching and imposing the common support.

Estimation of exposure impact
The estimate of the impact of chronic malnutrition is presented in Table 3.By way of comparison, the results of three different PSM estimators are presented.The estimated PSM Kernel shows the impact of chronic malnutrition on children between 24 and 36 months of age, which indicates that the group exposed to chronic malnutrition were 8% less likely to show EVC compared to the unexposed group, with a 95% confidence interval of -0.106 to -0.054 (95% CI).The nearest perneighbor estimator (noreplacement) and weighting with adjusted regression (IPWRA) showed similar results of 9% (IC -0.144 to -0.034) and 6.3% (IC -0.119 to -0.005), respectively.After matching, the covariates in the two groups the mean standardized biases decreased from 31.Chronic malnutrition continues to be a problem in Peruvian children, with 14% of children between 24 and 36 months suffering from this condition in 2019.This prevalence is above the 9.6% of chronic malnutrition reported in Latin America and the global prevalence of 13.5% in children under five years old in 2017 [29,30].In addition, it has been described that in the first half of 2019, 33.2% of Peruvian children aged between 24 to 36 months were at risk of chronic malnutrition This study has limitations.The use of the secondary database of ENDES for analysis may introduce information biases by the respondents when providing the requested information as these may have provided inaccurate or false information.In addition, this bias may be present due to erroneous introduction of data at the time of data registration by the interviewers.Moreover, since the sample weights of the survey were not included in our analyses, the results cannot be generalized to the ENDES target population.Despite these limitations, it should be noted that ENDES follows the standardized design and procedures of the Demographic and Health Survey (DHS) program, which includes training of interviewers to reduce the introduction of biases during information collection in order to ensure the quality of the data collected.Similarly, a PSM analysis was used, providing a better balance of covariate measures in groups of children exposed and unexposed to chronic malnutrition with the aim of reducing the risk of confusion bias.This study is also one of the first national reports assessing the association between chronic malnutrition and an objective outcome of the ECD, such as EVC, with representative data.
In conclusion, the presence of chronic malnutrition has a significant negative impact on the propensity to develop EVC in Peruvian children between 24 to 36 months of age.Strategies for reducing malnutrition should be strengthened in population areas identified as vulnerable and should prioritize the components of ECD according to age groups for which their measurement is of interest to complement nutritional strategies with those that directly concern EVC.This would reduce the gaps in EVC among children in Peru with the aim of ensuring adequate ECD and equal opportunities for all Peruvian children.
Enter a financial disclosure statement that describes the sources of funding for the work included in this submission.Review the submission guidelines for detailed requirements.View published research articles from PLOS ONE for specific examples.This statement is required for submission and will appear in the published article if the submission is accepted.Please make sure it is accurate.Unfunded studies Enter: The author(s) received no specific funding for this work.Funded studies Enter a statement with the following details: Initials of the authors who received each award • Grant numbers awarded to each author • The full name of each funder • URL of each funder website • Did the sponsors or funders play any role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript?• NO -Include this sentence at the end of your statement: The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.• YES -Specify the role(s) played.• * typeset Competing Interests Use the instructions below to enter a competing interest statement for this submission.On behalf of all authors, disclose any competing interests that could be perceived to bias this work-acknowledging all financial support and any other relevant financial or nonfinancial competing interests.This statement will appear in the published article if the submission is accepted.Please make sure it is accurate.View published research articles The authors have declared that no competing interests exist.Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation from PLOS ONE for specific examples.NO authors have competing interestsEnter: The authors have declared that no competing interests exist.Authors with competing interestsEnter competing interest details beginning with this statement: I have read the journal's policy and the authors of this manuscript have the following competing interests: [insert competing interests here] /A" if the submission does not require an ethics statement.

Format
for specific study types Human Subject Research (involving human participants and/or tissue) Give the name of the institutional review board or ethics committee that approved the study • Include the approval number and/or a statement indicating approval of this research • Indicate the form of consent obtained (written/oral) or the reason that consent was not obtained (e.g. the data were analyzed anonymously) • Animal Research (involving vertebrate animals, embryos or tissues) Provide the name of the Institutional Animal Care and Use Committee (IACUC) or other relevant ethics board that reviewed the study protocol, and indicate whether they approved this research or granted a formal waiver of ethical approval • Include an approval number if one was obtained • If the study involved non-human primates, add additional details about animal welfare and steps taken to ameliorate suffering • If anesthesia, euthanasia, or any kind of animal sacrifice is part of the study, include briefly which substances and/or methods were applied • Field Research Include the following details if this study involves the collection of plant, animal, or other materials from a natural setting: Field permit number • Name of the institution or relevant body that granted permission • Data Availability Authors are required to make all data underlying the findings described fully available, without restriction, and from the time of publication.PLOS allows rare exceptions to address legal and ethical concerns.See the PLOS Data Policy and FAQ for detailed information.
Fig 1 shows the assessment of the assumption of overlap.
1 to 2.0 (Fig 2), of which 15 observed variables were in the range of [-0.05,+0.05]after matching, and 5 observed variables showed a higher matching to the aforementioned range.The matching bias in all covariates was less than [-0.98;+ 0.81].
[31].Although our country has shown a marked reduction in the prevalence of chronic child malnutrition from 30% in 2005 to 15% in 2015[32], these results show that efforts are still needed to reduce the prevalence of this condition due to the absence of substantial reductions between 2015 and 2019, and the considerable number of children who remain at risk of chronic malnutrition.Likewise, attention must be paid to the Peruvian Andean region, where according to information from 2018 the lowest proportion of children meet the EVC indicator[28]., with a prevalence of chronic malnutrition of up to 30.1% and a risk of chronic malnutrition in children under five years old of 58.2%[31].
Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation

Table 3 . Exposure effects of chronic malnutritrion on EVC.
), children residing in the jungle vs. the coast and mountain areas (53% vs. 51.9% and 47%, respectively), children of mothers with higher vs. secondary and without education (53% vs. 49.3 and 45%, respectively), boys of the intermediate vs. lower economic quintile (52.3% vs. 46.2%,respectively)andgirls vs. boys (54% vs. 45.6%,respectively)[28].These results demonstrate the presence of gaps according to the sociodemographic characteristics of Peruvian children which must be taken into account by the government in order to target strategies to ensure that a higher proportion of children achieve EVC indicators and to reduce the inequalities in achieving this indicator.EVC is one of the seven components of ECD and its measurement is of interest between 9 and 36 months of age.Indeed, in order to facilitate the work of decision-makers, national studies and technical reports should report chronic malnutrition according to age groups and not aggregated for all children under five years of age (60 months).This would enable focusing on nutritional and complementary strategies for the development of each of the components of the ECD according to specific age groups with potential vulnerability.
Figure 2. Covariates balance before and after matching.respectively