Mapping Nutrition Interventions in Guatemala to Identify Opportunities to Improve Coverage and Coordination

Background: Guatemala has one of the highest rates of chronic child malnutrition (stunting) in the world, with minimal improvements over time, despite intensifying efforts. In 2018, a mapping effort was conducted with nutrition stakeholders, modelled after similar successful efforts in other countries. Methods: A public-private consortium used a survey to collect information from non-governmental nutrition stakeholders in Guatemala, including institutional name, sector, geographic areas of activities, and types of activities implemented. The survey data were used to populate an online mapping tool (ArcGIS Online). The distribution of institutions and programmatic activities at the departmental level was compared to child stunting prevalence using heat maps and linear regression. Geographic distribution of nutrition-specic and nutrition-sensitive activities aligning with Guatemala's four national strategy pillars (primary care services; education and behavior change; water sanitation and hygiene; and household income generation) were examined. Results: 5,861 program activities from 99 institutions were mapped. The majority of institutions were national non-governmental, community-based, or private sector organizations. Heat maps and statistical analysis demonstrated some misalignment between the distribution of resources versus national strategy priority areas and nutrition need. There were a low number of education and behavior change activities at a national level and relatively few primary care and health-sector strengthening activities in some priority departments. Conclusions: Mapping stakeholder institutions and nutrition activities in Guatemala identied several areas where resource allocation could be optimized to better address child stunting. The mapping initiative has been incorporated into Guatemala's national planning process and can assist with future monitoring efforts.

nutrition-speci c and nutrition-sensitive activities aligning with Guatemala's four national strategy pillars (primary care services; education and behavior change; water sanitation and hygiene; and household income generation) were examined.
Results: 5,861 program activities from 99 institutions were mapped. The majority of institutions were national non-governmental, community-based, or private sector organizations. Heat maps and statistical analysis demonstrated some misalignment between the distribution of resources versus national strategy priority areas and nutrition need. There were a low number of education and behavior change activities at a national level and relatively few primary care and health-sector strengthening activities in some priority departments.
Conclusions: Mapping stakeholder institutions and nutrition activities in Guatemala identi ed several areas where resource allocation could be optimized to better address child stunting. The mapping initiative has been incorporated into Guatemala's national planning process and can assist with future monitoring efforts.

Background
Guatemala is a middle-income country in Central America with one of the highest prevalences of stunting (low length/height for age) in the world. In recent years, efforts to combat stunting at a national level have intensi ed, but progress has been very slow, with the rate of stunting declining at less than 0.5% per year. (1) An important feature of Guatemala's nutrition landscape is the extensive involvement of aid organizations who support the National Food and Nutrition Security System, including bilateral aid, nongovernmental organizations, and corporate partnerships.(2, 3) However, the efforts of these stakeholders are poorly coordinated, with no centralized regulatory authority or reporting, which may contribute to the slow national progress against stunting.
In other settings, national or regional mapping efforts have been successfully used to better understand the landscape of nutrition stakeholders and to identify gaps in coverage and opportunities for improved coordination. For example, in 2015, the government of Burkina Faso, along with the United Nations Network for Nutrition and REACH (Renewed Efforts Against Child Hunger and undernutrition), used mapping of nutrition interventions targeting malnutrition to build consensus on geographic strategies and priorities and increase accountability among stakeholders.(4) Similar work has also been conducted in Egypt, Ethiopia, Tanzania, Mozambique, and Uganda. (5)(6)(7) Inspired by these efforts, we collaborated on an effort to collate, categorize, and map nutrition activities conducted by aid organizations in collaboration with Guatemala's Secretariat of Food and Nutrition Security (SESAN). In this short report, we use these data to visualize the geographic and sectorial distribution of nutrition initiatives in Guatemala, with the aim of highlighting potential gaps in coverage and resource distribution.

Description of Context
In December 2017, a public-private consortium was formed to pursue an effort to identify and map nutrition activities across non-governmental stakeholders in Guatemala, with endorsement from the Vice President of Guatemala, who oversees the National Strategy for the Prevention of Chronic Malnutrition.

Data Collection
The consortium designed a survey (see Supplementary Appendix) to collect information from aid organizations operating in Guatemala. The survey asked organizations to provide identi able information including institutional name, sector, geographic areas of activities, and types of activities implemented, as well as anonymous information on total budget, monitoring and evaluation practices, bene ciaries, language pro ciency, and institutional collaborations.
Survey data were collected via a secure website using the Qualtrics XM (Drive Provo, UT, USA) platform.
The consortium invited 320 stakeholder institutions to complete the data collection survey from July 15-August 31, 2018 via an o cial announcement on SESAN's email list. Consortium members conducted direct and social media outreach to encourage organizations to complete the survey.

Data Visualization and Analysis
Identi able data captured by the survey were displayed using an interactive web-based mapping tool created with ArcGIS Online (Redlands, CA, USA; see Supplementary Appendix Figure 1). Survey data were imported into Stata Version 14 (College Station, TX, USA) and R Studio Software Version 3.4.1 (Vienna, Austria). We tabulated descriptive statistics (number and percentage) for characteristics of participating institutions. Implemented activities were grouped within the four priority areas of the National Strategy (provision of primary care services, education and behavior change, water sanitation and hygiene; and household income generation) and were also classi ed as either nutritionsensitive or nutrition-speci c following de nitions proposed by Ruel and Alderman, and Hossain et al. Table 1).(8-10) Linear regression was used to examine the correlation between implemented activities and departmental stunting prevalence, using data on population-level stunting prevalence from the 2014-2015 Demographic Household Survey.(11) The geographic distribution of stunting prevalence, institutions, nutrition-sensitive and nutrition-speci c activities, and activities within each of the four national priority areas were visualized using heat maps generated in Datawrapper, Mapcite (Excel add-in 2016; Microsoft Corporation, Redmond, WA).

Ethical Approvals
Data collection for the mapping tool was determined to be exempt from ethics review by the Duke University Institutional Review Board.

Institution characteristics
Ninety-nine participating stakeholder institutions provided geographically localizable data on 5,861 nutrition-related activities. Organizational descriptive characteristics are summarized in Table 1. The majority of participating organizations were national non-governmental, community-based, or private sector organizations (59%), with bilateral aid, international non-governmental, religious, and academic organizations comprising the remainder. Overall, 42% of responding institution's activities were distributed in the six departments prioritized by the Government of Guatemala for their high rates of malnutrition and poverty, whereas the majority (58%) were located in non-priority departments. Nearly half (45%) reported conducting both nutrition-sensitive and nutrition-speci c activities, whereas 22% conducted only nutrition-speci c and 33% only nutrition-sensitive activities. Eighty-one percent of institutions reported an active monitoring and evaluation program. The ve most commonly mentioned program impact indicators included child growth (90%), breastfeeding and complementary feeding adequacy (65%), hygiene promotion (55%), micronutrient and food supplementation for women of childbearing age (25%), and prenatal care (23%). Table 1 Key characteristics of institutions and nutrition-related activities in the national mapping effort.

Geographic distribution of institutions and activities
We compared the geographic distribution of nutrition activities and institutions to child stunting prevalence and national priority areas. Figure 1 depicts the distribution of nutrition stakeholder institutions and activities compared to stunting prevalence in children under 5 years by department.
Misalignments between strategic priority and resource allocation were apparent. For example, the department of Totonicapán has the highest prevalence of child stunting in the nation but has notably fewer institutions and activities than other regions ( Figure 1A-C, solid arrow). Similarly, Chimaltenango has an intermediate prevalence of stunting but the highest institutional concentration in the country ( Figure 1A-C, dashed arrow). Finally, Escuintla has relatively low prevalence of stunting, but a disproportionately high concentration of nutrition activities ( Figure 1A-C, arrowhead). A linear regression of the number of active, responding institutions versus departmental stunting prevalence (Figure 2) shows that in general, priority departments have more responding nutrition stakeholder institutions compared to most non-priority departments. However, consistent with Figure 1, there are examples of relative under-resourcing of some priority departments (e.g., Totonicapán) and potential over-resourcing of some non-priority departments (e.g., Chimaltenango).
Distribution of nutrition-sensitive, nutrition-speci c, and priority area activities Based on both international guidelines and the Guatemalan National Strategy, effective nutrition policy requires complementary implementation of both nutrition-sensitive and nutrition-speci c interventions.
(8-10) Figure 3 gives the distribution of nutrition-sensitive and nutrition-speci c interventions by department. Variability in optimal implementation was observed. Some high-priority departments (Huehuetenango, Figure 3A-B, dashed arrow) showed good balance between nutrition-sensitive and nutrition-speci c interventions. However, others (Sololá, Figure 3A-B, solid arrow) showed a marked mismatch in implementation of nutrition-sensitive vs. nutrition-speci c interventions, with more emphasis on nutrition-speci c activities. We also mapped the distribution of activities according to the National Strategy's four priority areas ( Figure 4).(8) Most notable in this visualization is the relative absence of primary care and health-sector strengthening activities in most of the priority departments other than Sololá (Fig. 4A) as well as the overall low number of education and behavior change activities at a national level (Fig. 4B).

Discussion And Conclusions
Here we report on ndings from a 2018 nutrition stakeholder mapping effort in Guatemala, the rst publicly available data set of its kind in the country, and compared the distribution of resources with departmental child stunting prevalence and the priorities of the National Strategy for the Prevention of Chronic Malnutrition.(8) Our visualizations showed some signi cant misalignments between the concentrations of institutions and nutrition activities and departmental prevalence of child stunting and national priority areas. However, in some cases, over-resourced, non-priority departments do have a stunting prevalence that is high in the global context, which needs to be considered in decisions about resource allocation. Complementary deployment of multi-sectoral nutrition-speci c and nutrition-sensitive initiatives was inconsistent, with a balanced distribution of approaches in some departments but not others, and important potential de ciencies in activities aligned with key pillars of the National Strategy (especially primary care and education/behavior change) were noted. (8) These ndings should be interpreted cautiously, considering that only about one-third of nutrition stakeholders in the country responded to the survey. It is possible that there could be some systematic bias in the organizations that did and did not respond, in terms of location or focus area(s). However, our ndings on resource mismatch are similar to those from other countries where stakeholder mapping has been used to understand the landscape of nutrition interventions. For example, in the last decade, Activity REACH (Renewed Efforts Against Child Hunger and undernutrition) has worked with 22 countries, including Egypt, Burkina Faso, Ethiopia, Tanzania, Mozambique, and Uganda, and used mapping to support planning and resource allocation processes. (4,7) The initial efforts outlined in this paper laid the foundation for SESAN to adopt a similar approach. Importantly, the survey and visualization tools are being maintained by SESAN, with plans to update the visualizations going forward. We hope this will lead to ongoing conversations among stakeholders in Guatemala on how to more effectively allocate resources and collaborate to support the National Strategy for the Prevention of Chronic Malnutrition.  Linear regression of the relationship between number of nutrition stakeholder institutions per department and the departmental stunting prevalence in children under 5 years. Each department is plotted as a dot, and priority department names are bolded. Dotted lines represent the 95% con dence intervals around the regression.