Artigos

Introduction
Health indicators are proposed to assess biological and socioeconomic risks in communities around the world. They include information on mortality, morbidity, infrastructure and access to health services, and the presence of socioeconomic, environmental and behavioral risk factors. They can be used to identify areas of risk and prioritize health interventions.1
For child health, the most commonly used indicators are: infant and neonatal mortality, proportion of low-birth-weight babies, incidence of infectious diseases, prevalence of nutritional disorders as assessed by anthropometric indices and biochemical measures, and duration of exclusive and complementary breastfeeding. These indicators are essential for monitoring and evaluating children and directing resources to promote healthy child development.2
In 2019, the Early Childhood Friendly Municipal Index (IMAPI) was launched in Brazil to assess the performance of Brazilian municipalities in providing an enabling environment for early childhood development (ECD). This initiative is in line with the concept of nurturing care proposed by the United Nations Children's Fund (UNICEF), which refers to the existence of a stable environment created by parents and other caregivers that ensures children's good health and nutrition, protects them from threats, and provides early learning opportunities through responsive interactions and emotional support.3-6
Despite the robust way in which this index was generated, combining 30 of these indicators from different categories with a careful selection method and clinical and biological plausibility, there are no studies in the scientific literature evaluating its ability to monitor and identify areas at risk of ECD.3,4 In addition, the attempt to create a single indicator to translate a more or less favorable environment for ECD can facilitate the actions of health managers. However, grouping variables together can lead to misinterpretations and biases in estimates due to the variability and potential interactions between them. The incorporation of variables from disparate domains of public health or social determinants has the potential to result in a composite indicator that falls short of its intended representation, either underestimating or overestimating its intended value.7
Therefore, a qualitative and quantitative assessment of the indicators of the IMAPI dimension is essential for the appropriate use of their values when comparing areas or communities, when designing policies and deciding on the allocation of available resources.6
In this context, the aim of this study was to estimate the ability of IMAPI to discriminate between Brazilian municipalities in terms of their ability to promote child development.

Methods
This is an ecological, descriptive and analytical study based on secondary data from official sources. All 5,570 Brazilian municipalities were studied. Only six municipalities were not studied due to lack of information of interest.
In addition, information from the 2010 Municipal Human Development Index (MDHI-2010)8 and data on the schooling of children aged six to 14 (percentage of children in this age group enrolled in the appropriate level of education for their age) and demographic density (number of inhabitants per km²), both from 2010, and gross domestic product (GDP) per capita (total sum of final goods and services produced divided by the number of inhabitants) from 2019 were used. These data were taken from the Atlas of Human Development in Brazil (AtlasBR), prepared by the United Nations Development Program (UNDP), the Institute for Applied Economic Research (IAEP) and the João Pinheiro Foundation (JPF).9 All of this information is publicly and freely available.
IMAPI was created to describe the municipal contexts that are more or less conducive to ECD in Brazil, to support decision making in early childhood and to evaluate the performance of Brazilian municipalities in providing an enabling environment for ECD. It has been developed and revised on the basis of the set of indicators that translate the conceptual model of "Nurturing Care" and includes a set of 30 indicators assigned to four domains: Good Health (14 indicators), Adequate Nutrition (four indicators), Opportunities for Early Learning (seven indicators), Safety and Security (five indicators) and Responsive Care (one indicator).3,4
Although the Care Nutrition conceptual model consists of 5 domains, Responsive Care was not used to calculate the final scores (zero to 100). The calculation of the other domains was based on the analytical weight of the indicators generated according to the SMART method.10 Thus, each indicator was assigned a weight value ranging from 2.71 to 4.24, with the highest weight assigned to three indicators in the Good Health domain (home visits in the first ten days of life, low birth weight and prematurity) and the lowest to one indicator in the Safety and Security domain (air pollution).
Finally, the IMAPI score was calculated as the average of the scores of the four domains, ranging from zero to 100, indicating that the higher the score, the better the municipality performed.3
To represent child development, in the absence of a gold standard indicator, the MHDI was used because it is an indicator of human development of the population at the municipal level, not necessarily child development, but a proxy indicator widely used in the scientific literature. This made it possible to understand the phenomena and dynamics related to the development of the municipality, which served as the sample unit for this study. The MHDI is a measure composed of indicators for three dimensions of human development: Longevity (MHDI), Education (MHDI) and Income (MHRDI) and ranges from 0 to 1. The higher the score, the better the conditions of human development in the municipality.8
The quantitative data available in the databases searched were checked by the researchers for internal consistency before estimates and associations were calculated. Data were extracted from the information sources via electronic files or transcribed into spreadsheets,6,8 with double typing and subsequent validation to correct errors. Furthermore, a specific evaluation of the available information revealed no outliers or missing values. A single database was created with the information collected, allowing the study of associations to be aggregated by Brazilian municipality.
The Kolmogorov-Smirnov test was used to test the normality of the distribution of numerical variables according to the values of their means and standard deviations (x = normal ((x-mean) / standard deviation)), where x corresponds to the variable under test. Specifically, the means, standard deviations, medians, interquartile ranges (p25-p75), and maximum and minimum values of the IMAPI and its domains were described. In addition, the variance and the coefficient of variation were calculated and used as indicators of the ability of each IMAPI domain to discriminate between Brazilian communities.
The correlations of the variables studied were calculated using Spearman's correlation coefficient (?), according to the non-parametric distribution characteristic of the main variables studied (IMAPI and MDHI-2010). Correlations were defined as poor, fair, moderately strong, and very strong for ? values less than 0.3, between 0.3 and less than 0.6, between 0.6 and 0.8, and greater than 0.8, respectively. The data were expressed using a correlation matrix and scatterplots to determine the behavior of Brazilian municipalities with respect to these indicators.11
In addition, a panel of experts in children's health, composed of eight researchers from the study group associated with this research was created to evaluate the suitability of the final indicators that made up the IMAPI to represent the ECD.14 The opinions of these experts were collected individually. The team was constituted by four pediatricians, two nutritionists with expertise in pediatric nutrition, and two experienced epidemiologists.
This panel used the Delphi method, which is based on the principle that predictions made by a structured group of experts are more accurate than those made by unstructured or individual groups.14 Each member, isolated from the direct influence of the others and characterized by anti-protagonism, individually filled out a spreadsheet and classified the 30 indicators used to create the IMAPI as inadequate, debatable, or adequate for promoting ECD. The classifications inadequate, debatable and adequate were assigned values of 0, 1 and 3, respectively. A suitability score was then created for each indicator, ranging from 0 to 24. As a result, indicators with scores above 17 (18 to 24) were considered "suitable" to promote ECD, those with scores between 11 and 17, and those with scores below 11 (zero to ten) were classified as "controversial" and "inappropriate", respectively.
The statistical package used was STATA 14, and significant statistical associations were those with p-values <0.05 (maximum alpha error 5%).
The project was submitted to and approved by the Research Ethics Committee of the Federal University of São Paulo, in accordance with the guidelines of the National Research Ethics Committee (process number: 49072321.4.0000.5505, opinion number: 4.868.854).

Results
The mean IMAPI score in the 5,570 Brazilian communities was 43.79 (SD=32) and the median was 44 (IQR 40-48). The "Initial Learning" domain had the highest mean (68.19; SD=2.62), median (69; IIQ=59-78), and variance (159.33) compared to the other domains and the IMAPI. The distributional characteristics of the IMAPI and its domains are shown in Table 1. The "Safety and Security" domain had the lowest mean (22.52; SD=6.53), the lowest median (22; IIQ=18-26) and a small variance in the values assigned to the municipalities, similar to the IMAPI variance and much smaller than the variances of the other three domains. The distribution of values for each domain is shown in Figure 1.

Figure 1. Histograms of the distribution of the values of the four IMAPI domains assigned to the 5,570 Brazilian municipalities.

Of the 30 indicators that make up the IMAPI, (53.3%) were considered by the expert panel to be suitable for promoting ECD according to the criteria established here. Of the indicators in the "Good Health" domain, 11 out of 14 (78.6%) were considered suitable, and in the "Opportunities for Early Learning" and "Safety and Security" domains, three out of seven (42.9%) and two out of five (40%), respectively, were considered suitable. In addition, none of the four indicators in the "Adequate Nutrition" domain were considered suitable.
On the other hand, five of the 30 indicators (16.7%) were classified as inappropriate. Of these, one was from the Good Health domain, two from the Adequate Nutrition domain and two from the Safety and Security domain. The remaining nine indicators (30%) were considered controversial (Figure 2).

Figure 2. Percentage of aptitude of the 30 IMAPI indicators and their domains according to the panel of experts.

Spearman's correlation coefficients between IMAPI and MHDI scores were 0.494 (p<0.001) (Figure 3), and between IMAPI and the MHDI dimensions of income, longevity, and education were 0.495; 0.513; and 0.426, respectively (Table 2). Figure 3 shows the trend line, correlation and dispersion between IMAPI and MHDI scores.
Among the IMAPI domains and the MHDI dimensions, the positive and moderately strong correlations of the Initial Learning domain with the three MHDI dimensions stand out. However, there are reasonable negative correlations between the Safety and Security domain and the three dimensions of the MHDI (Table 2).
The study of IMAPI and its domains with the other indicators studied showed a moderately strong correlation only between GDPPC and the Initial Learning domain (?=0.667). The other correlations, although statistically significant, were moderate or weak (Table 2).
The intrinsic evaluation of the MHDI with its dimensions resulted in very strong correlations, except for the correlation between the MHDI and the MHME, which was moderately strong (?=0.726). On the other hand, no very strong correlations were found between IMAPI and the domains of Good Health (?=0.614), Adequate Nutrition (?=0.634), and Opportunities for Early Learning (?=0.647). In particular, the correlation between IMAPI and the "Safety and Security" domain was low (?=0.033) and not statistically significant. In addition, this domain showed negative correlations with the Initial Learning domain (?=-0.335) and with the MHDI and its dimensions (Table 2).

Figure 3. Scatter plot and Spearman's correlation coefficient of the IMAPI and MDHI-2010 values for the 5,564 Brazilian municipalities studied.

Discussion
IMAPI represents the first attempt to classify Brazilian municipalities in terms of their capacity to promote ECD according to the "nurturing care" model, using a set of indicators available in the health information systems of the Brazilian Ministry of Health.3,4,15 The development of IMAPI was a Brazilian response to the global call to develop data science initiatives to support municipalities in promoting nurturing care.16-21 Although there are other experiences/proposals to measure the environment for ECD or child health in Brazil22 , none of them have grouped indicators at the municipal level or have been based on the structure proposed by UNICEF (Nurturing Care Framework)5 to generate a single index, as IMAPI has done in 2020.3
The objective of this study has been to evaluate the 30 variables that were included in the final version of the IMAPI, after all selection stages carried out by the team that proposed the indicator. We did not set out to reduce the number of IMAPI variables; however, we analyzed them individually and in groups based on their relevance and effectiveness in contributing to a municipal indicator of child development.
The evaluation of the distribution characteristics of the IMAPI values of the 5,570 Brazilian municipalities showed a discrete variance of the IMAPI and, consequently, a limited statistical ability of this indicator to discriminate municipalities in terms of what it proposes. The distribution of the maximum and minimum values includes only 53% of the scoring possibilities, making the municipalities more similar around the measures of central variation. On the other hand, the isolated evaluation of the domains showed a greater ability for them to discriminate between municipalities, with the exception of the "Safety and Security" domain, which probably contributed the least to the ability of the final IMAPI score to discriminate between municipalities.
There was a reasonable positive correlation between the IMAPI scores and the MHDI and its dimensions, suggesting that better community socio-economic scenarios are more conducive to ECD. The correlation matrix between the IMAPI and its domains and the MHDI and its dimensions shows that there were no very strong correlations. Only the correlations between the "Initial Learning" domain, composed of seven IMAPI indicators (23.3%), and the MHDI and its dimensions were moderately strong12. Despite the fact that these are indices representing general living conditions, these results can be attributed to the fact that IMAPI and MHDI do not use the same indicators in their composition and assign different weights to the respective domains/dimensions. An exception is the composition of the education axis, which is similar between these indices. In addition to the correlation of the Opportunities for Early Learning domain with the MHDI and its dimensions, none of the seven indicators that make up this domain were considered by the expert panel to be inappropriate to represent the ECD. Thus, this domain is a candidate to be the one that is best linked to the IMAPI proposal to differentiate between Brazilian municipalities.
On the other hand, the "Good Health" domain, despite having poor correlations with the MHDI and its dimensions, is composed of 14 of the 30 IMAPI indicators (46.7%), and only one of them was considered inappropriate to represent the ECD by the panel of experts. In other words, the set of indicators that make up this domain seems to be conceptually better suited to express what IMAPI is trying to achieve, even though they are not correlated with the traditional MHDI used in Brazil.
In the process of evaluating indicators proposed by experts in the field, the following criteria are considered essential: (1) the integrity of the data that make up the indicator, in terms of the validity and availability of the information; (2) the consistency of the indicator, in terms of the attributes of the individual, space and time; and (3) the plausibility of the inference of the estimated indicator with other indicators from official data sources. In this sense, the "Safety and Security" domain proved to be the one that least met these criteria, as it had the lowest variance among the four IMAPI domains and limited ability to represent the municipality's capacity to promote ECDs.
In this sense, the domain "Safety and Security" proved to be the one that least met these criteria, as it had the lowest variance among the four IMAPI domains and, according to the panel of experts, limited ability to represent the community's capacity to promote ECD. More importantly, there was a reasonable negative correlation between this domain and the "initial learning" domain, suggesting that the worst results from one of these domains could potentially offset the best results from the other in the composition of the IMAPI. Furthermore, this reasonable negative correlation also occurred between the Safety and Security domain and the MHDI and its dimensions. This finding can be attributed to the composition of this domain, which encompasses five indicators that reflect socio-economic challenges, despite the fact that one of them (air pollution) is frequently observed in regions with higher income levels.3,4,23 This evidence points to the inability of the "Safety and Security" domain to discriminate between municipalities in promoting ECD, and therefore its participation in IMAPI or its indicator structure should be revised to better represent what it is intended for.
At the same time, the correlations between the "Adequate Nutrition" domain and the MHDI and its dimensions were mostly low, and none of the four indicators that make up the domain were considered suitable to represent the capacity of municipalities to promote ECD. Although these correlations are positive and the variance of their values is greater than that of the "Security" domain, the participation of this domain in IMAPI should also be considered with caution. Therefore, a review of its four indicators should be carried out, with the possibility of replacing them with others that better express the characteristics related not only to child nutrition, but also to maternal nutrition. It is noteworthy that maternal nutrition, child malnutrition, and exclusive and total breastfeeding practices are particularly associated with child growth and development. However, the prevalence rates of these indicators were not included in this domain.24-27
It is worth noting that the method used to construct the IMAPI follows well-established criteria that are consistent with the purpose for which it was proposed.3 However, one of the domains was removed from the IMAPI calculation (the responsive care domain), which may have affected the representation of the care-nutrition model. Furthermore, although the IMAPI includes 30 indicators, which makes it more robust in its composition, some of them were not suitable for representing ECD according to the form of assessment carried out in this study. This is probably due to the fact that relevant indicators for assessing child health and development were excluded from the IMAPI because information was not available at the municipal level, which is a quality criterion. In particular, in our analysis, the indicators in the areas of "Adequate Nutrition" and "Safety and Security" are limited to represent ECD, both in terms of their conceptual quality and their ability to discriminate between Brazilian municipalities.
In general, a good indicator must be measurable, feasible, valid, timely, reproducible/sustainable, relevant and understandable. It must also have good quality sources and reflect the phenomenon it is intended to measure/evaluate. Evaluating the quality of indicators should involve the experts responsible for developing them and those who will interpret the data produced. These specialists must be familiar with the procedures needed to monitor trends and the contexts from which the data originate, considering the characteristics, advantages and limitations of the meaning of the information generated.28-32
In this sense, the IMAPI was prepared only with available indicators and with adequate information in the databases consulted by the researchers responsible for its preparation. The lack of municipal information that could influence the ECD, such as the nutritional status of children under five, was not taken into account in the final composition of the IMAPI and its domains.3
This study used the correlation of the IMAPI and its domains with the MHDI and its dimensions, the variance of the resulting scores of municipalities on the IMAPI and its domains, and a conceptual evaluation by a panel of experts in the process of estimating the ability of the IMAPI to discriminate between Brazilian municipalities in terms of their ability to promote child development. This method allowed the IMAPI to be evaluated in three different ways: quantitative discriminatory, qualitative conceptual and inferential with a reference indicator.
Nevertheless, it is possible that this method was not yet the most comprehensive for assessing IMAPI's ability to estimate the ECD. The use of the MHDI and its dimensions as a substitute or reference indicator may not have represented the ideal state of the ECD intended by IMAPI, due to the characteristics of the composition of indicators from its three dimensions (income, longevity and education). While the MHDI does indeed reflect the characteristics of the environments in which children in their first years of life live, it does not make use of data that is directly related to them. In addition, the MHDI used is from 2010, which does not correspond exactly to the period used to select the IMAPI indicators, which may lead to misinterpretations in the associations. Despite these limitations, the MHDI is the most established human development indicator used in Brazil to evaluate municipal disparities and, consequently, to guide the strategies of managers at the three levels of government (federal, state and municipal). In addition, the opinion of the expert panel may have been influenced by the small number and professional background of its members. Although they were specialists in child health, the inclusion of other specialists, such as psychologists, in the team of researchers who made up the Delphi panel would increase the accuracy of its results.
In this context, the results of this study suggest that IMAPI is a conceptually promising indicator in terms of its proposal and intention to represent ECD. Nevertheless, the estimated variances, the negative correlations identified, and the expert opinion indicate the necessity for a reassessment of the significance of some of the selected indicators in terms of their capacity to represent the intended purpose. In addition, the importance and the need for greater rigor in the filling of Brazilian databases on health information is emphasized, so that they can support the construction of robust indicators that adequately discriminate between areas, supporting decision-making and the generation of public policies, avoiding errors of inference and interpretation of results.
Finally, the complexity of developing a comprehensive indicator of ECD highlights the challenges of allocating socioeconomic resources to political strategies and assistance related to the care necessary for promoting health in early childhood. This challenge is made even greater by the fact that ECD is multifactorial and requires comprehensive care for children and their families, involving various public administration sectors related to population health.

Data availability statement

The research data is available upon request to the corresponding author.

References
1. Borges JPA, Santos LA, Simões AL de A. Indicadores de qualidade na assistência à saúde da criança na atenção primária a saúde: uma revisão integrativa. Braz J Hea Rev. 2020;3(3):7101-16.
2. BRASIL. Ministério da Saúde. Secretaria de Atenção à Saúde. Departamento de Ações Programáticas Estratégicas. Política Nacional de Atenção Integral à Saúde da Criança: orientações para implementação. Brasília (DF); 2018.
3. IMAPI. Índice Município Amigo da Primeira Infância. Manual operacional. 52 p. 3.ed. 2020. Avaliable from: https://imapi.org/download/metodologia_manual-operacional.pdf
4. Buccini G, Pedroso J, Stefanie Bertti Coelho, Dias G, Bertoldo J, Sironi A, et al. Nurturing care indicators for the Brazilian Early Childhood Friendly Municipal Index (IMAPI). Matern Child Nutr. 2022;18(S2):e13155.
5. WHO. World Health Organization, United Nations Children’s Fund, World Bank Group. Nurturing care for early childhood development: a framework for helping children survive and thrive to transform health and human potential. Geneva; 2018.
6. Índice Município Amigo da Primeira Infância - IMAPI [Internet]. imapi.org. [cited 2024 Jul 30]. Available from: https://imapi.org/
7. Saisana M, Saltelli A, Tarantola S. Uncertainty and Sensitivity Analysis Techniques as Tools for the Quality Assessment of Composite Indicators. J. R. Statist. Soc. A. 168 (2); 2005. p. 307-323.
8. IDHM Municípios 2010 | United Nations Development Programme [Internet]. UNDP. 2013. Available from: https://www.undp.org/pt/brazil/MDHI-munic%C3%AECDos-2010
9. Ipea, PNUD E FJP lançam Atlas do Desenvolvimento Humano [Internet]. www.ipea.gov.br. [cited 2024 Jul 30]. Available from: https://www.ipea.gov.br/sites/MDHI/18-sites-ipea/manual-editorial/construcao-do-livro/51-fases-do-processo-de-publicacoes-do-ipea
10. CDC Division for Heart Disease and Stroke Prevention. Evaluation Guide. Writing SMART Objectives [Internet]. [cited 2024 Jul 17]. Available from: https://www.cdc.gov/dhdsp/docs/smart_objectives.pdf
11. Goodwin P, Wright G. Decision analysis for management judgment. 5th ed. Chichester, West Sussex, United Kingdom: Wiley; 2014.
12. PNUD - Programa das Nações Unidas para o Desenvolvimento) Relatório sobre o Desenvolvimento Humano. Anuais: 1990-2004. 2013.
13. Chan YH. Biostatistics 104: correlational analysis. Singapore Med J. 2003;44(12):614-9.
14. De Villiers MR, De Villiers PJT, Kent AP. The Delphi technique in health sciences education research. Medical Teacher. 2005;27(7):639-43.
15. Brasil. Ministério da Saúde. DATASUS (Departamento de Informática do SUS) [Internet]. Available from: https://datasus.saude.gov.br/
16. Richter L, Black M, Britto P, Daelmans B, Desmond C, Devercelli A, et al. Early childhood development: an imperative for action and measurement at scale. BMJ Global Health. 2019;4(Suppl 4):e001302.
17. Buccini G, Juliana Gomes Pimentel, Pedroso J, Eugênia S, Bertoldo J, Sironi A, et al. Prioritizing nurturing care at the municipal and district level with the Brazilian Early Childhood Friendly Municipal Index (IMAPI). Matern Child Nutr. 2022;18(S2):e13312.
18. Buccini G, Coelho Kubo SE dos A, Pedroso J, Bertoldo J, Sironi A, Barreto ME, et al. Sociodemographic inequities in nurturing care for early childhood development across Brazilian municipalities. Matern Child Nutr. 2022;18(S2):e13232.
19. Pérez?Escamilla R, Venancio S, Buccini G. Monitoring nurturing care environments for early childhood from the national to the municipal level. Matern Child Nutr. 2022;18(S2):e13327.
20. Pedroso J, Coelho Kubo SE, Silva PO, Ferreira de Castro G, Pimentel JL, Pérez?Escamilla R, et al. Population?based approaches for monitoring the nurturing care environment for early childhood development: A scoping review. Matern Child Nutr. 2022;18(S2):e13276.
21. Thrive Nurturing Care For Early Childhood Development.Country profiles for early childhood development [Internet]. [cited 2024 Jul 30]. Available from: https://nurturing-care.org/wp-content/uploads/2023/10/ECD_Countdown2030_Global.pdf
22. SEADE. Fundação Sistema Estadual de Análise de Dados. Governo do Estado de São Paulo. O que e? o I?ndice Paulista da Primeira Infa?ncia?. [Internet]. [cited 2024 Jul 30]. Available from: http://www.ippih.seade.gov.br/frontend/#/
23. Habermann M, Souza M, Prado R, Gouveia N. Socioeconomic inequalities and exposure to traffic-related air pollution in the city of São Paulo, Brazil. Cad Saude Publica. 2014;30(1):119-25.
24. Abouzahr C, Mikkelsen L, Rampatige R, Lopez A. Mortality statistics: a tool to enchance understanding and improve quality. Pac Health Dialog. 2012;18(1):247-70.
25. Boccolini CS, Carvalho ML, Oliveira MIC. Factors associated with exclusive breastfeeding in the first six months of life in Brazil: a systematic review. Rev Saude Publica. 2015;49:91.
26. Brasil. Ministério da saúde. Saúde da criança: aleitamento materno e alimentação complementar. Editora MS, Brasília, 2015. [Internet]. [cited 2024 Jul 30]. Available from: https://www.gov.br/saude/pt-br/assuntos/saude-de-a-a-z/s/saude-da-crianca/publicacoes/saude-da-crianca-aleitamento-materno-e-alimentacao-complementar/view
27. Boccolini CS, Boccolini PDMM, Monteiro FR, Venâncio SI, Giugliani ERJ. Breastfeeding indicators trends in Brazil for three decades. Rev Saude Publica. 2017;51:108.
28. OPAS. Organização Pan-americana da Saúde. Indicadores de Saúde: Elementos conceituais e práticos. Washington, DC: OPAS, 2018.
29. Weber E, Downward GS, Ebi KL, Lucas PL, Detlef van Vuuren. The use of environmental scenarios to project future health effects: a scoping review. Lancet Planet Health. 2023;7(7):611-21.
30. Li C, Cheng G, Sha T, Cheng W, Yan Y. The Relationships between Screen Use and Health Indicators among Infants, Toddlers, and Preschoolers: A Meta-Analysis and Systematic Review. Int J Environ Res Public Health. 2020;17(19):7324.
31. De Moraes FF, De Magalhaes Camara V, Froes Asmus CIR. Development of Environmental Health Indicators for the Child Population: Report on a Brazilian Experience. Ann Glob Health. 2017;83(2):227-33.
32. Carneiro VCCB, Ribeiro de Oliveira P de T, Rassy Carneiro S, Cardoso Maciel M, Pedroso J da S. Impact of expansion of primary care in child health: a population-based panel study in municipalities in the Brazilian Amazon. BMJ Open. 2022;12(3):e048897.