Nurturing care indicators for the Brazilian Early Childhood Friendly Municipal Index (IMAPI)

Abstract The Nurturing Care Framework (NCF) calls for establishing a global monitoring and accountability systems for early childhood development (ECD). Major gaps to build low‐cost and large‐scale ECD monitoring systems at the local level remain. In this manuscript, we describe the process of selecting nurturing care indicators at the municipal level from existing routine information systems to develop the Brazilian Early Childhood Friendly Index (IMAPI). Three methodological steps developed through a participatory decision‐making process were followed. First, a literature review identified potential indicators to translate the NCF domains. Four technical panels composed of stakeholders from federal, state and municipal levels were consulted to identify data sources, their availability at the municipal level and the strengths and weakness of each potential indicator. Second, national and international ECD experts participated in two surveys to score, following a SMART approach, the expected performance of each nurturing care indicator. This information was used to develop analytical weights for each indicator. Third, informed by strengths and weaknesses pointed out in the previous steps, the IMAPI team reached consensus on 31 nurturing care indicators across the five NCF domains (Good health [n = 14], Adequate nutrition [4], Responsive caregiving [1], Opportunities for early learning [7] and Security and safety [4]). IMAPI represents the first attempt to select nurturing care indicators at the municipal level using data from existing routine information systems.

panels composed of stakeholders from federal, state and municipal levels were consulted to identify data sources, their availability at the municipal level and the strengths and weakness of each potential indicator. Second, national and international ECD experts participated in two surveys to score, following a SMART approach, the expected performance of each nurturing care indicator. This information was used to develop analytical weights for each indicator. Third, informed by strengths and weaknesses pointed out in the previous steps, the IMAPI [7] and Security and safety [4]). IMAPI represents the first attempt to select nurturing care indicators at the municipal level using data from existing routine information systems.

K E Y W O R D S
Brazil, child development, cities, index, indicator, monitoring, nurturing care

| INTRODUCTION
Optimal early childhood development (ECD) follows a gradual process that is contingent on children and their families having access to nurturing care environments needed by children to reach their full motor, language, socio-emotional and cognitive development (Black et al., 2017;Shonkoff et al., 2012). Nurturing care is defined as an environment that is responsive, emotionally supportive, In this context, Brazil, which is the largest and the most populous country in Latin America (with 211 million inhabitants), provides an important opportunity for testing a municipal-level ECD index because the operationalization of ECD programmes occurs across 5570 municipalities in the context of great socio-economic and environmental disparities (Aristides dos Santos et al., 2019). Hence, translating and operationalizing the NCF into integrated ECD strategies through effective actions according to the needs and specific determinants of each community is a major challenge. Furthermore, over the past 33 years, three legal frameworks aimed at guaranteeing the rights of children have been created-the Federal Constitution of 1988, the 1990 Child and Adolescent Statute and the 2016 Legal Framework for Early Childhood. As a result, in recent years, the ECD agenda in the country has expanded greatly, and investments made in programmes focused on ECD, such as Criança Feliz ('Happy Child'), a home visiting programme that has already reached about 3000 Brazilian municipalities Girade, 2018). Furthermore, Brazil has a large volume of high-quality public information databases on health, education and social development (including cash transfer programmes); however, data integration for nurturing care monitoring purposes does not exist at any level of government. Thus, we aimed to describe the process of identifying and selecting nurturing care indicators at the municipal level from existing Brazilian databases to develop the Brazilian Early Childhood Friendly Municipal Index (IMAPI, 'Índice Município Amigo da Primeira Infância').

| METHODS
IMAPI was developed following a systematic methodology ( Figure 1).
In this manuscript, we describe the participatory decision-making process used in the first three methodological steps to identify and select nurturing care indicators at the municipal level. The name of indicators is reported as 'indicator name' throughout.

| Participatory decision-making process
A multisectoral and participatory decision-making process was designed to facilitate group communication and to reach consensus on the nurturing care indicators in Brazil. Consultative participatory methods have been used to enhance transparency, accountability, equity and efficiency in the decision-making process, which are adequate to identify and select nurturing care indicators (Devente et al., 2016;Elwyn et al., 2017;Okoli & Pawlowski, 2004). In this study, the participatory decision-making process included consultations with participants within each of the three methodological steps ( Figure 2).

Key messages
• The Nurturing Care Framework (NCF) calls for establishing monitoring systems for early childhood development (ECD) to achieve the 2030 Sustainable Development Goals.
• A three-step participatory methodological process was designed and piloted to select municipal-level nurturing care indicators using a routine information system.
• Brazil is an important country to identify municipal-level nurturing care indicators because the implementation of ECD programmes happens across its 5570 municipalities in the context of great inequities.
• The Brazilian Early Childhood Friendly Municipal Index (IMAPI) represents the first attempt to select nurturing care indicators at the municipal level using existing routine information systems.

| Participants selection and characteristics
Technical stakeholders, experts and the IMAPI team engaged in the different steps of the participatory decision-making process ( Figure 2). Detail on participants' institutions is described in Table S1.

In
Step 1, technical panels were convened to identify nurturing care indicators as well as routine information systems available at the municipal level. A formal invitation describing the technical panel's goals and characteristics of individuals who could serve as a participant was sent to government institutions (n = 15) as well as nongovernment institutions (n = 4). Each institution could determine and recommend one or more participants based on the technical panels' goals. Although the number of participants varied across technical panels, all invited institutions sent at least one participant. Technical panel participants were individuals that managed or had in-depth knowledge of national databases with information about health, education and social development in Brazil. In Step 2, expert surveys were designed to validate the selection of municipal-level nurturing care indicators and score each of the SMART attributes to inform an evidence-based analytical weight per indicator. Invited experts were identified within the Brazilian and international institutions and networks highly engaged with ECD programming in Brazil. Expert surveys' participants included individuals with experience in ECD and public policy, who had in-depth knowledge of the socio-economic, geographical and political context of Brazil. Fourteen experts were invited by email, and 12 accepted the invitation to participate. Of those, 11 experts participated in the first survey, and seven in the second survey. In Steps 1-3, the IMAPI team engaged throughout the participatory decision-making process. Each of the four rounds started with an input statement, which consisted of a list of nurturing indicators named 'ECD Indicators Versions 1-4' produced by the IMAPI team. At the end of each round, this list of nurturing care indicators was revised by the IMAPI team to summarize modifications and consensus within that round, which generated an output statement, which, in turn, became the input statement for the next round.
The IMAPI team was composed of the principal investigators, research assistants and senior advisors who collectively had expertise in epidemiology, maternal-child nutrition, implementation science and data science and machine learning.
Step 1. Conceptual model This step aimed to use the NCF conceptual framework to identify potential nurturing care indicators at the municipal level. Initially, the first author generated an input statement with an initial list of indicators (ECD Indicators Version 1) based on a careful review of documents and technical materials measuring actions and policies for early childhood, such as Countdown 2030, World Bank, World Health Organization, NCF website and São Paulo Early Childhood Index (IPPI) (Naudeau et al., 2011;SEADE, 2017;UNICEF and&Countdown to 2030, 2019;World Health Organization, United Nations Children's Fund, & World Bank Group, 2018). At this stage, the goal was to collect the largest number of potential indicators; thus, we included indicators that contributed to any of one of the five domains of the NCF, operationalized as follows: Good health (indicators related to healthcare from prenatal to the first years of life), Adequate nutrition (indicators related to the promotion of healthy eating, access to food and child nutrition), Responsive caregiving (indicators related to family skills and child care in the home environment), Opportunities for early learning (indicators related to the access and quality of formal education) and Security and safety (indicators related to protections or vulnerabilities to which the child may be exposed in the family or community environment).
Indicators considered important for placing the NCF into the diverse municipal contexts were allocated to the category of 'additional indicators' (e.g. population size and the region of the country), which also underwent revisions, additions or removal following the same methods followed for the selection of the nurturing care indicators. Second, in Round 1, the IMAPI team discussed the 'ECD Step 2. Analytical weight This step aimed to estimate an analytical weight for each municipallevel nurturing care indicator following a SMART process. Because the NCF assumes that all domains have the same level of importance, the purpose of the analytical weighting exercise was to improve reliability when generating IMAPI indexes by giving higher performance analytical weight to indicators with better data quality attributes (Köhler, 2016;OECD, 2008). Additional indicators were not included in the calculation of IMAPI indexes; thus, they did not receive scores for analytical weight. Round 3 was composed of two surveys for experts following a SMART approach (Centers for Disease Control, 2011). Surveys were set up into an Excel spreadsheet using the indicators listed in the 'ECD indicators Version 3' and were sent by email to experts. A detailed methodological note on how questions were asked and analysed can be found in Table S3 Table 1.
Step 3. IMAPI indicators indicator within each NCF domain was either included, relabelled to clarify the construct it was intended to represent or excluded due to the non-availability of the database after multiple attempts.
Importantly, data access and extraction started early in the decisionmaking process, that is, since the first technical panel, allowing various attempts to request the database over the project's timeframe. During data extraction, data quality was scored by the IMAPI team through the SMART surveys following the four quality attributes for each indicator (periodicity of data, data source, type of access to data and population profile). Although data quality did not drive the decision-making process regarding inclusion/exclusion of an indicator, it did drive how the IMAPI team defined the calculation method, that is, ways to aggregate data to improve quality.

| RESULTS
Step 1. Round 1 started with 67 potential indicators identified through document review and finalized with 48 indicators allocated to NCF domains. Indicators were excluded if they did not discriminate among Brazilian municipalities. For example, in Brazil access to HIV treatment is free and universal; hence, this indicator is homogeneous across municipalities. In Round 2, after four technical panels, a total of 35 indicators were preselected across the NCF domains-Good health (n = 16), Adequate nutrition (n = 3), Responsive caregiving (n = 6), Opportunities for early learning (n = 4) and Security and safety (n = 6) ( Figure 2).
Step 2. In Round 3, experts reclassified three indicators into another NCF domain (criterion S, Indicator Allocation). Greater importance was given to the attribute population profile (i.e. population representativeness) (mean = 4.55), followed by access to data (4.45), periodicity (4.20) and data source  (Table 1).
Step 3.  and neonatal health (Darmstadt et al., 2014). Another strength of IMAPI was the systematic engagement of key Brazilian stakeholders in the technical panels as well as national and international ECD experts to identify, select and reach consensus on indicators. Engaging with stakeholders in the decision-making process of selecting indicators can increase the likelihood that stakeholders will support project goals and implement decisions in the long term (Devente et al., 2016).
Considering IMAPI's goal to combine the selected indicators into a single overall index and sub-indexes for each NCF domain, taking into account an evidence-based analytical model is considered critical to improving reliability by giving higher weight to components with stronger impact on the outcome (Köhler, 2016). This is particularly true in the context of comparing municipalities within a given country (Köhler, 2016). We articulated within the participatory decisionmaking process a sound approach to account for SMART properties of each indicator considering specificity, attributes of quality and influence on ECD considering the NCF. In fact, the SMART approach has been described as a strong method to convert any type of weight assignment technique (i.e. relative/absolute) into numbers/weights (OECD, 2008;Velasquez & Hester, 2013). It also has been successfully used as a multicriteria decision analysis framework in healthcare with a focus on low-and middle-income countries (Németh et al., 2019). Through the SMART process, we gained an in-depth understanding of the different dimensions of each indicator to support the interpretation of IMAPI indexes. Hence, the SMART approach can be considered a strength of our methodology.
Furthermore, we identified some challenges related to data access and quality as well as the need to expand coverage of information and include new indicators to properly monitor the nurturing care at the municipal level. Regarding data access, even though we were using mostly routine governmental information system indicators, some of the databases were not publicly available. Thus, the request to access data to estimate multiple indicators, for example, 'visits by national home-visiting parenting skills programme', 'child immunization' and and quality and diversity of diet) and nutritional status(e.g. prevalence of stunting, prevalence of overweight/obesity, prevalence of low weight for age and prevalence of low weight for height) as originally planned by the research team. Due to the scarcity of information collected routinely to monitor critical factors influencing ECD outcomes in the existing national databases, two indicators from cross-sectional research were incorporated in the adequate nutrition ('household food insecurity') (Gubert & Pérez-Escamilla, 2013) and security and safety ('air pollution') (Brentani, 2020) Girade, 2018). Although other ECD initiatives and programmes are being implemented at the municipal level, there is no national system to collect and monitor this information for all municipalities. Thus, Criança Feliz was identified as a proxy of municipality's effort to supporting parenting skills and ultimately supporting responsive caregiving practices. The lack of information on responsive caregiving across countries has been reported previously (UNICEF and&Countdown to 2030, 2019). This missing information likely reflects the small financial investment to enable responsive caregiving in Brazil (Arregoces et al., 2019) and globally.
In the process of developing IMAPI, we faced some challenges that must be considered when interpreting nurturing care contexts of municipalities. First, the availability of indicators at the municipal level.
Many indicators did not have disaggregated information for municipalities and were excluded from this version of IMAPI, including 'basic sanitation', 'excessive alcohol consumption' and 'maternal mental health'. On the other hand, one advantage of using data aggregated at the municipal level is that the selection of indicators was not limited to any restrictions on Brazilian protection laws to access individual or identifiable information. Second, the lack of official data/estimation of children under 5 years old in the Brazilian population since 2016 made timeliness an important limitation for 20% of IMAPI's indicators. Fortunately, an estimation of the Brazilian population, including the population of children under 5 years old, was recently launched and will allow updates of IMAPI indicators for more recent years (Freire et al., 2019;IBGE, 2020.). Third, one challenge we can anticipate for the analytical weight when using IMAPI to follow the performance of municipalities over time is that SMART properties of indicators may also change over time (OECD, 2008). Fourth, we acknowledge that the lack of several indicators in the responsive caregiving domain is a current limitation of IMAPI.
In spite of these challenges, we were able to design and pilot a systematic methodology for identifying and selecting a set of nurturing care indicators at the municipal level. In summary, the greatest strength of IMAPI is the systematic consensus process to identify indicators using the best information available in existing Brazilian databases to monitor the domains of the NCF. The selection of IMAPI indicators was followed by the (i) use of a data engineering approach to create an index for each domain as well as a single index of the NCF and (ii) analysing whether IMAPI indexes can discriminate nurturing care environments across the 5570 municipalities in Brazil, which are described elsewhere.