Global, Regional, and National Burden of Protein–Energy Malnutrition: A Systematic Analysis for the Global Burden of Disease Study

Background: Statistical data on the prevalence, mortality, and disability-adjusted life years (DALYs) of protein–energy malnutrition are valuable for health resource planning and policy-making. We aimed to estimate protein–energy malnutrition burdens worldwide according to gender, age, and sociodemographic index (SDI) between 1990 and 2019. Methods: Detailed data on protein–energy malnutrition from 1990 to 2019 was extracted from the Global Burden of Disease (GBD) database. The global prevalence, deaths, and DALYs attributable to protein–energy malnutrition and the corresponding age-standardized rates (ASRs) were analyzed. Results: In 2019, the global prevalence of protein–energy malnutrition increased to 14,767,275 cases. The age-standardized prevalence rate (ASPR) showed an increasing trend between 1990 and 2019, while the age-standardized deaths rate (ASDR) and age-standardized DALYs rate presented a significantly decreasing trend in the same period. Meanwhile, there was a clearly ASPR, ASDR, and age-standardized DALYs rate downtrend of the prediction curve when the SDI went up. Conclusions: PEM still has a relatively serious disease burden in the world, especially in children and the elderly. At the same time, this phenomenon will be more obvious due to the aging of the world’s population. Effective prevention measures should be strengthened to continuously improve public health conditions.


Introduction
Malnutrition was defined as "a subacute or chronic state of nutrition, in which a combination of varying degrees of under-or overnutrition and inflammatory activity has led to changes in body composition and diminished function" [1,2]. Protein-Energy Malnutrition (PEM) is a series of diseases due to the malnutrition of all macronutrients, including marasmus, intermediate states of kwashiorkor-marasmus, and kwashiorkor. PEM is a common nutritional problem worldwide and can be seen in both developed and developing countries. The prevalence of the PEM in the older communities varied by region, from 0.8% to 24.6%, and it is also affected by the gender sampling frame and rurality [3]. The number of cases of children with PEM is declining globally but also varies by region; for example, it continues to decline in Asia but is increasing in Africa [4]. In addition, PEM poses a threat to public health, especially in children and the elderly, by impairing the immune response, which can lead to death [3,5]. Although the global

Data Source
The latest release of the Global Burden of Disease, Injuries, and Risk Factors Study (GBD) 2019 results was applied. The data was obtained on the official GBD website (http://ghdx. healthdata.org/gbd-2019 (accessed on 20 January 2022)) according to operation guidelines, without any inclusion or exclusion criteria. Data sources for the burden of protein-energy malnutrition were extracted with the Global Health Data Exchange (GHDx) query tool (http://ghdx.healthdata.org/gbd-results-tool (accessed on 20 January 2022)). Protein-energy malnutrition includes moderate and severe acute malnutrition, commonly referred to as "wasting," and was defined in terms of weight-for-height Z-scores (WHZ) on the WHO 2006 growth standard for children. We quantified non-fatal PEM burden in four mutually exclusive and collectively exhaustive categories, reflecting distinct gradations of disability that can occur: moderate wasting without oedema (WHZ < −2 SD to < −3 SD), moderate wasting with oedema (WHZ < −2 SD to < −3 SD), severe wasting without oedema (WHZ < −3 SD), and severe wasting with oedema (WHZ < −3 SD). For PEM, ICD 10 codes are E40-E46. 9 and E64.0, and ICD 9 codes are 260-263.9 [9]. We acquired data on the prevalence, mortality, and DALYs of protein-energy malnutrition and the respective age-standardized rate (ASR) of protein-energy malnutrition from 1990 to 2019. The 204 countries/territories were then categorized into five categories on the basis of the sociodemographic index (SDI) in 2019: high, high-middle, middle, low-middle, and low SDI. Besides, the human development index (HDI) values were derived from the World Bank.
The GBD estimation process is based on identifying multiple relevant data sources, including censuses, household surveys, civil registration and vital statistics, disease registries, health service use, and other sources. Each of these types of data is identified from a systematic review of published studies, searches of government and international organization websites, primary data sources such as the Demographic and Health Surveys, and contributions of datasets by GBD collaborators. First, individual-level and tabulated child anthropometry data from health surveys, literature, and national reports were used and centralized to inform the prevalence of weight-for-height Z-scores (WHZ) decrement in each category corresponding to our case definitions. Second, to inform the proportion of children under 5 years who have signs of organ failure manifested as oedema (i.e., kwashiorkor), a compiled dataset of surveys was conducted using Standardized Monitoring and Assessment of Relief and Transitions (SMART) methods. All data were extracted with the most detailed standard demographic identifiers available, including age, sex, country, year, and subnational location, if available [9].

Statistical Analysis
Estimated annual percentage change (EAPCs) and average annual percentage change (AAPCs) of age-standardized rates (ASRs) were calculated. We assumed that the natural logarithm of ASR was linear along with time; hence, Y = α + βX + ε (X represents the calendar year, Y represents ln(ASR)), ε represents the error term, and β indicates the positive or negative trends of ASRs). The formula of the EAPC was EAPC = 100 × (exp(β) − 1) and its 95% confidence intervals (CI) were calculated according to the linear model. When EAPC and the upper boundary of CI are negative, ASR represents a descending trend. In contrast, ASR is considered to be in an upward trend.
Average Annual Percent Change (AAPC) is a single number which represents the occurrence of disease in a population via applying the geometrically weighted averages for annual percent changes. The analysis was performed by the JointPoint Regression Program 4.9.0.1 (National Cancer Institute, Bethesda, MD, USA) provided by the United States National Cancer Institute Surveillance Research Program. To obtain the AAPCs, the software was applied to track trends in GBD data over time and then fit an underlying model possible to the data via connecting different line segments on a logarithmic scale. The segments are called "JointPoints", and each is tested for significance by a Monte Carlo permutation method. The analyses were performed using R statistical software version 4.1.2 (https://www.r-project.org (accessed on 25 February 2022)). p < 0.05 was regarded as statistically significant.
For the AAPC of prevalence, East Asia was top, High−middle SDI second, with Eastern Sub−Saharan Africa at the bottom ( Figure 1A). In the AAPC of deaths, most regions had a negative number, with Central Europe, High−income North America, Australasia, Western Europe, and South Asia possessing the most negative number simultaneously ( Figure 1B). As for the AAPC of DALYs, Western Europe had the biggest value, and South Asia had the smallest ( Figure 1C). The relative values of DALYs are YLLs and YLDs. Only High−income North America achieved a positive value in the AAPC of YLLs, and East Asia received the smallest negative number. However, East Asia got the biggest AAPC of YLDs. Eastern Sub−Saharan Africa was at the bottom in the AAPC of YLDs (Supplementary Figure S1).

Age and Sex
For the prevalence rate, the prevalence basically increased with age, except children aged 1-4 had the highest prevalence ( Figure 3A). For the death rate, the prevalence increased significantly with age, peaking at 95+, except in children aged 1-4 ( Figure 3B). With the DALYs rate, the value of minors decreased with age, the value of adults increased significantly with age, and female children aged 1-4 years were significantly higher than male children ( Figure 3C). The trend of the DALYs rate also appeared in the YLL rate and YLD rate ( Figure S4).  biggest EAPC of prevalence, and Guatemala had the smallest ( Figure 1B). Mali had the highest number of ASDRs (Supplementary Figure S2A). The Czech Republic had the biggest EAPC of ASDR, and the Democratic People's Republic of Korea had the smallest (Supplementary Figure S2B). Eritrea had the highest age-standardized DALYs rate, while Singapore had the smallest number (Supplementary Figure S3A). The Czech Republic had the biggest EAPC of DALYs, and Cambodia had the smallest (Supplementary Figure S3B).

Age and Sex
For the prevalence rate, the prevalence basically increased with age, except children aged 1-4 had the highest prevalence ( Figure 3A). For the death rate, the prevalence increased significantly with age, peaking at 95+, except in children aged 1-4 ( Figure 3B). With the DALYs rate, the value of minors decreased with age, the value of adults increased significantly with age, and female children aged 1-4 years were significantly higher than male children ( Figure 3C). The trend of the DALYs rate also appeared in the YLL rate and YLD rate ( Figure S4).

The Socio-Demographic Index (SDI) and Human Development Index (HDI)
SDI refers to the socio-demographic index, and HDI refers to the human development index. Their values can be used to judge the degree of economic development of a country or region so as to compare the relationship and causes between protein malnutrition and the level of national economic development.
For the relationship between prevalence and SDI, there was clearly an ASPR downtrend of the prediction curve when the SDI went up (R = −0.59, p < 2.2 × 10 −16 . At the regional level, the ASPRs of Global, South Asia, Southeast Asia, and Western Europe were higher than predicted, and the Southern Sub-Saharan Africa, Central Asia, High-income Asia Pacific, Southern Latin America, Oceania, Caribbean, Andean Latin America, Highincome North America, Central Latin America, Tropical Latin America, Australasia, and Central Sub-Saharan Africa ASPRs were lower than predicted (Figure 4). At the national level, it also showed a negative trend between prevalence and SDI (R = −0.33, p = 2 × 10 −6 ) (Supplementary Figure S6A). The EAPC of prevalence was negatively correlated with SDI (p = −0.53 (95% Cl: = −0.63-0.41), p < 0.001) ( Figure 5A).

The Socio-Demographic Index (SDI) and Human Development Index (HDI)
SDI refers to the socio-demographic index, and HDI refers to the human development index. Their values can be used to judge the degree of economic development of a higher than predicted, and the Southern Sub-Saharan Africa, Central Asia, High-income Asia Pacific, Southern Latin America, Oceania, Caribbean, Andean Latin America, Highincome North America, Central Latin America, Tropical Latin America, Australasia, and Central Sub-Saharan Africa ASPRs were lower than predicted (Figure 4). At the national level, it also showed a negative trend between prevalence and SDI (R = −0.33, p= 2 × 10 −6 ) (Supplementary Figure S6A). The EAPC of prevalence was negatively correlated with SDI (p = −0.53 (95% Cl: = −0.63-0.41), p < 0.001) ( Figure 5A). For the relationship between deaths and SDI, there was clearly an ASDR downtrend of the prediction curve when the SDI went up (R = −0.73, p < 2.2 × 10 −16 ). At the regional level, the ASDRs of Global, Southern Sub-Saharan Africa, Eastern Sub-Saharan Africa, Andean Latin America, Central Latin America, Tropical Latin America, Central Sub-Saharan Africa, and Southeast Asia were higher than predicted, and Central Asia, South Asia, Oceania, North Africa, and the Middle East were lower than predicted (Supplementary Figure S5A). At the national level, it also showed a negative trend between deaths and SDI (R = −0.69, p < 2.2e × 10 −16 ) (Supplementary Figure S6B). The EAPC of deaths was negatively correlated with SDI (p = −0.42 (95% Cl: −0.53-0.29), p < 0.001) ( Figure 5B). For the relationship between deaths and SDI, there was clearly an ASDR downtrend of the prediction curve when the SDI went up (R = −0.73, p < 2.2 × 10 −16 ). At the regional level, the ASDRs of Global, Southern Sub-Saharan Africa, Eastern Sub-Saharan Africa, Andean Latin America, Central Latin America, Tropical Latin America, Central Sub-Saharan Africa, and Southeast Asia were higher than predicted, and Central Asia, South Asia, Oceania, North Africa, and the Middle East were lower than predicted (Supplementary Figure S5A). At the national level, it also showed a negative trend between deaths and SDI (R = −0.69, p < 2.2 × 10 −16 ) (Supplementary Figure S6B). The EAPC of deaths was negatively correlated with SDI (p = −0.42 (95% Cl: −0.53-0.29), p < 0.001) ( Figure 5B).
For the relationship between DALYs and SDI, there was clearly an age-standardized DALYs rate downtrend of the prediction curve when the SDI went up (R = −0.73, p < 2.2 × 10 −16 ). At the regional level, the age-standardized DALYs rates of Global, Southern Sub-Saharan Africa, Eastern Europe, the Caribbean, Central Sub-Saharan Africa, and Western Europe were higher than predicted, and Central Asia, Oceania, North Africa, and the Middle East were lower than predicted (Supplementary Figure S5B). At the national level, it also showed a negative trend between deaths and SDI (R = −0.66, p < 2.2 × 10 −16 ) (Supplementary Figure S6C). The EAPC of DALYs was negatively correlated with SDI (p = −0.64 (95% Cl: −0.72-0.54), p < 0.001) ( Figure 5C).
We divided the SDI into five levels: High SDI, High-middle SDI, Middle SDI, Lowmiddle SDI, and Low SDI. The number of prevalence cases in High SDI and High-middle SDI regions was significantly lower than that in Middle SDI, Low-middle SDI, and Low SDI regions. The High SDI and High-middle SDI regions' cases were dominated by people over the age of 15, while the Middle SDI, Low-middle SDI, and Low SDI regions' cases were dominated by people under the age of 15. Meanwhile, the Low−middle SDI region had the highest number of cases in children under 5 years of age ( Figure 6). As for the death cases, the global death cases were mainly children under 5 years old. The Low-middle SDI and Low SDI regions had the largest number of overall cases, but they are decreasing year by year. In terms of age composition, the High SDI had always been dominated by the 75+ age group, and the High-middle SDI and Middle SDI areas gradually transitioned from children under 5 years old to 75+ over time, and Low-middle SDI and Low SDI areas had always been mainly for children under 5 years old. In particular, between 1995 and 2002, the total number of cases in the Low-middle SDI region increased abnormally (Supplementary Figure S8). For DALYs cases, the total number of cases in the world and regions decreased other than High SDI. Among them, Low-middle SDI and Low SDI had the largest total number of cases, and the High SDI region had the smallest total number of cases. In terms of age composition, the Middle SDI, Low−middle SDI, and Low SDI regions were dominated by the age group under 5 years old. The proportion of people under the age of 5 in the High−middle SDI and Middle SDI regions has been decreasing. The Low−middle SDI region also increased abnormally between 1995 and 2002 (Supplementary Figure S9). SDI regions. The High SDI and High-middle SDI regions' cases were dominated by people over the age of 15, while the Middle SDI, Low-middle SDI, and Low SDI regions' cases were dominated by people under the age of 15. Meanwhile, the Low−middle SDI region had the highest number of cases in children under 5 years of age ( Figure 6). As for the death cases, the global death cases were mainly children under 5 years old. The Lowmiddle SDI and Low SDI regions had the largest number of overall cases, but they are decreasing year by year. In terms of age composition, the High SDI had always been dominated by the 75+ age group, and the High-middle SDI and Middle SDI areas gradually transitioned from children under 5 years old to 75+ over time, and Low-middle SDI and Low SDI areas had always been mainly for children under 5 years old. In particular, between 1995 and 2002, the total number of cases in the Low-middle SDI region increased abnormally (Supplementary Figure S8). For DALYs cases, the total number of cases in the world and regions decreased other than High SDI. Among them, Low-middle SDI and Low SDI had the largest total number of cases, and the High SDI region had the smallest total number of cases. In terms of age composition, the Middle SDI, Low−middle SDI, and Low SDI regions were dominated by the age group under 5 years old. The proportion of people under the age of 5 in the High−middle SDI and Middle SDI regions has been decreasing. The Low−middle SDI region also increased abnormally between 1995 and 2002 (Supplementary Figure S9).

Discussion
This study is the first experiment to use the GBD2019 database to investigate the global burden of PEM. We comprehensively assessed the prevalence, deaths, DALYs,

Discussion
This study is the first experiment to use the GBD2019 database to investigate the global burden of PEM. We comprehensively assessed the prevalence, deaths, DALYs, YLLs, YLDs, and corresponding ASRs of PEM and compared data across countries, regions, age, sex, and SDI. Globally, both the ASDR and age-standardized rate of DALY decreased from Nutrients 2022, 14, 2592 13 of 17 1990 to 2019, but ASPR increased slightly. There is a relationship between economic level and malnutrition [10,11]. The improvement of the economic level and medical science can cause the improvement of the expected life span, which could increase in ASPR of PEM. At the same time, because the PEM diagnostic standards continuously improved between 1990 and 2019, the screening of PEM is more comprehensive, which may lead to the continuous increase in PEM's ASPR, and also lead to abnormal turning points in the JointPoint diagrams [2,12,13]. The decline in ASDR and ASR DALYS can also be attributed to the improvement of medical technology.
The burden of PEM has an obvious relationship with the socioeconomic level. Socioeconomic indicators HDI and SDI were negatively correlated with PEM burdens, which means that higher socioeconomic levels have lower PEM burdens. High HDI represents a higher level of prevention and treatment, and higher SDI represents a higher social security capability. These are all favorable factors to reduce the burden of PEM.
The age distribution of patients with PEM has a good concentration. We found that PEM patients are mainly concentrated in the children group and older group. For the children group, PEM is the most serious malnutrition in children. Child patients generally lack the amino acids required for growth and development. The lack of amino acids will affect the normal growth of cells and collectives. For example, the lack of amino acids will cause the main growth regulation of cells, Mechanical Target of Rapamycin Complex 1 (MTORC1), to synthesize and inhibit the growth of cells and the body, which leads to slow development and immune function defects [14][15][16]. PEM causes 56% of children's deaths in developing countries [17]. The main risk factors for children's PEM are environmental, poverty, gender, food culture, and immunization [16,[18][19][20], which will be reflected in the burden level of different regions. However, the high SDI place can guarantee a better nutritional level for children. Compared with the condition of the children group, the elderly group is different. PEM can be seen in about 50% of the elderly in hospitalization. With the increase in age, the elderly groups have gradually experienced a loss of appetite, insufficient diet, and weakened intestinal absorption capacity, which can easily lead to PEM. The elderly with PEM are prone to decreased concentration of ornithine, histidine, glutamic acid, and glutamine, which leads to the symptoms of decreased physical function and decreased immune function, which eventually leads to death [21,22]. Therefore, government and medical institutions need to pay special attention to the development of children and elderly PEM, actively test relevant indicators, and formulate positive strategies to prevent death.
As in this experiment, women had been found to be at a higher risk of PEM than men in several studies [23,24], but no specific studies have demonstrated whether the specific reasons are related to gender inequality, physiological differences, or life expectancy, etc. [3]. This also suggests the direction of our future research in order to provide women with more preventive and therapeutic measures. At the regional level, ASPR, ASDR, age-standardized DALYs rate, EAPC, and AAPC vary from place to place. South Asia, Southeast Asia, and East Asia occupy the top positions in the ASPR rankings, while East Asia also has the highest EAPC of ASPR, the lowest EAPC of ASDR, the lowest EAPC of DALYS, and the highest AAPC of prevalence. The rapid rise of ASPR in East Asia may be related to the general rapid aging of countries in East Asia [25,26]. Moreover, the elderly are more likely to obtain PEM. At the same time, South Asia, Southeast Asia, and East Asia are generally developing countries, and the ability to provide formal medical services is not as good as that of developed countries [27]. However, in East Asian countries such as China, the number of PEM has decreased significantly in recent years due to rapid economic development, the substantial improvement in the quality of medical services, and major efforts in PEM prevention [28]. At the same time, China has the largest population base in East Asia, so it is the most likely reason for the decline in ASDR and DALYs [3]. Rapidly developing economies and incomplete healthcare systems also pose enormous challenges to PEM public health services in East Asia [26]. Although Eastern Sub-Saharan Africa had the lowest AAPC of prevalence, it had the highest ASDR and age-standardized DALYs rate, which is closely related to the lower total population, economic level, and medical level of this region [29,30]. At the same time, maize as the main energy source in Eastern Sub-Saharan Africa also leads to the apparent lack of tryptophan and lysine in the region [31]. This region needs certain dietary guidance policies to promote the reduction in the PEM burden.
This study has performed a comprehensive review of the PEM burden at the global, regional, and national levels, but there exist some limitations. Because of the complexity and breadth of the data, the PEM burden should be interpreted cautiously. Fewer data are available from countries and regions with lower SDI values, and the disease burden may be underestimated because of different levels of registration management. Moreover, though the data of the GBD study are considered of high quality, differences in data collecting, extracting, coding, and quality of data sources inevitably compromise the robustness and accuracy of GBD estimates. Obviously, the quality of data collection coding and data sources is qualitatively related to the economic and political level of countries and regions. In developed countries, especially developed countries with relatively complete medical systems, such as the United Kingdom, the United States, and France, the data obtained are relatively reliable and complete, while in some developing countries, such as most African and Central Asian countries, due to imperfections in the economy, war, or the imperfection of the medical system, statistics may be lost or wrong. At the same time, the true burden of the disease may be underestimated due to the difficulty of diagnosing PEM. Last but not least, the fluctuations in disease burden data may partly represent the detection bias associated with adjustments in screening protocols rather than real changes.

Conclusions
This study provides a reference for monitoring the burden and trends of PEM at the global, regional, and national levels, which is crucial to conducting interventions that might slow down the rising burden of PEM. Our findings found that PEM still has a relatively serious disease burden in the world, especially in lower SDI regions, children, and the elderly. Consequently, more attention should be paid to developing early prevention and treatment measures for PEM in lower SDI regions and countries, such as safeguarding the food supply, eliminating hunger, and improving overall nutritional status. At the same time, particular attention to children and elderly PEM is needed. Effective steps, such as enhancing nutrition-related health education, strengthening nutritional support, and early aggressive treatment, should be formulated to relieve the burden of PEM. Further studies are required to identify more useful public health interventions.  Figure S7. The association between Human Development Index (HDI) and estimated annual per-centage change (EAPC) in Age-standardized prevalence rate (ASPR), Age-standardized death rate (ASDR), and age-standardized disabil-ity-adjusted life years of protein-energy malnutrition. HDI (2019) was negatively associated with EAPC in (A) ASPR (ρ = −0.53, p < 0.001), (B) APSR (ρ = −0.42, p < 0.001), and (C) age-standardized DALY (ρ = −0.64, p < 0.001). Figure S8. The pro-portion of the five age groups for protein-energy malnutrition death cases between 1990 and 2019 globally, and in high, high-middle, middle, low-middle, and low SDI quintiles. The populations were divided into five age groups: under 5, 5-14 years, 15-49 years, 50-69 years, and 70+ years. SDI values can be used to judge the degree of economic development of a coun-try or region. Figure S9. The proportion of the five age groups for protein-energy malnutrition DALY cases between 1990 and 2019 globally, and in high, high-middle, middle, low-middle, and low SDI quintiles. The populations were divided into five age groups: under 5, 5-14 years, 15-49 years, 50-69 years, and 70+ years. SDI values can be used to judge the degree of economic development of a country or region. Figure S10. Joinpoint regression analysis of the age-standardized prevalence rates for PEM from 1990 to 2019. (A) Age-standardized prevalence rate in high SDI quintiles; (B) Age-standardized prevalence rate in high-middle SDI quintiles; (C) Age-standardized prevalence rate in middle SDI quintiles; (D) Age-standardized prevalence rate in low-middle SDI quintiles; (E) Age-standardized prevalence rate in low SDI quintiles. The prevalence of High SDI, High-middle SDI, Middle SDI quintiles increased in 1990-2011, and the Low-middle SDI, Low SDI decreased at the same time. And the prevalence of all SDI groups went through a U-shaped curve in 2011-2019. Figure S11. Joinpoint regression analysis of the age-standardized death rates for PEM from 1990 to 2019. (A) Age-standardized death rate in high SDI quintiles; (B) Age-standardized death rate in high-middle SDI quintiles; (C) Age-standardized death rate in middle SDI quintiles; (D) Age-standardized death rate in low-middle SDI quintiles; (E) Age-standardized death rate in low SDI quintiles. As for the jointpoint of deaths, High-middle SDI, Middle SDI, Low SDI decreased in 1990-2019 and the High SDI, Low-middle SDI de-creased in 2000-2019. Figure S12. Joinpoint regression analysis of the age-standardized DALY rates for PEM from 1990 to 2019. (A) Age-standardized DALY rate in high SDI quintiles; (B) Age-standardized DALY rate in high-middle SDI quintiles; (C) Age-standardized DALY rate in middle SDI quintiles; (D) Age-standardized DALY rate in low-middle SDI quintiles; (E) Age-standardized DALY rate in low SDI quintiles. The jointpoint of all SDI groups except the High SDI group decreased by the time. Table S1. The death cases and age-standardized death of PEM in 1990 and 2019, and its temporal trends from 1990 to 2019. Table S2. The DALY cases and age-standardized DALY of PEM in 1990 and 2019, and its temporal trends from 1990 to 2019.

Institutional Review Board Statement:
The data used in this article were from the GBD public database and do not require institutional review board approval or other ethics approval or consent of the study subjects.

Informed Consent Statement:
The institutions that conducted the surveys were responsible for obtaining informed consent from the participants. Data Availability Statement: Data are available in a public, open access repository. All of the data are publically available. Data are available on request.