Comparison of Child Undernutrition Anthropometric Indicators Across 56 Low- and Middle-Income Countries

Key Points

Question  How do different assessments of anthropometric failure in children compare?

Findings  In this cross-sectional study of 530 906 children in 56 low- and middle-income countries, there were substantial differences in the estimates of children experiencing anthropometric failure when using different approaches to measurement. Furthermore, children with simultaneous stunting, underweight, and wasting had significantly higher odds of diarrheal disease compared with children who exhibited no failure.

Meaning  A clearer picture of the prevalence of single and co-occurring anthropometric failure obtained through different methods of measuring undernutrition may accelerate progress toward meeting the targets of the United Nations’ Sustainable Development Goal 2 focused on ending hunger.

Importance  The United Nations’ Sustainable Development Goal Target 2.2 seeks to end all forms of malnutrition by 2030 by meeting targets, including the elimination of stunting and wasting in all children younger than 5 years. Such indicators are used to monitor childhood undernutrition but may not provide a complete picture at a population level.

Objective  To compare global estimates of the prevalence of undernutrition using conventional indicators of anthropometric failure (AF; stunting, underweight, and wasting); the Composite Index of Anthropometric Failure (CIAF); and a proposed classification system called Categories of Anthropometric Failure (CAF) as well as to investigate the association of the conventional indicators, CIAF, and CAF with diarrheal disease as an assessment of the validity of each measure.

Design, Setting, and Participants  Cross-sectional study of the prevalence of undernutrition among children in 56 low- and middle-income countries using data from the nationally representative Demographic and Health Surveys. The study included 530 906 children younger than 5 years. Data were collected from June 2005 to December 2018 and analyzed from September 27, 2020, to February 4, 2021.

Main Outcomes and Measures  Undernutrition identified according to conventional indicators (stunting, underweight, and wasting), the CIAF, and the proposed CAF classification system was estimated and compared. Six logistic regression models were used to examine the association between different classifications of anthropometric failure (AF) and morbidity.

Results  A total of 530 906 children (mean [SD] age, 29.0 [17.2] months; 272 355 [51.3%] boys and 258 551 [48.7%] girls) from 56 low- and middle-income countries were included in the analysis. Estimates of undernutrition generated using the conventional indicators of stunting, underweight, and wasting were lower than estimates generated using the CIAF in all countries. The CAF classification system pointed to considerable variation across countries in children with multiple AFs, which does not correspond to the overall prevalence of undernutrition. For example, 7.5% of children in Niger and 7.1% of children in Timor-Leste were stunted, underweight, and wasted, while 56.0% of children in Niger and 71.1% of children in Timor-Leste were undernourished according to the CIAF. In addition, children who had stunting, underweight, and wasting had 1.52 (95% CI, 1.45-1.61) times the odds of diarrhea compared with children who exhibited no AFs.

Conclusions and Relevance  The results of this study highlight the importance of using different approaches to aid understanding of the entire spectrum of AF with regard to research and development of policies and programs to address AF. The use of the CIAF and the CAF classification system may be useful for treatment to prevent AFs and could accelerate progress in meeting targets for the Sustainable Development Goal.

The United Nations’ Sustainable Development Goal (SDG) 2 aims to eliminate hunger by the year 2030.¹ Progress toward meeting this goal is defined through a set of targets related to nutritional well-being, agricultural productivity, and sustainability of food systems.² Target 2.2 seeks to end all forms of malnutrition by 2030 by meeting targets including the elimination of stunting and wasting in all children younger than 5 years.³ Underweight was used as an indicator to track progress toward ending hunger under the Millennium Development Goals; however, use of this indicator came under scrutiny because it can overstate progress given the increasing dual burden of undernutrition and overnutrition in many countries around the world.^4,5 Stunting, wasting, and underweight are often labeled collectively as states of anthropometric failure (AF)^6,7 and serve as proxies for severe malnutrition in the absence of nutritional intake data. Stunting, wasting, and underweight are defined as having a height-for-age z score, weight-for-height z score, and weight for age z score, respectively, of less than −2 SDs from the World Health Organization growth standards reference median.⁸

Although stunting and wasting indeed represent different biological processes, understanding them as separate indicators of nutritional status is a challenge given that they share many of the same causes.^9-11 Anthropometric failure is not static; children pass through different states of AF and may experience 1 or multiple failures at different times in their lives.¹⁰ An extensive literature review failed to find any independent causes of wasting that were not also associated with stunting.^4,12 In the last decade, a cyclical association between wasting and stunting has been found. Children with wasting are more likely to develop stunting, and in some places, these conditions may follow seasonal trends and environmental stressors.^10,13-16 Generally, stunting is considered to be relatively insensitive to marginal or short-term nutritional insufficiency. In contrast, underweight and wasting may be the result of acute starvation and/or disease, but neither indicator is able to clearly differentiate between recent and chronic nutritional deficiencies.^9,17 Understanding how the different states of AF are associated with undernutrition is further complicated by the potential role of infectious diseases in reducing appetite, increasing metabolic requirements, and increasing nutrient loss.¹⁸

Global interest in the co-occurrence of different types of AF is relatively new; the first global studies of which we are aware were published in 2017.¹⁰ Approaches to measuring childhood nutritional status largely still focus on stunting, underweight, and wasting as individual conditions. Existing World Health Organization indicators for malnutrition do not consider children who meet the criteria for more than 1 category of AF and as such, do not provide a method by which an overall estimate of child malnutrition at a population level can be ascertained.¹⁹ To provide a more complete picture of undernutrition at the population level, the Comprehensive Index of Anthropometric Failure (CIAF) was proposed as a way to calculate an aggregated estimate of the burden of childhood malnutrition by combining all children experiencing any single type or combination of AF into 1 summary measure^19-22; however, as a combined estimate, the CIAF loses the ability to distinguish between children facing different combinations of AF, which may be of specific interest for interventions to reduce the burden of malnutrition.

Previous research suggests that children who experience multiple AFs concurrently, especially those who had stunting, underweight, and wasting simultaneously, may have an elevated risk of morbidity and mortality,^10,19,23,24 although the cause of co-occurring failures remains poorly understood.^10,23,24 To address these challenges, we propose the Categories of Anthropometric Failure (CAF) classification system as a way to examine the burden of malnutrition at a population level, which corresponds to the disaggregated categories of AF as described in other studies.^19,24 The CAF classification system disaggregates the CIAF into all possible combinations of AF: stunting only; underweight only; wasting only; stunting and underweight; wasting and underweight; and stunting, underweight, and wasting. Of note, the combination of stunting and wasting (but not underweight) is theoretically impossible.²³ The eFigure in the Supplement details the differences between the conventional measures of AF, the CIAF, and the CAF classification system.

The aim of this study is to compare the prevalence of undernutrition using different approaches to measuring AF at the population level. Herein, we present a pooled and cross-country comparison of the prevalence of undernutrition identified by the conventional indicators (stunting, wasting, and underweight), the CIAF, and the CAF classification system using representative data from 56 low- and middle-income countries globally. We also assess the association of the conventional indicators, CIAF, and CAF with diarrheal disease as an assessment of the validity of each measure.

In this cross-sectional study, we used data from the most recent Demographic and Health Survey (DHS) in the 56 countries, with surveys conducted since 2010. Demographic and Health Surveys are nationally representative, cross-sectional surveys that have been conducted at regular intervals in more than 85 countries since 1984.^25,26 The DHS uses a multistage stratified cluster design.²⁷ Within each selected household, all children younger than 6 years are eligible for biomarker collection if they are a usual resident or had slept in the household the previous night.²⁸ For consistency with international definitions, we include only children 5 years and younger in our analyses. Sampling weights are provided for each survey to calculate nationally representative statistics.

Children’s anthropometric measurements were taken using standardized procedures that are consistent across surveys. Weight was measured using a digital scale, and length was collected by 2 trained individuals using a portable measuring board.²⁸ Data were collected from June 2005 to December 2018. The DHS data collection procedures were approved by the ORC Macro International, Inc, Institutional Review Board as well as by the relevant body in each country responsible for approving research studies with human participants. This study was exempt from further review, because it was based on an anonymous public use data set with no identifiable information on the survey participants. Participant oral informed consent was originally obtained at the time of data collection. This report follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cross-sectional studies.

The study population for the pooled analysis included 750 652 children 5 years and younger from 56 countries. We excluded 219 746 children missing anthropometric data. The final sample included 530 906 children.

Stunting, wasting, and underweight are the primary outcomes of interest for comparing the prevalence of undernutrition. We defined stunting, wasting, and underweight as having a height-for-age z score, weight-for-height z score, or weight-for-age z score of less than −2 SDs from the World Health Organization’s growth standards reference median.⁸ We calculated the CIAF as the summed total of children who experience any type or combination of AF (specifically those with stunting, underweight, or wasting). For the CAF, we calculated the 6 possible categories of AF separately (stunting only, underweight only, wasting only, stunting and underweight, wasting and underweight, and stunting, underweight, and wasting). In addition, we coded the presence of child diarrheal disease as a binary variable based on whether the child had experienced diarrhea in the last 2 weeks.

To compare the prevalence of undernutrition using different indicators, we calculated the prevalence of stunting, underweight, and wasting for each country according to the conventional indicators and the percentage of children who had stunting, underweight, or wasting who also exhibited another AF. We then calculated the CIAF for each country and compared it with the percentage of children who had stunting, underweight, and wasting according to the conventional definition. In addition, we calculated the CAF score for each country.

To assess the association of the different indicators of AF considered in this study and diarrheal disease, we conducted a pooled logistic regression analysis, with child diarrheal disease as the outcome measure. We ran 6 different models examining different indicators of AF as the exposure variable; 3 use the conventional measures of AF (stunting, underweight, and wasting), 1 model uses the CIAF, and 1 model uses the CAF on the pooled sample from the 56 countries included in the study. We adjusted for age-, sex-, and country-level fixed effects. P < .05 was chosen a priori to represent statistical significance, and 2-sided hypothesis tests were used. Stata, version 14.2 (StataCorp) was used to perform all analyses, and figures were produced using the R statistical program’s package ggplot2 (R Foundation for Statistical Computing).^29,30 Data were analyzed from September 27, 2020, to February 4, 2021.

Figure 1 shows the prevalence estimates of stunting, underweight, and wasting according to the conventional definitions and the CIAF calculated for the 56 countries included in the study (full results are available in eTable 1 in the Supplement). Burundi had the highest prevalence of stunting (57.9%). Timor-Leste had the highest prevalence of underweight (44.2%). India had the highest prevalence of wasting (21.0%). Timor-Leste had the highest burden of undernutrition as measured by the CIAF (71.1%). Although the overall prevalence of stunting in Timor-Leste and Burundi was similar (57.6% and 57.9%, respectively), the percentages of children who were underweight and wasting were significantly lower in Burundi than in Timor-Leste (12.7% lower and 15.7% lower, respectively). When comparing the aggregate burden of undernutrition in each country as measured by the CIAF, the CIAF suggested a 10% higher prevalence of undernutrition in Timor-Leste than in Burundi, a difference that is not directly ascertainable from looking at each conventional indicator alone. In the case of India, the substantial overlap in children experiencing multiple AFs is evident when comparing the prevalence of stunting (38.4%), underweight (35.7%), and wasting (21.0%) with the aggregate estimate provided by the CIAF (55.2%).

Figure 2 shows the percentage of children with stunting, underweight, or wasting with 1 or concurrent AFs (full estimates are available in eTable 2 in the Supplement) to illustrate the wide variation in the co-occurrence of AF in children globally. There were 13 countries in which more than 50% of children with stunting had at least 1 additional concurrent AF, and there were 7 countries in which less than 20% of children with stunting had an additional AF. Most children with underweight also experience another type of failure, but there still remains an important percentage of children with underweight with no other concurrent AF. Ghana had the highest percentage of children who were only underweight without an additional, concurrent AF (18.4%), and Guatemala had the lowest (3.1%). Guatemala had the highest percentage of children who experienced wasting in combination with another type of AF (90.7%), whereas only 25.4% of children who experienced wasting in Armenia had another concurrent failure.

Figure 3 presents the prevalence of each disaggregated type of AF as defined by the CAF classification system across all 56 countries included in the study (full estimates are available in eTable 3 in the Supplement). Among the 29 countries where the prevalence of stunting was greater than 30%, the percentage of children who had stunting with other concurrent forms of AF varies considerably. Within these countries, the percentage of children who were stunting and underweight ranged from 7.0% in Rwanda to 28.6% in Timor-Leste, and the percentage of children who had stunting, underweight, and wasting ranged from 0.5% in Guatemala to 7.5% in Niger. However, in other countries, the prevalence of stunting was lower, but the proportion of children who experienced stunting in combination with another AF was substantial. For example, in the Republic of Maldives and Senegal, the prevalence of stunting was 18.0% and 18.4% respectively, but approximately 40% of those children were stunting and underweight. In Gambia, the prevalence of stunting was 24.2%, but 11.1% of those children with stunting, underweight, and wasting. Among the countries where more than 10% of children were wasting, the proportion of children who had stunting, underweight, and wasting ranged from 1.6% in São Tomé and Príncipe to 7.5% in Niger. Again, other countries with a lower prevalence of wasting also had a substantial proportion of children who had stunting, underweight, and wasting. For instance, 5.8% of children in Burundi experienced wasting, but 63.7% of those children had stunting, underweight, and wasting.

The Table presents the results of 6 logistic regression models showing the association of the 3 conventional indicators of AF, the CIAF, and the CAF with child diarrheal disease. All indicators of AF were associated with increased odds of diarrhea. Of the conventional indicators of AF, model 3 (child is underweight vs not underweight) showed that children who were underweight compared with those who were not had the highest odds (1.32; 95% CI, 1.28-1.36) of having experienced diarrhea in the last 2 weeks compared with the estimates obtained from the models including the other forms of AF (model 2 [stunting vs no stunting] and model 4 [wasting vs no wasting]). Model 6 (CAF) included the different combinations of AF as measured by the CAF. In this model, children who had stunting, underweight, and wasting had 1.52 (95% CI, 1.45-1.61) times the odds of diarrhea compared with children who exhibited no AFs, which was the highest among the different combinations of failure included in the CAF.

The results of this study provide important considerations for how undernutrition, as defined by AF, is measured globally, which may have implications for achieving SDG Target 2.2. As of 2015, only 16.5% out of 188 countries had eliminated stunting, the same percentage that had eliminated wasting.³¹ First, the results suggest that the picture of AF across countries is largely dependent on the measurement approach used. Second, there is substantial overlap between the 3 conventional indicators used to assess undernutrition at a population level and estimates of undernutrition using these indicators vary considerably across countries, thus potentially blurring international comparisons of child nutritional status. Furthermore, focusing only on the overall prevalence of stunting, underweight, and wasting without disaggregation may obscure the proportion of children in a population most at risk for poor health outcomes owing to multiple, simultaneous AFs.^19,32 Third, results of the present study suggest that further examination of the association between multiple, concurrent AFs, as defined by the CAF, and other forms of child morbidity may be beneficial to programs seeking to address undernutrition, but more research is needed to further elucidate this association. Without examining the co-occurrence of different types of AF, policy and intervention may be somewhat misaligned with the burden of disease.

The use of distinct indicators for program monitoring may also give an incomplete picture of progress. Changes observed over time in stunting, underweight, or wasting can be difficult to interpret because they may give mixed messages on both the rate and direction of change in undernutrition.³³ For instance, Nandy et al³³ highlighted data from 2 consecutive Demographic and Health Surveys in Zimbabwe, which provided conflicting information about changes in child nutritional status: the number of children who exhibited stunting and wasting increased between surveys, but the number of underweight children declined. Part of the difficulty in interpreting these indicators may be owing to the considerable overlap in the population experiencing multiple, concurrent failures.²⁰

The current focus on the 3 conventional indicators of undernutrition represents decades of work to build evidence and consensus around these measures. The use of anthropometry to assess malnutrition grew from early efforts to create an internationally comparable classification system for estimating the burden of undernutrition at a population level.³⁴ A timeline showing the evolution of studies^35-46 using anthropometry in assessing undernutrition is shown in Figure 4. Low height-for-age and weight-for-height were proposed as measures of undernutrition in the early 1970s, which were subsequently termed stunting and wasting.^35,36 The use of −2 SD from the reference population median as a cutoff was introduced in 1977 and endorsed by the World Health Organization in 1986,^37,44 although it was noted at the time that the choice of −2 SD as the cutoff was somewhat arbitrary.^37,47 The use of a statistically determined cutoff has since been questioned given that there is no biological basis for its definition as a clear threshold, and the association between anthropometry and poor health operates along a continuous gradient.⁴⁸ In 2008, studies on maternal and child undernutrition promoted the use of stunting and wasting to assess child nutritional status globally and presented them as distinct concerns with distinct interventions.^4,36

The use of the CIAF provides a simplified way to assess undernutrition by offering a single, comprehensive estimate; however, it does not provide information on children experiencing multiple AFs. Therefore, the CAF may be a useful tool for policy makers and interventionists to better prevent and address specific combinations of AF, especially if coupled with a deeper understanding of the differing causes of different types of concurrent failure. It should be noted that both the CIAF and the CAF rely on different configurations of the conventional, widely used indicators of AF. A detailed comparison of the advantages and disadvantages of each approach is provided in eTable 4 in the Supplement. Future research should question whether all categories of AF are truly useful in assessing undernutrition within a population and their association with morbidity and mortality.¹⁹

The results of this study highlight the importance of considering different methods of measuring AF to aid understanding of the entire spectrum of AF. By examining the composition of the different categories of AF within a population, new insights can be gained about the burden of disease, which may lead to a prioritization of different programmatic and policy approaches in countries with similar overall burdens of AF. For example, the overall prevalence of AF as measured by stunting and the CIAF is similar in Mali and Mozambique; however, the CAF provides important insight into the composition of AF within each country, suggesting a much larger burden of multiple AFs in children in Mali than in Mozambique. Thus, despite a similar overall burden of AF as defined through conventional indicators, the CAF illustrates that different programmatic approaches may be necessary in each country to most effectively reduce AF given the dramatically different burden of disease when considering concurrency.

The use of the CIAF and the CAF may be useful in responding to calls to shift focus from treatment to prevention of undernutrition¹⁰ by contributing to the discourse on whether to prioritize individual vs community intervention. Previous research on disaggregated measures of AF derived from the CIAF in India found that children who had stunting, underweight, and wasting tended to live in the poorest households.¹⁹ Because economically disadvantaged children are more likely to have multiple AFs simultaneously, the presence of multiple vs singular AFs may better distinguish between chronic and acute undernutrition than the traditional distinction between stunting vs wasting. More research using the CAF could help elucidate individual and ecological factors of multiple failures to better design interventions to prevent AFs altogether or to prevent children with 1 failure from developing multiple failures.

Progress toward reducing morbidity and mortality among children who have AF may be limited if interventions are successful at reducing stunting or wasting individually without addressing the complex interplay between the co-occurrence of the different categories of AF.¹⁰ Furthermore, given the cyclic association between stunting and wasting, addressing them as separate conditions may limit the effectiveness of programmatic efforts that view them as distinct; however, different states of AF respond to treatment differently.¹⁰ In particular, there are many causes of stunting, and the results of the present study emphasize that not all countries in which a large proportion of children with stunting also have large proportions of children with wasting. Thus, focusing prevention efforts to specifically address children with the highest risk of morbidity and mortality associated with AF may allow country programs to better focus efforts on reducing adverse health outcomes among children at highest risk. In particular, interventions designed to address wasting in populations where wasting is an important factor of stunting may focus intervention efforts on specific seasons or addressing risks specific to children by sex, socioeconomic status, or maternal characteristics.¹⁶ Thus, prevention efforts that consider the complex and upstream common causes of concurrent failures may ultimately have a more substantial impact on preventing children from experiencing multiple forms of AF as well as the poor health outcomes associated with concurrent failures than those that focus on singular etiologic pathways.

This study has several important limitations. First, the data presented are cross-sectional, and thus, we cannot account for children who move between categories of AF over time. Second, the results may be subject to measurement error that could affect the comparability of results between countries. Furthermore, approximately 29% of the children in the initial sample were missing anthropometric data across all countries. Because we did not find any patterns related to missing data, we believe that the data are missing at random; thus, we do not believe that the missing data bias our results. Third, we focused the analysis on undernutrition and did not consider obesity; overweight was added to the CIAF in 2007.²¹

In this study, the CAF found considerable variation across countries in children with multiple AFs that did not correspond to the overall prevalence of undernutrition. Current progress is limited in eliminating stunting and wasting. Progress must improve if countries around the globe are to meet the targets set forth in the SDG 2. Examining different methods of measuring undernutrition may contribute to a deeper understanding of the spectrum of AF as well as inform policies and programs to address AFs through prevention and treatment. In turn, accelerating progress toward meeting the SDG 2.2 target may be possible.

Accepted for Publication: December 21, 2021.

Published: March 11, 2022. doi:10.1001/jamanetworkopen.2022.1223

Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2022 Gausman J et al. JAMA Network Open.

Corresponding Authors: S. V. Subramanian, PhD, Harvard Center for Population and Development Studies, Nine Bow Street, Cambridge, MA 02138 (svsubram@hsph.harvard.edu); Rockli Kim, ScD, Korea University College of Health Science, Hana Science Building, Room 355, 145 Anam-ro, Seongbuk-gu, Seoul 02841, South Korea (rocklikim@korea.ac.kr).

Author Contributions: Dr Subramanian had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Gausman, Kim, Li, Subramanian.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Gausman, Tu.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Gausman, Li, Rajpal.

Administrative, technical, or material support: Li.

Supervision: Kim, Joe, Subramanian.

Conflict of Interest Disclosures: None reported.