putting child kwashiorkor on the map

CMAM FORUM TECHNICAL BRIEF, MARCH 2016
PUTTING
CHILD KWASHIORKOR
ON THE MAP
AUTHORS: Jose Luis Alvarez, Nicky Dent, Lauren Browne, Mark Myatt and André Briend
02
PUTTING
CHILD KWASHIORKOR
ON THE MAP
This report was produced by an independent expert group led by the CMAM Forum and funded by
UNICEF, ACF and ECHO (via the CMAM Forum).
Jose Luis Alvarez, Nicky Dent, Lauren Browne, Mark Myatt and André Briend.
Coordinated by Nicky Dent and Jose Luis Alvarez
CONTACT: please write to Jose Luis Alvarez at [email protected]
PROPOSED CITATION: Alvarez JL, Dent N, Browne L, Myatt M, and Briend A. Putting Child
Kwashiorkor on the map. CMAM Forum Technical brief. March 2016.
ACKNOWLEDGEMENTS
Core Mapping Group: UNICEF: Julia Krasevec, Diane Holland; WHO: Monika Blöessner, Zita Weise Prinzo; ACF-UK: Saul Guerrero.
Technical Advisory Group: CDC (Carlos Colorado-Navarro, Leisel Talley, Oleg Bilukha); CRED/University Uclouvain (Chiara Altare); Jimma
University (Tsinuel Girma); KEMRI (Jay Berkley); Mwanamugimu Nutrition Unit, Mulago Hospital, Uganda (Hanifa Numusoke) ; MSF,
ALIMA (Kevin PQ Phelan); Washington University in St. Louis (Mark Manary, Indi Trehan); Valid International (Paul Binns).
Many thanks for additional review comments from ACF (Benjamin Guesdon, Victoria Sauveplane); London School of Hygiene and
Tropical Medicine (Severine Frison); MSF (Kerstin Hanson, Saskia van der Kam); SCF (Jessica Bourdaire); UNICEF regional offices (Cecile
Basquin, Christiane Rudert, Helene Schwartz); University of Copenhagen (Henrik Friis, Pernille Kæstel); University of Southampton
(Alan Jackson); University of Westminster (Nidia Huerta Uribe).
For data sharing thanks to Government health and nutrition ministries of Burkino Faso (Bertine Dowrot Ouaro), Central Africa Republic
(Gisele Molomadon), Chad (Adoum Daliam), the Democratic Republic of Congo (Damien Ahimana, Jean Pierre Banea, Nicole Mashukano),
the Gambia (Samba Ceesay), Guinea (Mamady Daffe), Guinea Bissau (Ivone Menezes Moreira), Ivory Coast (Theckly Ngoran), Kenya
(Lucy Gathigi), Liberia (Kou Baawo), Sierra Leone (Aminata Koroma), Togo (Mouawiyatou Bouraima) and Nigeria Bureau of Statistics
(Isiaka Olarewaju); Afghanistan: UNICEF/Ministry of Public Health/ Agha Khan University; Guatamala (FEWS NET and ACF-Spain,
with USAID funds); Pakistan: IRC international Maryland USA, NIPS Islamabad Pakistan; the Philippines: Philippine National Nutrition
Cluster, the National Nutrition Council, UNICEF, and ACF Philippines.
All headquarters and country offices including the following who have helped access the data:
ACF (Benjamin Guesdon, Rachel Lozano, Danka Pantchova, Damien Pereyra, Victoria Sauveplane); ACF intern (Kaiser Esquillo, Sabine
Appleby); ALIMA (Géza Harczi, Ali Ouattara, Susan Shepherd); College of Medicine, Department of Paediatrics, University of Malawi,
Blantyre, Malawi (Emmanuel Chimwezi, Wieger Voskuijl); Concern Worldwide (Kate Golden, Ros Tamming); ECHO (David Rizzi) FEWS
NET (Gilda Maria Walter Guerra, Christine McDonald); Food Security and Nutrition Analysis Unit/FAO (Nina Dodd, Rashid Mohamed);
GOAL (Amanda Agar); International Medical Corps (Caroline Abla, Suzanne Brinkman, Amelia Reese-Masterson) ; International Rescue
Committee (Bethany Marron, Casie Tesfai) ; KEMRI (Jay Berkley, Kelsey Jones); LSHTM (Severine Frison); Médecins Sans Frontières
(Kerstin Hanson, Kevin PQ Phelan, Saskia van der Kam); MRC International Nutrition Group and MRC The Gambia Unit (Helen
Nabwera); Plan International (Unni Krishnan); Save the Children (Christoph Andert, Jessica Bourdaire); Terre des Hommes (Charulatha
Banerjee); United Nations High Commissioner for Refugees (Vincent Kahi, Eugene Paik Caroline Wilkinson, Joelle Zeitouny); UNICEF
(Victor Aguayo, Bulti Assaye, Arshidy Awale, Fanceni Balde, Amina Bangana, Faraja Chiwile, Patrick Codjia, Nguyen Dinh Quang,
Martin Eklund, Katherine Faigao, Denis Garnier, Lucy Gathigi Maina, Rene Gerard Galera, Aashima Garg, Mariama Janneh, Vandana
Joshi, Angela Kangori, Wisal Khan, James Kingori, Edward Kutondo, Chirchir Langat, Anne-Sophie Le Dain, Leo Matunga, Bonaventure
Muhimfura, Grainne Moloney, Mueni Mutunga, Simeon Nanama, Mamadou Ndiaye, Mara Nyawo, Jecinter Akinyi Oketch, Lucy Oguguo,
Magali Romedenne, Christiane Rudert, Kalil Sagno, Maria Claudia Santizo, Lilian Selenje, Flora Sibanda-Mulder, Ismael Ngnie Teta, Noel
Marie Zagre); World Vision International (Sarah Carr, Colleen Emary, Simon Karanja, Tim Roberton); Zerca y Lejos (Patricia Postigo y
Mamen Segoviano). Particular thanks to Helene Schwartz and Sara Gari-Sanchis of the UNICEF West Africa Regional Office.
PUTTING
CHILD KWASHIORKOR
ON THE MAP
03
CONTENTS
ABBREVIATIONS
05
EXECUTIVE SUMMARY
06
KEY FINDINGS
08
INTRODUCTION 09
1 BACKGROUND
1.1 An introduction to kwashiorkor 10
10
2 METHODOLOGY
13
2.1 Supervision and technical support
13
2.2 Data collection, management and analyses
13
2.3 Admission data 14
3 RESULTS FROM SURVEYS
15
3.1 Data received 15
3.2 Estimated oedema prevalence in affected areas by country
18
3.3 Proportion of oedema among SAM cases by country 20
3.4 Oedema prevalence and oedema as a proportion of SAM cases by age and sex
22
3.5 Distribution of MUAC and WHZ among oedema cases 24
4 ADMISSION AND OUTCOME DATA
26
4.1 Proportion of kwashiorkor among SAM cases admitted to therapeutic programmes
26
4.2 Mortality
29
4.3 Seasonality
32
5 COMPARISON OF SURVEY AND ADMISSION DATA
34
6DISCUSSION
36
6.1 Limitations, data issues and biases
36
6.2 The wider context of kwashiorkor
38
7RECOMMENDATIONS 39
7.1 Programmatic recommendations
39
7.2 Research priorities
40
8 CONCLUSION 42
REFERENCES
43
ANNEXES
44
Annex 1: Project information sheet shared with partners Annex 2: Letter of agreement 46
Annex 3: Surveys by contributing agency 47
Annex 4: Additional maps 48
Annex 5: Description of data sources
49
Annex 6: Survey dataset
49
Annex 7: UNHCR admission data 50
Annex 8: Other tables
51
44
04
PUTTING
CHILD KWASHIORKOR
ON THE MAP
MAPS
MAP 1
Oedema Prevalence 2006-2015
MAP 2
Proportion of SAM cases defined by MUAC <115mm or WFH<-3 or oedema with
kwashiorkor 2006-2015 MAP 3
Proportion of SAM cases defined by MUAC <115mm or oedema with kwashiorkor 2006-2015 18
20
20
FIGURES
FIGURE 1
FIGURE 2
FIGURE 3
FIGURE 4
FIGURE 5
FIGURE 6
FIGURE 7
FIGURE 8
Number of publications with “Kwashiorkor” or “oedematous malnutrition”
as keyword since 1945
Data filtering process Distribution of MUAC and WHZ in children with and without oedema ROC curves comparing the association of MUAC and WHZ with oedema KEMRI, Kilifi Hospital, Kenya, inpatient mortality data
Percentage of Kwashiorkor among SAM cases. Malawi National CMAM Data: 2010-2015
Uganda Mwanamugimu Nutrition Unit, Mulago Hospital, Inpatient Admissions: 2009-2014 Survey-admission scatter plot for global admission data 11
16
24
25
29
32
33
34
TABLES
TABLE 1
TABLE 2
TABLE 3
TABLE 4
TABLE 5
TABLE 6
TABLE 7
TABLE 8
TABLE 9
TABLE 10
TABLE 11
TABLE 12
TABLE 13
TABLE 14
TABLE 15
Data Sources for surveys received
Country breakdown of surveys, dates, and sample sizes
Oedema Prevalence 2006-2015
Proportion of SAM cases with kwashiorkor
2006-2015 listed in descending order of prevalence
Oedema prevalence by age and by sex 1990-2015
Oedema as a proportion of SAM cases defined by MUAC or oedema by age
and by sex 1990-2015
Oedema as a proportion of SAM cases defined by WHZ and oedema by age
and by sex 1990-2015
Country breakdown of median MUAC ranges among children with oedema
Admission data provided by agency
Overall summary of Kwashiorkor percentages by year and by country 2006-2015
ALIMA programme mortality, January 2011 - April 2015
ALIMA admission data from N’Djamena, Chad, 2014
Mortality by category, community-based programme, Malawi, 2009-2011, Trehan et al Broad analysis of admissions to health facilities in Haut Artibonite, Haiti 2009 to 2013,
demonstrating percentage of admissions with oedema Limitations of survey and admission data 15
17
19
21
22
23
23
25
26
27
30
31
31
35
36
PUTTING
CHILD KWASHIORKOR
ON THE MAP
05
ABBREVIATIONS
CAR
CFR
CI
CMAM
CSV
DHS
DRC
HAZ
IDP
MAM
MICS
MoH
MUAC
NGO
RAF
ROC
SAA
SAM
SMART
UNHCR
UNICEF
WHO
WHZ
Central African Republic
Case Fatality Rate
Confidence Interval
Community-based Management of Acute Malnutrition
Comma-Separated-Value
Demographic and Health Surveys
Democratic Republic of the Congo
Height-for-Age z-score
Internally Displaced Person
Moderate Acute Malnutrition
Multiple Indicator Cluster Survey
Ministry of Health
Mid Upper Arm Circumference
Non-Governmental Organisation
R Analytic Flow
Receiver Operating Characteristic
Sulphur Amino Acids
Severe Acute Malnutrition
Standardized Monitoring and Assessment of Relief and Transitions
United Nations High Commissioner for Refugees
United Nations Children’s Fund
World Health Organization
Weight-for-Height z-score
06
PUTTING
CHILD KWASHIORKOR
ON THE MAP
EXECUTIVE SUMMARY
Putting Kwashiorkor on the Map started as a call for sharing data to give an idea of prevalence and raise the profile
of kwashiorkor. In order to help fill data gaps and obtain a more comprehensive understanding of the global situation
for kwashiorkor, Phase Two of the project was launched in September 2014 with funding assistance from UNICEF.
A Kwashiorkor Mapping Core Group was established to manage the project outputs including data collection,
interpretation and documentation.
The aims of Phase Two are:
1
2
To refine and update the initial kwashiorkor map, provide a broad estimate of the numbers
and location of cases of kwashiorkor and identify high burden countries/areas.
To strengthen the evidence base and support advocacy for inclusion of kwashiorkor in relevant
methodology discussions at global level.
Extent of the problem
This report highlights the importance of kwashiorkor as a public health problem, as reflected by its prevalence and
also by the proportion of SAM cases it represents in surveys. Kwashiorkor, is an acute condition, and standard crosssectional surveys are not adapted to assess the real importance of this problem. The high proportion of kwashiorkor
reported among SAM children admitted for treatment in some areas where its prevalence is low shows the difficulty
in assessing the extent of this problem. For example, the reported prevalence of oedema during the last ten years was
less than 1% in most of the countries were data was available but when examining the estimate proportion of SAM
cases with kwashiorkor, figures ranged between 50% in Malawi, to 32% in the Democratic Republic of Congo and just
1.6% in Pakistan. This suggests that kwashiorkor is probably far more extensive than what cross-sectional surveys
show. Certain types of studies, such as incidence studies or community studies with regular active case finding, may
be better suited to more accurately describe the burden of oedema in countries.
Distribution of kwashiorkor
Despite its limitations, this report gives, for the first time, a representation of the geographic distribution of kwashiorkor,
based on 2,515 datasets with information on more than 1,736,000 individual children collected from 55 countries during
the time period 1992 to 2015. It shows that this form of malnutrition occurs most frequently in some parts of Africa,
specifically around the equator. This is consistent with what has been reported for more than 40 years in West Africa.
DRC is the highest burden country in the world with respect to oedema prevalence and surveys from a significant
number of countries in Africa indicated that more than a third of SAM cases defined by MUAC <115mm or oedema had
kwashiorkor, including Malawi, Rwanda, Zambia, Togo, and Cameroon. Notably, the data from Malawi estimated that
half of all SAM cases had kwashiorkor. Once again, Malawi, DRC, Haiti and Zambia were found to have some of the
highest rates of kwashiorkor admissions. Oedema prevalence was greatest in the youngest age groups (6-17 and 18-29
months) and no difference on prevalence was found among sexes but when SAM was defined by MUAC <115mm or
oedema, the proportion of males with oedema among those with SAM was consistently higher than that of females.
Association with background malnutrition and with mortality
This report also highlights the high variation of malnutrition associated with oedema. Many oedematous children
would not be classified as having SAM if only MUAC or WHZ were considered.
Arguably, the interpretation of nutritional status with WHZ is flawed in children with oedema due to weight increase
caused by oedema, but this report shows that oedema cases often have low WHZ (the median WHZ score for children with
oedema was -1.55 and for those without oedema, -0.62). MUAC measures also tended to fluctuate in generalised oedema.
PUTTING
CHILD KWASHIORKOR
ON THE MAP
07
MUAC is less sensitive to changes in hydration status and seems better for assessing the general nutritional status
of children with oedema that does not extend up to the child’s upper arms (i.e., +++ oedema). This latter assumption
is supported by the ROC curves in this report that describe the association between anthropometry and oedema,
showing that MUAC more readily identifies children with oedema, compared to WHZ.
Mortality associated with kwashiorkor also varies across studies, with some reporting lower, identical or higher
mortality compared to non-oedematous malnutrition. These discordant observations may be related to a different
level of associated malnutrition. In children with SAM, the presence of oedema is considered as an aggravating factor
associated with a higher risk of death as reported by a number of studies but some of the patterns analysed may be
indicative of no association. However, the lack of actual and reliable data hinders the assessment and comparison of
the mortality rates between the 3 types of SAM, as well as the identification of prognostic factors that could guide the
treatment of these patients.
Poor association between prevalence surveys and admission data
Another important finding of this report is that standard cross-sectional surveys do not adequately reflect the clinical
importance of kwashiorkor, since there appears to be a lack of a relationship between admission data and kwashiorkor
prevalence obtainedd from surveys. The possible reasons for this discrepancy are many and should be explored. A
possibly poorly adapted survey methodology with insufficient standardisation for collection of oedema data should
be considered first. The recommendation of national protocols in terms of referral to inpatient care, the existence of
community-based management of acute malnutrition in the country or the level of community mobilisation activities
were not assessed in relation to each annual national admission dataset. It is therefore uncertain whether variations
in the level of inpatient care for oedematous children were due to country policy or severity of cases.
It is possible that duration of oedematous malnutrition is not the same in different settings, in particular as a result of
the very different degree of associated malnutrition. This may have an influence on the associated mortality and/or on
the rapidity of recovery, both of which can have an influence on the probability of finding oedematous cases during a
nutritional survey.
High variations of association with background malnutrition and with mortality and poor association between results
of prevalence surveys and admission data are factors that could also be related to the shift in treatment with the
introduction of community-based management of acute malnutrition (CMAM) around 2005, where oedema + or ++
(variation between agencies) moved from being treated only in an inpatient setting to being treated in an outpatient
setting. Ideally inpatient admissions could be interpreted on a timescale in relation to the country protocols at that time,
but given the historical data used, exact admission protocols at the time were not available.
Data collection and standardization
It is made clear throughout the report, that better collection methodologies on kwashiorkor data and improvements to
the current survey reporting systems are needed. Additionally, a global database that includes admissions for oedema
should be included in each country’s surveillance system.
The high proportion of kwashiorkor reported among SAM
children admitted for treatment in some areas where its
prevalence is low shows the difficulty in assessing the extent
of this problem and suggest that kwashiorkor is probably
far more extensive than cross-sectional surveys show.
08
PUTTING
CHILD KWASHIORKOR
ON THE MAP
Key Findings:
01
There is a need for better, standardised and routine data collection
to correctly identify the burden of kwashiorkor across the world.
02
High rates of oedema were reported in Central and Southern Africa,
as well as Haiti (DRC, Yemen and Zimbabwe reported prevalence
rates between 1-2%). When examining the proportion of SAM cases
with kwashiorkor, surveys from a significant number of countries
indicated that more than a third of SAM cases had kwashiorkor,
including Malawi, Rwanda, Zambia, Togo and Cameroon.
03
Mortality rates are highest among children who have MUAC <115mm
and oedema (marasmic kwashiorkor).
04
Oedema prevalence was greatest in the youngest age groups
(6 - 17 and 18 - 29 months).
05
There is a need for more studies to provide evidence-based prognoses
and specific treatment recommendations relating to kwashiorkor.
PUTTING
CHILD KWASHIORKOR
ON THE MAP
09
INTRODUCTION
Putting child kwashiorkor on the map initiative
Putting Kwashiorkor on the Map28 started as a call for sharing data to give an idea of prevalence and to raise the profile
of kwashiorkor. In Phase One of the project a short commentary and map of “kwashiorkor” based on a database of
557 surveys (from 1992-2006) compiled by Brixton Health was released in October 2013 by the CMAM Forum (http://
www.cmamforum.org). The initial data outputs indicated that there is a problem of high caseloads or prevalence of
oedematous malnutrition, although its distribution could not be ascertained due to significant gaps in data. Phase One
also led to the establishment of an informal Technical Advisory Group (TAG) to define parameters for any future data
collection and liaison with organisations that were willing to share data.
In order to help fill data gaps and obtain a more comprehensive understanding of the global situation for kwashiorkor,
Phase Two of the project was launched in September 2014 with funding assistance from UNICEF, ACF and ECHO
(via the CMAM Forum) to further compile existing information on kwashiorkor. A Kwashiorkor Mapping Core Group
was established to manage the project outputs including data collection, interpretation and documentation. This was
comprised of representatives from UNICEF and WHO nutrition departments, ACF operations department and the CMAM
Forum, with tasks shared between members. The TAG group was continued and data sourcing was extended to key
international non-govermental organisations (NGO), United Nations and governments involved with conducting surveys
or nutritional programmes.
The overall goal of the project is to draw attention to kwashiorkor and stimulate improved data collection and further
research to ultimately support better detection or management of these children. This step is to assemble the information
on kwashiorkor that currently exists. The audience includes practitioners, researchers, academics and policy makers.
The aims of Phase Two were:
1
2
To refine and update the initial kwashiorkor map, provide a broad estimate of the numbers
and location of kwashiorkor and identify high burden countries/areas.
To strengthen the evidence base and support advocacy for inclusion of kwashiorkor in relevant
methodology discussions at global level.
Specific questions to be answered using any combined datasets included:
◆ Geographic distribution of kwashiorkor with a more robust and recent dataset
◆ Proportion of SAM cases that have bilateral pitting oedema
◆ Comparison of kwashiorkor prevalence co-incidental to low MUAC and to low WHZ
◆ Concurrence of kwashiorkor and low MUAC
While the limitations of current data collection tools to analyse the burden of an acute condition such as kwashiorkor
are recognised, it was considered that collation of a large number of surveys from different areas and settings, and
including recent national surveys, would contribute to learning more about the global situation and help assess where
we are at the current time. The intiative also assembled admission data, where possible. It is hoped that assembling
existing information could help highlight potential trends, identify countries with higher and lower burdens and
encourage more rigorous and standardised data collection in the future, stimulate research and generally raise
attention to child kwashiorkor.
10
PUTTING
CHILD KWASHIORKOR
ON THE MAP
1 BACKGROUND
1.1 An introduction to kwashiorkor
Severe acute malnutrition (SAM) is currently defined by the World Health Organization (WHO) and the United Nations
Children’s Fund (UNICEF) by a mid-upper arm circumference (MUAC) less than 115mm or by a weight-for-height Z-score
(WHZ) less than -3 or by presence of bilateral pitting oedema (WHO UNICEF Joint statement 2009). This report focusses
on bilateral pitting oedema and assembling information on the prevalence and outcomes of children with this sign.
The WHO ICD 10 classification1,2 defines kwashiorkor as a “form of severe malnutrition with nutritional oedema with
dyspigmentation of skin and hair.” We use the term kwashiorkor and oedema for forms of SAM associated with bilateral
pitting oedema (referred to as oedema in this report), without necessarily including associated dyspigmentation of skin
and hair or any other clinical signs. We chose this definition because it is frequently used in-country, specifically for
admission to therapeutic programmes, and it is the simplest clinical sign of kwashiorkior that can be assessed. Skin
dyspigmentation, on the other hand, may be slight and remain unnoticed, and hair changes require a longer time
to take place, as noted by Cicely Williams3. This approach was also adopted by the Wellcome Trust, which defines
kwashiorkor based only on the presence of oedema4. It is acknowledged that this condition is also referred to as
“nutritional oedema” or “oedematous malnutrition.”
The Wellcome classification also introduced the term “marasmic kwashiorkor” for children with oedema and a weightfor-age indice less than 60%. This index is difficult to interpret in the presence of oedema and is not routinely measured
in most nutritional programmes.
Part of the problem of
identifying kwashiorkor
is that this condition is
often transient, meaning
that children usually
spontaneously recover
or die within a few
days or weeks of onset
Kwashiorkor is therefore a clinical syndrome indicated by the
typical onset of bilateral pitting oedema in the lower limbs that can
gradually spread to the upper sections as it becomes more severe.
The grade of oedema has been classified5 as + mild: both feet; ++
moderate: both feet, plus lower legs, hands, or lower arms; +++
severe: generalised oedema including both feet, legs, hands, arms
and face. Other clinical features that may be found in these patients
include: apathy, irritability, fatty liver, “flaky-paint” dermatosis and
scanty lustreless hair6,7. A comprehensive characterisation of the
clinical signs and laboratory test abnormalities has not been done so far.
The condition affects hundreds of thousands of children every year in the poorest countries of the world, killing many of
them – and yet it seems not to attract global attention. Given the high mortality risk associated with this form of SAM, the
low level of understanding of kwashiorkor and limited new research are surprising. There is no mention of kwashiorkor in
the comprehensive implementation plan on maternal, infant and young child nutrition adopted by the 2012 World Health
Assembly, which sets global nutrition targets for 2025. The condition is also overlooked in the 2013 “Maternal and Child
Nutrition” Lancet series8, which does not acknowledge its importance in public health terms or mention the existence
of effective treatment that is capable of preventing many deaths every year. Likewise it is not mentioned in the Global
Nutrition Report 2014 and 20159.
Currently, there is no reliable estimate of the number of children suffering from kwashiorkor around the world. The last
published global map showing the prevalence of kwashiorkor was produced in 195410 and was done at a time when
the diagnostic criteria were more vague. It is not clear how this map was obtained, and in absence of reference to
PUTTING
CHILD KWASHIORKOR
ON THE MAP
11
FIGURE 1
NUMBER OF PUBLICATIONS WITH “KWASHIORKOR”
OR “OEDEMATOUS MALNUTRITION” AS KEYWORD (IN ENGLISH) SINCE 1945
140
Number of Publications
120
100
80
60
40
20
0
1945
1950
1955
1960
1965
1970
1975
1980
1985
1990
1995
2000
2005
2010
2015
Year
community surveys, it may just be based on hospital records as was usual at that time. Oedema is not included in the
Joint Child Malnutrition Estimates compiled by UNICEF, WHO and the World Bank11, despite being a key diagnostic
and admission criterion to therapeutic services (MUAC as well is not included in these joint estimates, despite being
an important diagnostic and admission criterion). Similarly, cases of kwashiorkor are seldom documented in standard
national surveys, such as Demographic and Health Surveys (DHS) and Multiple Indicator Cluster Surveys (MICS). This
gap was highlighted in a letter to the Lancet in 201312 and more recently by the article: Omitting oedema measurement:
how much acute malnutrition are we missing, defining prevalence, overlap between oedematous malnutrition and
wasting and regional estimates13. This article, through an assessment of 852 cross-sectional surveys (conducted from
1992-2011, with 95% from 2000) estimates that “the median prevalence of edema cases was very low (0.2%) in all
regions except Central and South Africa (0.6%). The mean was also fairly low in all regions (0.6% or less) except in
Central and South Africa (1.2%).” It also highlights that “though these prevalences may represent a large proportion
of SAM cases two-thirds of oedema cases are missed by measuring wasting only and over 80% missed by assessing
severe wasting.”
Part of the problem of identifying kwashiorkor is that this condition is often transient, meaning that children usually
spontaneously recover or die within a few days or weeks of onset, and due to this short duration, kwashiorkor is poorly
captured by cross-sectional surveys commonly used to assess the importance of malnutrition. This problem was clearly
described in 1972 by Cicely Williams14 with kwashiorkor as a form of protein calorie malnutrition (PCM): “Acute cases
of P.C.M. are not often seen on surveys; they are more likely to arrive as outpatients or in hospital (The more chronic
the disease the more likely it is to be identified in surveys.)”. In addition, some surveys do not include oedema as an
indicator. For those that do, inexperienced surveyors, lack of training and poor follow up of oedematous cases treated
elsewhere during the survey are likely to contribute to underdiagnosis of oedema.
12
PUTTING
CHILD KWASHIORKOR
ON THE MAP
While there is indirect evidence that the number of children suffering from kwashiorkor has declined over the last 30
years or more, possibly in parallel with the reduction in infectious diseases (especially the increased coverage of measles
immunisation)15,16, the public health significance of kwashiorkor and its associated morbidity and mortality warrant
further attention, especially since kwashiorkor has been documented as the most common form of SAM in some parts of
Africa. 12,17,18, 19
Much of the research around kwashiorkor goes back some 30 or 40 years. As stated in the recent review: Kwashiorkor:
still an enigma – the search must go on20, kwashiorkor was the topic of high quality research and intense debate in the
1950-1970s, but “surprisingly, it later became an ‘orphan disease,’ with very few groups actively involved in disentangling
its mechanism and decreasing number of publications over the years…” (see Figure 1 above). While recent research into
kwashiorkor in Malawi has been promising21, significant gaps remain, with even the aetiology and physiopathology of
the condition poorly understood.
There is generally a lack of concrete up-to-date data to demonstrate whether mortality is higher or lower for children with
oedema compared to acutely malnourished children with status assessed by WHZ (<-3) or MUAC (<115mm), although
kwashiorkor (especially marasmic kwashiorkor) is still anecdotally associated with higher rates of hospitalisation and
higher mortality. Some studies suggest presence of oedema is associated with a higher risk of death 22,23,24,25, whereas
one community study from Malawi suggests a lower mortality risk among kwashiorkor children compared with children
with non-oedematous malnutrition26.
Kwashiorkor cases are more frequently admitted to inpatient or hospital care, which is more costly to caretakers and
health staff and national health systems, and increases the risk of contracting nosocomial diseases, indirectly increasing
risk of mortality. This could be due to national guidelines, some of which indicate that all grades of oedema should
be treated as inpatients, whereas others recommend treating oedema grade + and ++ in outpatient programmes, as
recommended in the latest WHO recommendations on treatment of SAM27. The higher frequency of referral to hospital
could also be due to late identification. While few sociological studies could be found, anecdotally kwashiorkor is often
not associated by communities as being a nutritional problem and is frequently misdiagnosed or undetected as a form
of SAM by health workers. This was often flagged as an issue in various coverage surveys analysing barriers to uptake
of care (uploaded on the Coverage Monitoring Network website, http://www.coverage-monitoring.org/ ). The higher
rate of hospital admission could also be due to associated complications due to late referral.
Much of the research on oedema goes back some 30 to 40
years and hence, significant gaps exist in understanding
the burden of kwashiorkor around the world
PUTTING
CHILD KWASHIORKOR
ON THE MAP
13
2 METHODOLOGY
2.1 Supervision and technical support
Roles and responsibilities were established within the Kwashiorkor Mapping Core Group (ACF, CMAM Forum, WHO
and UNICEF) and external consultants to manage agreements with NGOs, data storage on a secure server, contractual
arrangements, coordination, communication and liaison with data contributors, construction of the database, the
analysis and the reporting work. Particular care was taken to ensure that raw survey data were cleaned and not
duplicated. Technical consensus was generated on relevant points for data management.
The TAG guided the type of information to be collected, the database construction, the analyses and the final report.
This was made up of representatives from CDC; CRED/University Uclouvain; Jimma University Ethiopia; KEMRI;
Mwanamugimu Nutrition Unit, Uganda; Médecins Sans Frontières (MSF); Washington University in St. Louis and Valid
International. The report was also shared with additional experts and data contributors for review.
2.2 Data collection, management and analyses
The CMAM Forum, with ACF, put out a request for sharing anthropometric surveys to NGOs, United Nations (UNICEF,
WHO, UNHCR, WFP) and governments (directly or via UNICEF) involved with nutrition programmes. In addition, a
request for any admission and outcome data or case studies distinguishing cases of kwashiorkor was made. Requests
were accompanied by an information sheet of the project (see Annex 1) and a data sharing letter of agreement which was
signed between agencies and ACF-UK (for NGOs) or a signed permission letter, facilitated by UNICEF (for government
data) (see Annex 2).
From January to September 2015, all anthropometric datasets collected and deemed eligible were collated in a central
database constructed by the data management team with feedback from the TAG. Datasets were added to an existing
database of 557 datasets held by Brixton Health and used in Phase One.
The raw data of any nutritional survey adopting the Standardized Monitoring and Assessment of Relief and Transitions
(SMART) methodology or similar with population proportional to size (PPS) or exhaustive sampling, simple random
sampling, or systematic sampling and including the key variables of age, sex, weight, height, MUAC and oedema
were utilised for the project. MICS or DHS datasets were acceptable if they collected all of the variables required and
PPS sampling was used within strata. Age was limited to 6-59 months, since only a very small number of surveys
include children 0-6 months and MUAC is not usually taken on infants under 6 months old.
Raw datasets were used in preference to narrative reports to enable standardisation and avoid transcription errors.
Narrative reports were also requested to provide additional metadata.
A database management system was developed. This provides a range of functions including data import, calculation of
anthropometric indices, flagging of extreme values by SMART or WHO flagging criteria, checking for duplicate datasets,
reporting on data quality, setting inclusion criteria for data analysis, data analysis, mapping and reporting.
The datasets received were all converted to CSV (comma-separated values text file) format, stripped of unnecessary
variables, restructured to the common format and cleaned according to the pre-determined standards. Variables were
re-coded as needed to ensure that the codes were consistent throughout the datasets. Where necessary, measurement
units were converted (e.g. MUAC records changed from centimetres to millimetres).
14
PUTTING
CHILD KWASHIORKOR
ON THE MAP
Datasets with any of the key variables (data columns) missing, besides that of the cluster identifiers, were not included
in the database. For each dataset eligible for inclusion, any obvious data entry errors were fixed or deleted, and
extreme values were also removed.
If the validity of any survey’s data was questioned (e.g. due to a large number of repeated records), the affected dataset
was removed from the database. An analysis to determine databases with abnormal results for oedema (i.e. outliers)
was implemented; given that only 1 survey was classified as an outlier the authors decided to include all surveys in
the analysis.
The R Analytic Flow (RAF) scientific workflow software was used to organise the database. Using the R Language and
Environment for Statistical Computing, scripts were written to import, restructure, check, calculate anthropometric
indices, identity duplicate datasets and analyse the data and produce maps, tables, and graphs showing the proportion
of SAM cases that have bilateral pitting oedema.
In order to provide a representative figure for each country, maps and tables were generated taking into account the
following considerations:
◆ SUBNATIONAL SURVEYS VS NATIONAL SURVEYS: National surveys, when stratified by regional
level with the necessary information provided, were separated into datasets by region.
◆ BORDER DETERMINATION: Use of geolocation was not possible due to coordinate data not being
provided by the vast majority of surveys. Where historical national and/or regional divisions differed
from current divisions, the name of the survey site was utilised to help determine its current location.
If the location could not be accurately determined, its location was described as “unspecified”.
◆ COUNTRIES WITH NO OR LITTLE DATA: If no data was received, the affected countries are
marked on maps with grey colour to indicate uncertainty regarding the national nutrition situation with
respect to oedema. The countries whose surveys did not identify at least 20 SAM cases (according to
the definition of SAM employed in each analysis) were excluded from the any map analysis requiring
calculation of the number of cases of SAM to avoid distorting the map, although the numbers are kept
in the tables.
◆ WHO FLAGS were used to remove records with outlying indices, but only for the analyses which
concerned that specific indice: height-for-age (HAZ)<-6.0 or HAZ>+6.0, WHZ<-5.0 or WHZ>+5.0.
Extreme MUAC values were censored using MUAC/A Z-scores (MAZ). Records with MAZ < -5 or MAZ
> +5 were removed. Records were excluded from particularly analyses bases on suspect values in the
relevant anthropometric index identified using WHO flagging criteria.
2.3 Admission data
Agencies were requested to share programme admission data if available, segregated by oedematous and nonoedematous
cases. These data were then combined to give total numbers of cases with oedema, total numbers of children with SAM and
the percentage of oedematous admissions for each year of available data per country. Data were separated into inpatient
admissions and outpatient admissions where available. Refugee information was recorded separately given the population
differences (for example, in a refugee context, coverage and thus representativeness of the admission data are supposed
to be much better than in the other contexts). Associated information of what national protocol was in place at the time of
the programme (ie pre or post CMAM) or what level of community mobilisation and case finding was operational was not
provided given the historical nature of most of the data and this may explain some of the high admission rates.
PUTTING
CHILD KWASHIORKOR
ON THE MAP
15
3 RESULTS FROM SURVEYS
3.1 Data received
1,958 eligible datasets were collected from 15 national Governments/UNICEF and 11 NGOs, FEWS NET, and two United
Nations bodies. These were mostly SMART surveys, with only one MICS survey included. These were added to the
original 557 datasets already held by Brixton Health and used for the mapping in Phase One to give a total of 2,515
datasets, with information on more than 1,736,000 individual children for analysis (for additional information on data
sources see Annexes 3, 5 and 6).
TABLE 1
DATA SOURCES FOR ELIGIBLE SURVEYS RECEIVED
Agency
# datasets
# reports
# countries
(total)
ACF
814363 40
Concern Worldwide
1086811
FEWS NET
221
FSNAU
20701
GOAL
14107
IMC
1502
IRC
302
MSF
95 025
Plan International
201
Save the Children
58
2
8
Terre des Hommes
7
0
3
UNHCR
193170 23
UNICEF/Government
Afghanistan, Burkina Faso, Central African Republic, Chad,
the Democratic Republic of Congo, the Gambia, Guinea, Guinea Bissau, Ivory Coast, Kenya, Nigeria, Pakistan, the Philippines, Sierra Leone, Togo
640220 22
World Vision
1105
Zerca y Lejos (local NGO)
TOTAL No OF UNIQUE SURVEYS
1
0
2277*
825
Proportion of datasets that also shared narrative reports
1
55 unique countries
36.2%
*2297 but 20 were partnerships between multiple organisations
16
PUTTING
CHILD KWASHIORKOR
ON THE MAP
Exclusion criteria were applied to the datasets received (see Figure 2). Most of the excluded datasets were missing
MUAC data. Any duplication of the surveys received was checked for and duplicate datasets were removed from the
database. WHO flags were used, with <1% of records flagged.
FIGURE 2
DATA FILTERING PROCESS (SURVEYS AND DATA INCLUDED IN ANALYSES WITH REASONS FOR EXCLUSION)
2,515
Reason for dataset
#
%
exclusion (varible(s) missing) datasets
MUAC
DATASETS RECEIVED
2,350 165
ELIGIBLE DATASETS
93.44%
2,277
NONDUPLICATE
DATASETS
96.89%
1,725,200
UNFLAGGED RECORDS
ALL
9859
7
4
11
18
7
11
29
18
165100
INELIGIBLE DATASETS
6.56%
73
TRUE DUPLICATE
DATASETS
3.11%
10,847
FLAGGED RECORDS
99.38%
0.62%
6,996
114
COMPLETE OEDEMA
CASE RECORDS
MUAC & oedema
Sex, weight
&/or height
Unknown codes
Inadequate dataset
info provided
INCOMPLETE OEDEMA
CASE RECORDS
Details on data received per country for all years can be seen in table 2.
Reason for individual
record exclusion
#
%
children
(varible(s) missing)
MUAC
MUAC weight
&/or height
Weight &/or height
Age
ALL
8776
8
7
17
15
22
114100
PUTTING
CHILD KWASHIORKOR
ON THE MAP
17
TABLE 2
COUNTRY BREAKDOWN OF SURVEYS, DATES, AND SAMPLE SIZES (ALL YEARS)
Country
Afghanistan
Albania
Angola
Bangladesh
Benin
Bolivia Botswana Burkina Faso Burundi Cameroon Central African Republic Chad Congo, Democratic Republic Cote d’Ivoire Djibouti Eritrea
Ethiopia
The Gambia
Guatemala
Guinea
Guinea-Bissau
Haiti
India
Indonesia
Jordan
Kenya
Liberia
Macedonia
Madagascar
Malawi
Mali
Mauritania
Mozambique
Myanmar
Nepal
Nicaragua
Niger
Nigeria
Pakistan
Philippines
Rwanda
Senegal
Sierra Leone
Somalia
South Sudan
Sri Lanka
Sudan
Tajikistan
Tanzania
Thailand
Togo
Uganda
Yemen
Zambia
Zimbabwe
55 COUNTRIES
# Surveys
Earliest
survey
Most recent
survey
Total # children
measured
43*1995 2013
11999 1999
221993 2002
262009 2014
7*2008 2014
3
2011 2011 1
2013 2013 50 2008 2014 25 1994 2013 9
1998 2013 58 1993 2015 201 1994 2015 264 1994 2014 49 1994 2014 7
2008 2014 32001 2014
2331994 2015
82012 2012
22015 2015
121995 2012
132008 2012
491994 2014
82008 2014
31999 2002
22014 2014
1072000 2014
521993 2013
11999 1999
42001 2014
161994 2012
142007 2011
562007 2014
111992 2012
222000 2015
122006 2014
2Unspecified Unspecified
382001 2011
1072005 2015
182000 2012
122010 2015
211993 2012
72012 2012
58*1994 2014
2271993 2015
1401993 2014
31997 2002
1361996 2015
51999 2004
71995 2014
22004 2004
18*2009 2014
741996 2014
22013 2014
52009 2013
12009 2009
48,878
906
17,361
13,480
7,930
1,775
164
40,446
14,742
5,642
36,443
124,096
227,390
24,233
2,516
1,969
155,494
6,769
625
9,603
7,216
39,764
5,182
1,749
802
71,475
31,230
865
3,180
16,277
10,968
36,432
3,867
14,391
7,650
1,017
49,411
66,398
14,200
6,220
13,534
8,445
64,028
237,498
96,959
2,586
109,099
4,337
4,903
1,812
11,976
48,503
816
2,095
700
2277 1992 20151,736,047
* includes 1 national survey that could not be divided into regions
18
PUTTING
CHILD KWASHIORKOR
ON THE MAP
3.2 Estimated oedema prevalence in affected areas by country
The reported prevalence of oedema in these surveys during the last ten years was less than 1% in most of the countries
where data was available. Some countries in Central and South Africa, as well as Haiti (the Caribbean), reported higher
rates, and Yemen, Zimbabwe and the Democratic Republic of Congo (DRC) reported rates between 1% and 2% (see Map 1
and Table 3). It must be taken into account that for some countries the number of surveys and children screened was very
limited in contrast to others and that surveys are usually conducted in areas with suspected nutritional problems and so
are not necessarily representative of the whole country. Only one survey was available from Zimbabwe, while more than
200 were available from each of the following countries: Chad, DRC, Ethiopia and Somalia.
MAP 1
OEDEMA PREVALENCE 2006-2015
The percentage of children with oedema is highest in
Zimbabwe (1.71), Yemen (1.10), The Democratic Republic
of Congo (1.04), and Zambia (0.81)
PUTTING
CHILD KWASHIORKOR
ON THE MAP
TABLE 3
OEDEMA PREVALENCE 2006 - 2015
Country
Afghanistan
Bangladesh
# Children
surveyed
# Oedema
cases found
16,021
81
% children with
oedema (95% CI)
0.51 (0.40, 0.63)
13,480
3
0.02 (0, 0.07)
Benin
7,930
5
0.06 (0.02, 0.15)
Bolivia
1775
3
0.17 (0.03, 0.49)
164
0
0 (0, 2.22)
40,446
50
0.12 (0.09, 0.16)
Botswana
Burkina Faso
Burundi
2,830
1
0.04 (0, 0.20)
Cameroon
4,312
33
0.77 (0.53, 1.07)
Central African Republic
36,005
153
0.42 (0.36, 0.50)
Chad
94,736
192
0.20 (0.18, 0.23)
194,334
2021
1.04 (1.00, 1.09)
21,742
79
0.36 (0.29, 0.45)
2,516
7
0.28 (0.11, 0.57)
347
0
0 (0, 1.06)
134,201
241
0.18 (0.16, 0.20)
6769
3
0.04 (0.01, 0.13)
625
3
0.48 (0.10, 1.40)
Guinea
7,932
4
0.05 (0.01, 0.13)
Guinea-Bissau
7,216
2
0.03 (0, 0.10)
Haiti
12,012
61
0.51 (0.39, 0.65)
India
5,182
5
0.10 (0.03, 0.23)
802
0
0 (0, 0.46)
Kenya
66,220
101
0.15 (0.12, 0.19)
Liberia
8,574
17
0.20 (0.12, 0.32)
Madagascar
2,280
0
0 (0, 0.16)
Malawi
5,645
44
0.78 (0.57, 1.04)
Mali
10,968
2
0.02 (0, 0.07)
Mauritania
36432
8
0.02 (0.01, 0.04)
Mozambique
1,085
2
0.18 (0.02, 0.66)
Myanmar
8,007
7
0.09 (0.04, 0.18)
Nepal
7,650
29
0.38 (0.25, 0.54)
Nicaragua
1,017
8
0.79 (0.34, 1.54)
0.03 (0.02, 0.06)
Congo, Democratic Republic
Cote d’Ivoire
Djibouti
Eritrea
Ethiopia
Gambia
Guatemala
Jordan
Niger
40,301
14
Nigeria
65,449
130
0.20 (0.17, 0.24)
Pakistan
6,545
4
0.06 (0.02, 0.16)
Philippines
6,220
0
0 (0, 0.06)
Rwanda
4,040
17
0.42 (0.25, 0.67)
0.04 (0.01, 0.10)
Senegal
8,445
3
32,755
121
0.37 (0.31, 0.44)
218,048
692
0.32 (0.29, 0.34)
44,457
114
0.26 (0.21, 0.31)
293
1
0.34 (0.01, 1.89)
40,417
117
0.29 (0.24, 0.35)
619
4
0.65 (0.18, 1.65)
Togo
11,976
42
0.35 (0.25, 0.47)
Uganda
26,079
52
0.20 (0.15, 0.26)
1.10 (0.51, 2.08)
Sierra Leone
Somalia
South Sudan
Sri Lanka
Sudan
Tanzania
Yemen
Zambia
Zimbabwe
816
9
2,095
17
0.81 (0.47, 1.30)
700
12
1.71 (0.89, 2.98)
19
20
PUTTING
CHILD KWASHIORKOR
ON THE MAP
3.3 Proportion of oedema among SAM cases by country
Map 2 shows these results when the WHO definition of SAM is used (MUAC, WHZ and oedema criteria) and map 3 shows
results when SAM is defined by a MUAC cut-off (<115mm) or presence of oedema. Similar results are obtained when
SAM is defined by WHZ <-3 or oedema (Annex 4). Country surveys with small sample sizes (<20) for SAM cases are not
included in Map 2 but are shown below the main table.
MAP 2
PROPORTION OF SAM CASES DEFINED BY MUAC <115mm OR WFZ<-3 OR OEDEMA WITH KWASHIORKOR
(2006-2015) (EXCLUDING COUNTRIES WITH <20 CASES OF SAM)
MAP 3
PROPORTION OF SAM CASES DEFINED BY MUAC <115mm OR OEDEMA WITH KWASHIORKOR (2006-2015)
(EXCLUDING COUNTRIES WITH <20 CASES OF SAM DEFINED BY MUAC AND OR OEDEMA)
PUTTING
CHILD KWASHIORKOR
ON THE MAP
21
TABLE 4
PROPORTION OF SAM CASES (DEFINED BY MUAC <115MM OR OEDEMA) WITH KWASHIORKOR (2006-2015)
LISTED IN DESCENDING ORDER OF PREVALENCE
Country
# SAM cases
Malawi
Rwanda
Zambia
Togo
Cameroon
Haiti
Congo, Democratic Republic
Cote d’Ivoire
Sierra Leone
Liberia
Central African Republic
Sudan
Djibouti
Somalia
South Sudan
Nepal
Kenya
Chad
Afghanistan
Uganda
Nigeria
Burkina Faso
Gambia
Ethiopia
Benin
Senegal
Guinea
India
Myanmar
Mauritania
Guinea-Bissau
Bangladesh
Niger
Pakistan
Mali
Madagascar
Philippines
# Oedema cases
% SAM cases with
oedema (95% CI)
86
4350.0 (39.0, 61.0)
34
1647.1 (29.8, 64.9)
38
1744.7 (28.6, 61.7)
102
4039.2 (29.7, 49.4)
91
3235.2 (25.4, 45.9)
154
5334.4 (27.0, 42.5)
6,012
1,926
32.0 (30.9, 33.2)
289
76
26.3 (21.3, 31.8)
530
119
22.5 (19.0, 26.3)
71
1622.5 (13.5, 34.0)
713
135
18.9 (16.1, 22.0)
706
11516.3 (13.6, 19.2)
47
714.9 (6.2, 28.3)
4,773
67514.1 (13.2, 15.2)
793
112
14.1 (11.8, 16.7)
212
2913.7 (9.4, 19.1)
787
10112.8 (10.6, 15.4)
1467
18512.6 (11.0, 14.4)
648
7812.0 (9.6, 14.8)
479
4910.2 (7.7, 13.3)
1233
1209.7 (8.1, 11.5)
530
48
9.1 (6.8, 11.8)
35
38.6 (1.8, 23.1)
2787 2207.9 (6.9, 9.0)
75
56.7 (2.2, 14.9)
48
36.3 (1.3, 17.2)
53
35.7 (1.2, 15.7)
101
55.0 (1.6, 11.2)
143
64.2 (1.6, 8.9)
236
73.0 (1.2, 6.0)
37
12.7 (0.1, 14.2)
114
32.6 (0.6, 7.5)
698
142.0 (1.1, 3.3)
247
41.6 (0.4, 4.1)
142
21.4 (0.2, 5.0)
49 00 (0, 7.3)
26 00 (0, 13.2)
COUNTRIES WITH <20 CASES OF SAM (IDENTIFIED BY MUAC<115MM)
Mozambique*
Nicaragua*
Guatemala*
Zimbabwe*
Yemen*
Bolivia*
Tanzania*
Sri Lanka*
Burundi*
Botswana*
Jordan*
Eritrea*
2
2100.0 (15.8, 100.0)
10
880.0 (44.4, 97.5)
4
375.0 (19.4, 99.4)
17
1270.6 (44.0, 89.7)
14
964.3 (35.1, 87.2)
5
360.0 (14.7, 94.7)
11
436.4 (10.9, 69.2)
5
1
20.0 (0.5, 71.6)
19
15.3 (0.1, 26.0)
1 00 (0, 97.5)
2 00 (0, 84.3)
2 00 (0, 84.2)
*not included in Map 2 due to less than 20 SAM cases found in country
22
PUTTING
CHILD KWASHIORKOR
ON THE MAP
Surveys from a significant number of countries in Africa indicated that more than a third of SAM cases defined by
MUAC <115mm or oedema had kwashiorkor, including Malawi, Rwanda, Zambia, Togo, and Cameroon. Notably, the
data from Malawi estimated that half of all SAM cases had kwashiorkor.
Furthermore, when SAM was defined by WHZ <-3 or oedema, Malawi, Rwanda and Zambia were also in the top five
for the countries with the highest percentages of SAM cases that had kwashiorkor. DRC also found a high percentage
of SAM children with kwashiorkor when SAM was defined this way (See Annex 4).
3.4 Oedema prevalence and oedema as a proportion of SAM cases
by age and sex
The prevalence of kwashiorkor was analysed by age and sex (see Table 5). Oedema was present in all age groups,
especially among children under 30 months, with a maximum prevalence in the age range 18-29 months and declining
thereafter. This age range is consistent to what has been reported previously6. Prevalence was similar for boys and
girls in all age groups.
TABLE 5
OEDEMA PREVALENCE BY AGE AND BY SEX 1990-2015
Age
group
#
Males
(mo.)
6 - 17
#
Males
with
oedema
Oedema
prevalence
for males
(95% CI)
#
Females
#
Females
with
oedema
Oedema
prevalence
for females
(95% CI)
#
Children
#
Children
with
oedema
Oedema
prevalence
both sexes
(95% CI)
205,386
915
0.45 (0.42,0.48) 198,683
9830.49 (0.46,0.53) 404,069
1,8980.47 (0.45,0.49)
18 - 29
216,398
1,157
0.53 (0.50,0.57) 211,963
1,0980.52 (0.49,0.55) 428,361
2,2550.53 (0.51,0.55)
30 - 41
202,690
765
0.38 (0.35,0.41) 198,653
7470.38 (0.35,0.40) 401,343
1,5120.38 (0.36,0.40)
42 - 53
172,998
494
0.29 (0.26,0.31) 169,217
4110.24 (0.22,0.27) 342,215
9050.26 (0.25,0.28)
54 - 59
81,289
228
0.28 (0.25, 0.32) 78,770
1980.25 (0.22,0.29) 160,059
4260.27 (0.24,0.29)
878,761 3,5590.41 (0.39,0.42) 857,286
3,4370.40 (0.39,0.41)1,736,047
6,996 0.39 (0.39,0.41)
All
When SAM was defined by MUAC <115mm or oedema (see Table 7), the proportion of males with oedema among those
with SAM was consistently higher than that of females. Males with low MUAC presented with oedema more often than
females, and this held true for all age group breakdowns. It should be noted, however, that MUAC tends to identify SAM
among females more often and also that social factors may discriminate against females in some contexts.
Among SAM cases (identified by MUAC or oedema), the proportion of children presenting with oedema increased
progressively with increasing age, with the highest percentage found among children 54-59 months. Therefore, it is
noteworthy that while overall oedema prevalence was highest in the two youngest age groups, the proportion of SAM
cases (by MUAC or oedema) with oedema was highest in the two older age groups. This is because MUAC preferentially
selects younger children, so less SAM by MUAC in older children leads to higher oedema percentages.
PUTTING
CHILD KWASHIORKOR
ON THE MAP
23
TABLE 6
OEDEMA AS A PROPORTION OF SAM CASES DEFINED BY MUAC OR OEDEMA BY AGE AND BY SEX 1990-2015
Age
group
(mo.)
#
#
Males
Males
with SAM with
by
oedema
MUAC/oedema
Oedema as
a proportion
of SAM
by MUAC
for males
#
#
Females Females
with SAM with
by
oedema
MUAC/oedema
(95% CI)
6 - 17
Oedema as
a proportion
of SAM
by MUAC
for females
(95% CI)
#
Children
with
oedema
by
MUAC/oedema
Oedema as
a proportion
of SAM by
MUAC for
both sexes
(95% CI)
9,819
862
18 - 29
4,902
1094
22.3 (21.2,23.5)
5,993
104017.4 (16.4,18.3)
10,895
2,13419.6 (18.9,20.3)
30 - 41
1,964
717
36.5 (34.4,38.7)
2,200
72032.7 (30.8,34.7)
4,164
1,43734.5 (33.1,36.0)
1,046
476
45.5 (42.5,48.6)
1,029
39138.0 (35.0,41.0)
2,075
86741.8 (39.7,43.9)
427
215
50.4 (45.5, 55.2)
433
19244.3 (39.6,49.2)
860
40747.3 (43.9,50.7)
42 - 53
54 - 59
All
18,158
8.8 (8.2,9.4)12,625
#
Children
with
SAM
3364 18.5 (18.0, 19.1)
22,280
965 7.6 (7.2,8.1)22,444 1,827 8.1 (7.8,8.5)
3,308 14.9 (14.4,15.3)
40,438
6,672 16.5 (16.1,16.9)
When SAM was defined by WHZ<-3 or oedema (see Table 7), the proportion of females with oedema among those with
SAM was consistently higher than that of males. Females with low WHZ presented with oedema more often than males,
and this held true for all age group breakdowns, except the 54-59 month range. This finding is in contrast to that shown
in Table 6, where males were found to have oedema more frequently.
When examining oedema prevalence among SAM cases defined by WHZ (see Table 6), the age pattern disappears. The
proportions of SAM children (identified by WHZ or oedema) with oedema appear to increase with age, peak at 30-41
months, and then decrease again.
TABLE 7
OEDEMA AS A PROPORTION OF SAM CASES DEFINED BY WHZ AND OEDEMA BY AGE AND BY SEX 1990-2015
Age
group
(mo.)
#
#
Males
Males
with SAM with
by
oedema
WHZ/oedema
Oedema as
a proportion
of SAM by
WHZ for males
(95% CI)
#
#
Females Females
with SAM with
by
oedema
WHZ/oedema
Oedema as
a proportion
of SAM by
WHZ for
females
(95% CI)
6 - 17
868
#
Children
with
oedema
by
WHZ/oedema
Oedema as
a proportion
of SAM by
WHZ for
both sexes
(95% CI)
7.3 (6.8,7.8)
7,507
18-29
7,665
113214.8 (14.0,15.6)
5,046
1,07921.4 (20.3,22.6)
12,711
2,21117.4 (16.8,18.1)
30-41
4,811
75215.5 (14.5,16.6)
3,405
74121.7 (20.3,23.1)
8,216
1,49318.1 (17.2,18.9)
3,884
48712.4 (11.4,13.5)
2,978
40713.5 (12.3,14.8)
6,862
89412.9 (12.1,13.7)
42-53
11,891
#
Children
with
SAM
54-59 2,268 2279.9 (8.7,11.2) 1,975
All
30,519
346611.3 (10.9,11.7) 20,911
96513.0 (12.2,13.8)19,398 1,833 9.5 (9.1,9.9)
1979.9 (8.6,11.3) 4,243
3,38916.2 (15.7,16.7)
51,430
4249.9 (9.0,10.8)
6,85513.3 (13.0,13.6)
24
PUTTING
CHILD KWASHIORKOR
ON THE MAP
3.5 Distribution of MUAC and WHZ among oedema cases
Figure 3 shows that children with kwashiorkor tend to have a lower MUAC and a lower WHZ (median MUAC 125mm,
median WHZ -1.55) than children without kwashiorkor (median MUAC 142mm, median WHZ -0.62). It also shows that
the range of MUACs and WHZ among children with oedema is highly variable and that a large number are well above the
115mm MUAC or WHZ <-3 cut-off used to define SAM in the absence of oedema.
FIGURE 3
DISTRIBUTION OF MUAC AND WHZ IN CHILDREN WITH AND WITHOUT OEDEMA
MUAC with oedema
WHZ with oedema
1500
Number of children
Number of children
1500
1000
500
0
80
100
120
140
160
1000
500
0
180
-6
-4
MUAC (median =125 mm)
-2
0
2
4
6
4
6
MUAC (median = -1.55)
MUAC without oedema
WHZ without oedema
500,000
500,000
Number of children
Number of children
400,000
300,000
200,000
100,000
0
300,000
100,000
80
100
120
140
160
MUAC (median =142 mm)
180
0
-6
-4
-2
0
2
WHZ (median =0.62)
PUTTING
CHILD KWASHIORKOR
ON THE MAP
25
Table 8 demonstrates that there is huge variation in the median MUAC among children with oedema in different countries,
varying from less than 115mm in several Sahelian West African countries and Madagascar to more than 130mm in Central
and East Africa regions, suggesting that marasmic-kwashiorkor is much more common in some settings.
TABLE 8
COUNTRY BREAKDOWN OF MEDIAN MUAC RANGES AMONG CHILDREN WITH OEDEMA
Median MUAC Countries
among children
with oedema
Please note, italicised countries were not included
in Map 2 due to <20 total SAM cases (according
to MUAC and oedema) found in that country
<110
Mali
110 - 114
Gambia, Senegal, Madagascar
115 - 119
Myanmar, Niger, Burkina Faso, Nigeria, Mauritania, Central African Republic, Guinea-Bissau
120 -124
Congo (Kinshasa), Eritrea, Chad, Guinea, Haiti, Cote d’Ivoire, Angola, Tajikistan
125 - 129
Ethiopia, Indonesia, Uganda, Afghanistan, Nepal, India, Mozambique, South Sudan
130 - 134
Benin, Burundi, Cameroon, Liberia, Rwanda, Djibouti, Kenya, Pakistan,
Sierra Leone, Malawi, Somalia
135 - 139
Sudan, Tanzania, Nicaragua
>139
Guatemala, Zimbabwe, Togo, Macedonia, Bangladesh, Zambia, Yemen, Bolivia, Sri Lanka
Unknown
Albania, Botswana, Jordan, Philippines, Thailand
FIGURE 4
ROC CURVES COMPARING THE ASSOCIATION OF
MUAC AND WHZ WITH OEDEMA
100
155
MUAC
160
2
3
1
WHZ
150
145
80
0
140
Sensitivity (%)
135
-1
60
130
Figure 4 shows the association between
MUAC and WHZ with kwashiorkor by
means of ROC curves (receiver operating
characteristic). Sensitivity in this case
measures the proportion of oedema cases
that fall below the cut-off shown on the
curve, while specificity measures the
proportion of children without oedema
that are above the same cut-off. The
ROC curve of MUAC is above that of
WHZ, indicating that MUAC is better at
identifying cases of oedema than WHZ.
125
40
120
-2
115
20
-3
110
105
100
0
0
20
40
60
100 - Specificity (%)
80
100
Children with
kwashiorkor have a
lower median MUAC
and a lower median
WHZ than those
without kwashiorkor,
but MUAC and WHZ
vary significantly
for children with the
condition.
26
PUTTING
CHILD KWASHIORKOR
ON THE MAP
4 ADMISSION AND OUTCOME DATA
Currently there is no existing global database that includes admissions for oedema in treatment programmes. In 2010,
UNICEF tried to rectify this by commissioning a SAM mapping29, followed by a pilot Annual SAM Update that has been
developed into a “Nutridash”30 system. However, this is not broken down by type of admission criteria. Some countries
and some agencies collect separate statistics for kwashiorkor and marasmus, but outcomes for SAM tend to use a
common denominator for all children discharged (as per SPHERE standards), regardless of admission criteria, so it is
not possible to ascertain from routine data where the highest mortality lies. This is partly due to the simplification of
reporting requirements over recent years that has led to a combining of kwashiorkor and marasmus admissions.
This section attempts to collate information from various sub-analyses of SAM across countries and specific agency
routine programme data or case studies to determine the percentage of admissions that can be attributed to kwashiorkor.
TABLE 9
ADMISSION DATA PROVIDED BY AGENCY
Agency
# countries
Years
ACF
8
2011 - 2014
ALIMA
5
2006 - 2015
College of Medicine, Malawi
1
2014 - 2015
KEMRI, Kenya
1
2000 - 2014
Mulago Hospital,Uganda
1
2009 - 2014
MRC, Gambia
1
2010 - 2015
MSF
132014
Save the Children
12
2010 - 2015
Terre des Hommes
1
2011 - 2015
19
2008 - 2015
1
2010 - 2014
World Vision
17
2006 - 2015
TOTAL No OF COUNTRIES
26
UNHCR
UNICEF/Government, Malawi
4.1 Proportion of kwashiorkor among SAM cases admitted
to therapeutic programmes
Table 10 is a summary of all the admission data received from different sources. It shows a high variability of kwashiorkor
admissions across different countries, years and admission modality. Although this dataset is incomplete, it shows small
consistent trends across agencies and countries. In almost all countries, inpatient admissions from kwashiorkor were higher
than outpatient and these often comprised up to 50% of total inpatient admissions. Malawi and several Central African
countries demonstrated some of the highest admission rates from kwashiorkor, especially Central African Republic and
DRC. Ethiopia, Haiti, Kenya, Uganda (inpatient data available only) and Zambia also showed high proportions of admissions
with kwashiorkor. Bangladesh, the Gambia and Somalia showed particularly low levels of oedematous admissions.
Similar results were retrieved from refugee camp data, provided by UNHCR. These were collated separately and not
combined with the main table to avoid skewing country results and can be found in Annex 7. In addition, camps in Chad,
Rwanda and Uganda showed high levels of kwashiorkor as a percentage of admissions for therapeutic care, while Nepal,
Sudan and South Sudan had low levels.
2006
2007
2008
2009
2010
2011
2012
2013
2014
I49.8% (433/869)41.7% (341/818) O38.6% (632/1,635) 33.7% (973/2,883)2.6% (257/9,733)15.6% (652/4,169) 12.5% (149/1,189)
CAR
29.4% (45/153) O
44.4% (176/396)
O47.3% (370/782)38.5% (1,121/2,915)29.8% (1,507/5,058)21.0% (90/428)21.4% (329/1,539)24.6% (6,314/25,668)
I100% (52/52)
O6.1% (7/115)13.3% (111/837)18.2% (814/4,470)6.6% (157/2,362)7.9% (605/7,643)21.7% (2,115/9,760)29.1% (2,689/9,248)22.8% (876/3,835)
Ethiopia
66.1% 69.3% O57.0%
57.4% 61.8% 49.0%
44.9%
I = Inpatient O = Outpatient
(14,006/24,591)(14,040/24,471)(12,554/20,322)(10,835/22,099) (11,868/26,417)
(4,253/7,651)
60.1% 55.6%
0.1% (3/3,894) (8,200/12,713) (7,150/10,812) (6,647/9,592) (5,405/8,987)
64.5%
1.0% (27/4,769) 0.0% (0/1,390)
I
1.0% (70/6,914) Malawi
1.3% (13/976) I40.1% (142/354)42.6% (180/423)29.0% (85/297)35.0% (81/231)59.0% (23/39)70.0% (14/20)
O 0.5% (1/202)0.0% (0/180)1.3% (1/78)
Kenya
I2.2% (1/46)1.8% (3/165)
India
GambiaI 1.7% (1/60)2.9% (3/104)0% (0/196)0% (0/160)0% (0/167)0% (0/70)
0% (0/934)
53.4% (206/386)52.3% (79/151)3.4% (80/2,384)
62.5% (10/16)
25.7% (19/74)35.0% (117/334)38.8% (187/482)50.0% (33/66)34.7% (1,841/5,310) I
Congo (DR)
100% (318/318)55.7% (162/291)58.5% (72/123)73.9% (329/445)63.3% (348/550)
O6.7% (86/1,280)13.3% (294/2,215)2.8% (435/15,594)2.4% (822/34,076) 3.2% (355/11,166)
Chad
I
25.0% (21/84)11.6% (239/2,060)37.1% (13/35)
30.0% (79/263) 37.0% (1,135/3,071) 27.2% (321/1,181)20.7% (279/1,351)
O2.0% (88/4,340)1.3% (95/7,424)
Burundi
28.0% (177/633)21.3% (221/1,040)
0% (0/104)
0.9% (2/232)
2015
Burkina Faso I
0% (0/221) 0% (0/23) O 0% (0/2)0% (0/116)0% (0/49)0.3% (1/350)
1.0% (9/925)
BangladeshI
Angola
O 8.0% (2,051/25,645)
O 0.9% (15/1,613) Afghanistan
I
4.1% (143/3,530)
Country
OVERALL SUMMARY OF KWASHIORKOR PERCENTAGES (TOTAL NUMBER OF KWASHIORKOR ADMISSIONS/TOTAL
NUMBER OF SAM ADMISSIONS) BY YEAR AND BY COUNTRY (2006-2015)
TABLE 10
PUTTING
CHILD KWASHIORKOR
ON THE MAP
27
TABLE 10 CONTINUED OVERLEAF
2006
2007
2008
2009
2010
2011
2012
2013
2014
O4.5% (63/1,414)3.1% (252/8,021)4.8% (885/18,353)5.0% (920/18,505)3.9% (212/5,462)
2015
O 0.1% (2/3,013)
O0.6% (34/5,357)0.2% (36/18,964) 2.1% (392/18,478)0.0% (0/30,609)0.0% (0/14162)
I69.0% (238/345)84.4% (961/1,139)17.8% (241/1,353)
O0% (0/1,327)0% (3/9,114)0.2% (17/7,862)0% (0/195)0.5% (57/10,520)
I1.6% (37/2,300)7.8% (109/1,384)1.2% (6/489)
O1.4% (51/3,663)0.2% (39/18,334)0.9% (108/12,359)
Pakistan
Somalia
3.9% (132/3,342)1.5% (26/1,680)
O0% (0/15)0% (0/82)1.6% (56/3,509)0.4% (14/3,813)0.8% (30/3,615)0.2% (33/13,767)
0.1% (1/1,585)
0% (0/7)27.3% (3/11)1.1% (5/470)8.3% (41/493)16.3% (80/492)14.7% (327/2,227)1.3% (1/77)
O1.3% (3/229)2.0% (21/1,060) 2.5% (259/10,544)
3.5% (421/11,925)1.6% (38/2,389)
I65.5% (466/711)60.4% (598/990)51.2% (608/1,188)44.0% (547/1,244)50.0% (532/1,063)
54.4% (657/1,208)
O31.3% (15/48)21.2% (84/397)27.3% (39/143)
I = Inpatient O = Outpatient
ZimbabweO 10.6% (5/47)5.8% (12/206)3.9% (4/103)
17.5% (63/360)
I47.0% (71/151)41.2% (40/97)
Zambia
Yemen
O
0.2% (12/5,848)0.2% (21/9,545)0.0% (0/444)
Uganda
O
0% (0/17)
Syria
I
0% (0/19)
Sudan
I
21.0% (22/105)11.2% (18/161)6.5% (5/77)
South SudanI
I20.5% (18/88)12.1% (47/387)
Nigeria
14.7% (78/529)15.6% (80/512)12.8% (37/289)
O0.1% (12/11,479) 2.0% (1,020/51,236)1.3% (1,151/87,493)2.1% (1,327/64,284)2.1% (378/17,673)
19.5% (585/3,001)
Niger
I
27.7% (304/1,096)
Nepal
2.1%(2/94)3.1% (1/32)
12.8% (25/195)0% (0/560)0.0% (0/572)
1.8% (50/2,764)1.0% (17/1,688)0.8% (11/1,432)1.9% (2/105)
O0.9% (5/47)
5.2% (1/19)7.2% (14/195)
O0.4% (1/267)0.7% (4/562)0.3% (1/381)
MyanmarI
Mauritania
I
0% (0/15)0.0% (0/20)
Mali
I
46.3% (1,668/3,600)
Country
28
PUTTING
CHILD KWASHIORKOR
ON THE MAP
TABLE 10 CONTINUED
PUTTING
CHILD KWASHIORKOR
ON THE MAP
29
4.2 Mortality
The question of whether there is greater mortality amongst children with oedema than amongst those with marasmus
is still unclear and there is a lack of data stratifying by grade of oedema (+, ++, +++) and by inpatient versus outpatient
admissions. While the following data are from small programmes, they have been included to demonstrate some of the
variety found at different sites and to encourage further monitoring of oedema levels in the future. Mortality data from
prevalence surveys was not available, so all the data in this section comes from admission data.
Figure 5 shows the case fatality rate (CFR) for children admitted in Kilifi County Hospital, Kenya, since 1999. The graph
clearly shows that the highest mortality exists among children who have oedema, and while the case fatality rate has
fallen over the years, it has remained high.
FIGURE 5
EMRI, KILIFI HOSPITAL, KENYA, INPATIENT MORTALITY DATA
40%
Kilifi Hospital, Kenya, Case Fatality Rate by SAM type
Kilifi Hospital, Kenya, Case Fatality Rate by SAM type
40%
35%
35%
30%
30%
25%
25%
20%
20%
15%
15%
10%
10%
5%
5%
0%
0%
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
MUAC<115mm only
MUAC <115mm and Oedema
Linear (MUAC<115mm only)
Linear (Oedema only)
MUAC<115mm only
MUAC <115mm and Oedema
Linear (MUAC<115mm only)
Linear (Oedema only)
Oedema only
Linear (Oedema only)
Oedema only
Linear (Oedema only)
Children with kwashiorkor tend to be
admitted into inpatient programmes rather
than outpatient, and they often comprise
up to 50% of total inpatient admissions.
30
PUTTING
CHILD KWASHIORKOR
ON THE MAP
Table 11 describes admission and mortality data from combined inpatient and outpatient care from various
nutritional programmes managed by ALIMA. This data shows a consistently higher mortality among children with
oedema in all countries.
TABLE 11
ALIMA PROGRAMME MORTALITY, JANUARY 2011 - APRIL 2015 (INPATIENT AND OUTPATIENT DATA)
Country Site
Total SAM % total SAM
admissions admitted
with
kwashiorkor
Mortality rate
for all children
in the
programme
Mortality rate
for children
with
kwashiorkor
Chad
Tchad Ngouri
12,137
4.8%
1.7%
4.9%
Tchad N’Djamena
27,139
3.0%
1.7%
11.1%
4.7%
2.3%
16.0%
MaliKangaba
6,433
Kolokani
9,5412.7%
1.2% 21.6%
Fana
3,4376.6%
1.1%
5.3%
Dioila
6,6918.5%
1.5%
7.4%
Koulikoro
3,6253.3%
1.7% 10.0%
Ouéléssébougou7,654
9.2%
1.6%
3.9%
Diré
5,7971.9%
1.8%
8.4%
Goundam
1,8162.6%
1.0%
6.4%
Niger
Niger Dakoro
61,838
1.8%
1.6%
5.8%
Niger Mirriah
83,250
2.9%
1.4%
20.9%
Further details were collected from ALIMA N’Djamena project in Chad. A total of 1,600 children were admitted to
inpatient stabilisation care and 11,900 to outpatient or ambulatory care, with kwashiorkor cases making up 19% and
1% of total admissions, respectively. The admissions to inpatient care were analysed by anthropometrics/age and
outcome, with the kwashiorkor admissions broken down into two groups: those with “straightforward” kwashiorkor
(defined as bilateral pitting oedema but no wasting) and those with marasmic kwashiorkor (defined as oedema and
MUAC <115mm). In N’Djamena, the uncomplicated kwashiorkor child had a better prognosis than the others, while
the marasmic kwashiorkor child fared the worst (see Table 12). The programme does not yet have clear results of the
prevalence of HIV in these children, but it is thought to be significant in the area.
PUTTING
CHILD KWASHIORKOR
ON THE MAP
31
TABLE 12
ALIMA ADMISSION DATA FROM N’DJAMENA, CHAD, 2014
At admission
%
#
Oedema and MUAC >115mm
Average age
Average MUAC
(mo.)
(mm)
% Died
80
8%
18.6
120.2
9%
(marasmic kwashiorkor)
128
13%
20.2
102.4
18%
MUAC <115mm (no oedema)
472
46%
15.2
101.9
13%
WHZ -3 and length < 65cm
173
17%
9.5
N/A
12%
WHZ -3 and length > 65cm
142
14%
14.9
117.9
4%
27
3%
4
N/A
18%
Oedema and MUAC < 115mm
< 6 months
TOTAL
1,022100%
Note all children in the admission table were greater than 6 months of age apart from the last category which included only
children under 6 months whether admitted with marasmus, kwashiorkor or marasmic kwashiorkor.
Published data from a community-based randomised trial in Malawi also gives information on the outcome of SAM
cases detected in the community and treated as outpatients 26. Of 2,767 admissions to outpatient therapeutic care with
uncomplicated SAM, 1,945 (70.0%) had kwashiorkor, 244 (8.8%) had marasmic-kwashiorkor (defined as oedema plus WHZ
<-3) and 578 (20.9%) had marasmus (WHZ<-3). No relation between the type of SAM and intervention group (antibiotics
vs no antibiotics) was observed for either the rate of nutritional recovery or the mortality rate. However, children with
kwashiorkor recovered at a higher rate than those with Simple marasmus. Marasmic-kwashiorkor cases had the lowest
recovery rate of the three groups of children; they recovered less often, significantly more slowly and had higher mortality
rates. In summary, as shown in Table 11, children with oedema overall have a lower mortality than SAM children without
oedema, but children with oedema and low WHZ have the highest mortality.
TABLE 13
MORTALITY BY CATEGORY, COMMUNITY-BASED
PROGRAMME, MALAWI, 2009-2011, TREHAN ET AL
#
Oedema with WHZ > -3
Oedema with WHZ < -3
(marasmic kwashiorkor)
Oedema (total)
WHZ < -3 without oedema Deaths %
1,945
65
3.3
244
42
17.2
2,189
107
4.9
578 43 7.4
32
PUTTING
CHILD KWASHIORKOR
ON THE MAP
In contrast to the above, a small inpatient dataset in the same area (Moyo Inpatient ward, Blantyre, Malawi, where
marasmus is defined as MUAC <115mm or WHZ <-3), monthly data for 2014 and part of 2015 showed significant variation
in mortality for children with oedema throughout the months of 2014, ranging from 5.6% for August to 42.9% for October.
The mortality for children with kwashiorkor in 2014 was only slightly higher than for those without (20.8% vs. 19.5%).
The 2015 data, however, showed a lower mortality amongst children with oedema, as compared to those without (14.1%
vs 21.4%).
Data received from MSF showed that in Tombuctu, Mali, the mortality rate for children with kwashiorkor, as compared
to non-kwashiorkor, was 3 times higher, reaching 25%; in Yida, South Sudan, the mortality rate for children with
kwashiorkor, as compared to non-kwashiorkor, was 4.5 times higher, reaching 22%; and interestingly, in both Paou and
Carnot (CAR), the mortality for non-kwashiorkor children, as compared to kwashiorkor, was 1.5 times and 5 times higher,
respectively, with all rates <10%. Summarising the MSF programmes for which data was received, mortality rates for
kwashiorkor children ranged from 1% to 25%, depending on the country and region.
4.3 Seasonality
There have been anecdotal reports that kwashiorkor is more prevalent during the rainy seasons. A seasonal pattern
cannot be shown easily by cross-sectional surveys (unless analysing repeated surveys in the same area during different
seasons)”, so this was broadly examined in the few admissions datasets where consistent admission data were available
on a monthly basis over several years. This data was collected and charted in the graphs below.
Five years worth of national Malawian CMAM data, provided by the Government of Malawi and UNICEF (see Figure 6
and Annex 8 for raw data), and data from an inpatient unit in Uganda (see Figure 7), showed no obvious seasonal pattern.
It would be useful to chart by season and explore different contexts, as the Ugandan data is from a more urban context.
FIGURE 6
PERCENTAGE OF KWASHIORKOR AMONG SAM CASES. MALAWI NATIONAL CMAM DATA: 2010-2014
% Kwashiorkor in Malawi from 2010-2014
75%
70%
65%
60%
55%
50%
45%
40%
35%
30%
Jan
2010
Feb
2011
Mar
2012
2013
Apr
2014
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
PUTTING
CHILD KWASHIORKOR
ON THE MAP
33
FIGURE 7
UGANDA MWANAMUGIMU NUTRITION UNIT, MULAGO HOSPITAL, INPATIENT ADMISSIONS: 2009-2014
% Kwashiorkor in Mulago Hospital from April 2009 to December 2014
80%
70%
60%
50%
40%
30%
Jan
2009
Feb
2010
Mar
2011
2012
Apr
2013
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
2014
MSF admission data (Mali, DRC, South Sudan, Niger and CAR) also suggest that the prevalence of kwashiorkor does not
appear to be seasonal in nature.
The lack of seasonal pattern cannot be substantiated with only these examples. Additional analyses will be needed in
areas with strong seasonal variations. For example, in the Sahel and West Africa where they have very dry and very wet
seasons, with subsequent large variation in the burden of malaria, diarrhoea and respiratory infections that may affect
oedema prevalence.
50% of SAM cases diagnosed in Malawi had kwashiorkor,
32% in Democratic of Congo and 1.6% in Pakistan.
34
PUTTING
CHILD KWASHIORKOR
ON THE MAP
5 COMPARISON OF SURVEY
AND ADMISSION DATA
Figure 8 compares the proportion of kwashiorkor as per admissions in different centres and the burden of kwashiorkor
as per prevalence of total SAM (data from surveys), matched by country and year. No clear pattern can be identified,
suggesting a lack of relationship between admission data and kwashiorkor prevalence obtained from surveys.
When comparing the data on admissions in refugee camps provided by UNHCR from 2008-2014 with the prevalence data
from camp surveys, similar results as in Figure 8 were obtained (data shown in Annex 7).
FIGURE 8
% of SAM children with oedema (data from admission records)
% of SAM children with oedema (data from admission records)
SURVEY-ADMISSION SCATTER PLOT FOR GLOBAL ADMISSION
DATA (NOT INCLUDING REFUGEE CAMPS)
100
100
80
80
60
60
40
40
20
20
0
0 0
0
20
40
60
80
% 20
of SAM children with
from prevalence surveys)
40 oedema (data 60
80
100
100
% of SAM children with oedema (data from prevalence surveys)
Data from ACF in Artibonite, Haiti, also showed a discrepancy with respect to survey and admission data taking into account
that surveys occurred before admissions started. All nutritional surveys carried out in the zone by ACF or other agencies
showed zero or close to zero levels of bilateral oedema (surveys in 2008 and 2009 showed global acute malnutrition levels of
4.3% (CI: 2.5-6.1%) and 4.4% (CI: 3.0-5.8%) and oedema 0.1%, 0.3% and 0.1% respectively (NCHS)); a post-earthquake survey
in Artibonite conducted in January 2010 found 2 cases of oedema, or 0.3% (WHO), and a national SMART in 2009 did not use
oedema as an indicator. However, when activities were extended to the rural areas, the team noted that the average caseload
of kwashiorkor rose to around 40% of total admissions. A breakdown of these cases was conducted and showed that 36%
of the children admitted to outpatient centres and 50% admitted to inpatient units had kwashiorkor.
PUTTING
CHILD KWASHIORKOR
ON THE MAP
35
TABLE 14
BROAD ANALYSIS OF ADMISSIONS TO HEALTH FACILITIES IN HAUT ARTIBONITE, HAITI (2009 TO 2013),
DEMONSTRATING PERCENTAGE OF ADMISSIONS WITH OEDEMA
2009
Total SAM
Total with oedema
% with oedema
2010
2011
2012
2013
18357014371773186
47
191
508
694
84
25.7%
33.5%
35.4%
39.1%
45.2%
A spatial pattern was noted according to the geographical area of residence or origin of children with kwashiorkor, but
information was not available as to whether this is due to more extensive screening of bilateral oedema by community
health workers or better awareness of the condition by the community in these areas.
There does not appear to be a relationship between
kwashiorkor prevalence based on surveys and that
determined by nutritional programme admission
data. However, the data from nutritional surveys and
that from programmes, when collated on a larger
scale, both suggest a number of the same countries
as being high burden for oedema.
36
PUTTING
CHILD KWASHIORKOR
ON THE MAP
6 DISCUSSION
6.1 Limitations, data issues and biases
Analyses presented in this report are based on “found data” derived from “available” information utilised in programming.
Thus, the current dataset is not ideal, since it has not been directly collected by the authors for research purposes. Data was
predominantly from areas experiencing a nutritional emergency or where funding was granted to conduct nutritional surveys
or provide nutritional services. The maps and graphs produced in this report give an indication of the current situation for
oedematous malnutrition at national level, but do not give a definitive picture. Individual country results should not be
generalised to all regions, but rather should stimulate further research into the condition at a sub-regional level.
The data used in this report was collected for purposes other than mapping oedema and suffers from a number of
selection biases, mainly:
◆
PLACE: A mapping sample should, ideally, be based on a reasonably even and exhaustive spatial sample.
The sample that constitutes the database is spatially clustered. The data gathering process concentrated
on collecting survey datasets from UN organisations and NGOs working in humanitarian contexts. Many
of the surveys and clinical datasets come from locations in which there is a suspected or confirmed high
prevalence of malnutrition in which UN and NGOs are present and active in the nutrition field.
◆
TIME: The database is limited with regard to analysing data as a time series and it is not possible to derive
trends due to the limited number of data points in certain locations, whilst in other locations there are
numerous data points, albeit temporally clustered. This temporal clustering is due to a similar selection bias
as with the spatial sample. A second temporal bias is that all data comes from cross-sectional prevalence
surveys, which may be problematic for conditions, such a oedema, believed to have a rapid onset & short duration.
Results should then be interpreted with care. Readers of this report are urged to focus on rankings rather that absolute
values. Other limitations can be found in table 15.
TABLE 15
LIMITATIONS OF SURVEY AND ADMISSION DATA
SURVEY DATA
Survey selection
Prevalence vs incidence
Lack of standardisation
Timing of surveys
As cross-sectional surveys are mainly small scale and conducted when and where a nutritional problem
is highlighted, or a nutritional intervention is planned, rather than always on a national scale, bias is
presented. Furthermore, SMART survey sampling use population proportional sampling where small
rural areas have a smaller representation given smaller populations in the dataset. Governments have
shared national surveys when they exist, thus improving the dataset.
Nutritional surveys measure prevalence, and oedema is better assessed using incidence rather than
prevalence due to the short duration of kwashiorkor (marasmus also might be affected too).
Surveys were received in 5 different formats (ENA software for SMART, Epi-Info, STATA, SPSS, and
Excel), which required file conversions to the comma-separated values format. Some datasets were
provided as raw data, while others had already been cleaned before receipt. Therefore, many surveys
were cleaned based on the contributing organisation’s standards, while others were cleaned based on
the project’s standards, resulting in some variability. Agencies may have used WHO and/or SMART
flagging criteria, censored or deleted records, or left flagged data alone but excluded them from the
analysis.
Oedema could be seasonal, and surveys are often conducted prior to and at the end of interventions,
depending on funding and not necessarily based on seasonal variation.

PUTTING
CHILD KWASHIORKOR
ON THE MAP
Measurement error
Poor training and lack of measurement skills can lead to under or overestimation of oedema, although
many surveys build in a “check” for oedematous cases by survey supervisors..
Missing variables &
data errors
The missing variables were most often MUAC and sometimes oedema. MUAC was not consistently
taken on younger but eligible children in some surveys, which may bias the age distribution of the
results. Data entry errors were common in the original datasets. The MUAC variable was most often
recorded incorrectly, so some errors may have been made in assuming the intended values. The
duplicate code could not account for cleaning differences amongst data entry persons, so this may have
prevented some duplicate surveys from being detected.
Loss of original raw
datasets
Permissions &
sourcing of data
Agencies found it difficult to find all the original raw datasets, especially from older surveys, as these
had not always been systematically stored at headquarter or country level.
Subnational
classifications
Variable use of
definitions & survey
metadata collected
Different age groups
Some countries did not provide permission for use of nutritional surveys outside of the country of origin.
This limits the extent that which the map can represent the actual nutritional situation worldwide. Obtaining
data permission can be a lengthy process and only limited access may be granted to some datasets.
Metadata was often not present or was coded in an opaque manner. For example, sometimes a livelihood
zone was used as a region (within a country), when in fact it spanned many regions, so results for
specific regions may not have been possible. This complicated interpretation as each livelihood zone
may span many regions and each region may have many relevant livelihood zones. Although current
classifications were used for the final analysis, some regional classifications may have been different at
the time of survey, or at the time of the survey classifications may not have been clearly defined.
Definitions of livelihood zones, pastoralists, urban, peri-urban and rural varied, as did inclusion in the
data collected. It would be useful to have a minimal set of metadata collected across surveys in the
future.
Few surveys collect data on children aged <6 months or children older than 5 years, so estimates in
these age groups are not possible.
ADMISSION DATA
Reliance on programme
data & small numbers
Lack of segregation
due to data collection
simplification
Survey selection
Outcome data based on
total discharges
Definition of oedema
grade & marasmickwashiorkor
Data is often collected haphazardly during emergencies, so there is great risk of recording errors and
data gaps. No global database of admissions exists, so numbers are small. National data was only
available for one country.
Data collection tends to follow national guidelines, which are often based on SPHERE standards or
UNICEF/WHO recommendations. There is a move to simplify the amount of data collected, with the
result that oedema, MUAC and WHZ admissions grouped together.
Assessments of mortality or defaulting by admission type are not available apart from more medicallyorientated NGOs.
A lack of standard definition for the grade of oedema (+, ++, +++) and absence of grade reporting were
frequently found. Similarly either marasmic-kwashiorkor children were not reported on or definitions
varied.
This was not routinely available data, so it was difficult to determine whether numbers were only higher
in programmes with strong training and routine case finding.
37
38
PUTTING
CHILD KWASHIORKOR
ON THE MAP
6.2 The wider context of kwashiorkor
The main goal of this report is to highlight the importance of kwashiorkor as a public health problem, as reflected by
its prevalence and also by the proportion of SAM cases it represents in surveys. Despite its limitations, this report
gives, for the first time, a representation of the geographic distribution of kwashiorkor and the main findings are
consistent with what has been reported for more than 40 years in West Africa, with a higher frequency of kwashiorkor
in the Guinean compared to the Sudanic Zone31. The explanation given at that time was that the proportion of energy
derived from protein seemed lower near the equator, where diet had a higher proportion of roots and tubers. The
hypothesis that kwashiorkor is due to protein deficiency has been challenged, as it fails to explain many of its clinical
features32 Also, recent studies failed to show an association between protein intake and risk of kwashiorkor33,34. The
cause of kwashiorkor is still unknown, despite all the research conducted to date, and various hypotheses are still
being tested in an attempt to better explain onset of the condition.. Some of these explanations are: lower sulphur
amino acid consumption, different types of soil, Enteric Environmental Disorder, gut microbiota, high HIV prevalence,
climatological contexts, etc.. 21, 35, 36, 37, 38, 39, 40, 41.
MUAC is apparently less sensitive to changes in fluid retention42 and seems better for assessing the general nutritional
status of children with oedema that does not extend up to the child’s upper arms (i.e., +++ oedema). This latter
assumption is supported by the ROC curves in this report that describe the association between anthropometry and
oedema, showing that MUAC more readily identifies children with oedema, compared to WHZ. A similar observation
was made earlier in Malawi43.
Mortality associated with kwashiorkor also varies across studies. Data from Malawi suggests that the prognosis of
oedematous malnutrition may be greatly influenced by an associated WHZ<-326. This suggests that children with
oedema and WHZ<-3 are in a very high risk category. With the known close relationship between MUAC and mortality44,
the clinical outcome of oedematous malnutrition in countries is likely to be highly influenced by this underlying risk
factor. Only one study was found to have examined the association of oedema with mortality adjusted for MUAC in a
multivariate analysis. This hospital database showed an odds ratio risk of mortality for children with oedema of 2.83
(95% CI= 1.29-5.26), but 4.76 (95% CI: 2.76-8.21) for those with both oedema and MUAC <115mm23. This suggests
that children with oedema and MUAC <115mm are in a very high risk category. All this should be taken into account
when comparing mortality of oedematous children between countries. Certainly, more studies are needed in order to
accurately determine mortality rates in each type of SAM.
MUAC appears to be less sensitive to changes in hydration
status and appears a better means to assess the general
nutrition status of children with oedema that doesn’t extend
up to the child’s arms.
PUTTING
CHILD KWASHIORKOR
ON THE MAP
39
7 RECOMMENDATIONS
7.1 Programmatic recommendations
Improve the current data collection system for kwashiorkor
First and foremost this report highlights the need for information on kwashiorkor to be collected in a more intensive,
systematic and standardised manner.
Despite the limitations of the cross-sectional survey, given it is the main tool used to detect acute malnutrition, the
assessment of bilateral pitting oedema (and MUAC) should be included in the list of key variables (age, sex, weight,
height) collected, especially in standard national surveys (i.e. SMART, MICS, Demographic Health Surveys (DHS)) and
those surveys conducted in areas where kwashiorkor is potentially prevalent. Ideally, context information should also
be collected in order to explain high oedema rates (e.g. seasonality, emergency-related indicators, IDP setting, etc).
Hopefully this will not only provide more data but also increase awareness among surveyors, communities and health
practitioners of how to screen for, detect and refer cases of oedema.
Encourage moving to active community-based screening of kwashiorkor to assess its burden
A correct assessment of the importance of kwashiorkor can only be obtained by active regular screening at the community
level of all at risk under five years old children in populations where it is known to be present. This cannot be achieved
by cross-sectional surveys alone, but the data collection process could be integrated into treatment programmes where
children are regularly screened for referral to treatment or national surveillance systems.
More training of health workers and caretakers on the recognition of oedema is needed, as well as creating awareness of
kwashiorkor at the community level to better understand how to assess and how to treat this condition. Special attention
could be given to the possible involvement of mothers in oedema detection. A pilot study has shown that mothers can be
trained to measure MUAC45. It would be worth investigating the viability of training mothers to assess for oedema. As this
condition seems transient, mothers are likely in the best position to detect it before it becomes too severe.
Review admission categories and encourage monitoring of marasmic-kwashiorkor
Further attention should be given to classification of admission categories and possible pooling of data across centres.
Ideally, admission and outcome data should be broken down by type of SAM (marasmus or kwashiorkor) to help monitor
where the highest caseloads are found. Standard definition or standardisation of the grading (+, ++, +++) would be useful.
To better understand the diversity of background nutritional status, a MUAC-based definition of marasmic-kwashiorkor
may be more relevant. The authors recommend using the presence of bilateral pitting oedema and MUAC cut offs as a
definition, given that WHZ is particularly influenced by fluid accumulation caused by oedema, thus the latter would be
more difficult to interpret. Since it is suspected that children with marasmic-kwashiorkor are at greater risk of dying, there
should be further study into mortality in this group. It would also be beneficial to increase awareness and monitoring of
marasmic-kwashiorkor children in surveys and nutrition programmes. As mortality is higher for these children, piloting
the inclusion of a marasmic-kwashiorkor category at selected centres would help determine its feasibility, a necessary
step before it could be suggested as a new admission category.
Improve the current survey reporting system
Furthermore, to benefit measurement of acute malnutrition in general, there is a need for systematic collection, storage,
and standardisation of nutritional survey data, software and definitions within narrative reports. The inadequacy of the
current state of surveys (raw data and narrative reports) is a huge obstacle to having a global view of SAM, and more
specifically, the problem of kwashiorkor.
40
PUTTING
CHILD KWASHIORKOR
ON THE MAP
During this project many agencies could only provide narrative reports and had lost the original raw data, which limits
the interpretation of the surveys and led to a number of surveys being excluded from the analysis. Narrative reports
varied hugely in definition, presentation and inclusion of metadata, and therefore could not be used in this exercise.
Inconsistencies were found across surveys, including lack of a standard format, varying codes for some indicators, loss of
original files (often with past employees who left or through corrupted files), no clear contact person, etc. Variation was
found in the type of software used, coding/labelling and units (gender, age, grade or absence/presence of oedema, use of
millimetres or centimetres etc). Many surveys did not collect oedema or MUAC as one of the key indicators, even though
these are admission criteria for services managing acute malnutrition. No DHS surveys were included, as these did not
have all the indicators, especially MUAC or oedema, and only one MICS survey had all the needed variables. The rest of
the MICS surveys provided were missing MUAC.
There is neither a standard data entry system nor a standard code system used among organisations. The creation and
adherence of such a system would facilitate reviews such as this one, so it is worth advocating for the preservation of
raw data given its potential value for future research.
Grade of oedema is usually not included in surveys, even though it is important for admissions, as many countries dictate
whether a child should be managed in inpatient or outpatient care. A lack of definition or standardisation of the grading
(+, ++, +++) was also found.
7.2 Research priorities
It would be useful to increase research efforts around the following areas:
Examine the mortality risk of children with oedema, marasmus and marasmic-kwashiorkor
More studies examining different admission categories (marasmus and kwashiorkor) and stratified by grade of kwashiorkor
(+, ++, +++) and treatment category (inpatient or outpatient) would be beneficial for identifying more prognostic factors
and determining specific treatment recommendations. In particular, a research priority should be to look in more detail at
the outcomes of children with kwashiorkor and low MUAC (marasmic-kwashiorkor) as defined by oedema and MUAC less
than 115mm.
Examine the geographical distribution of potential causal factors
More epidemiological studies and careful examination of dietary patterns in different parts of Africa and their association
with possible variation of intake of some key nutrients could help explain the geographical distribution of kwashiorkor,
that it is more frequent in central and humid parts of Africa, as well as identify potential causal factors. The pattern
of sulphur amino acid intake deserves special attention, as available data suggest it may play a role in the origin of
kwashiorkor36, 46. It appears that the last attempt to examine variations of sulphur amino acid intake in Africa goes back
to more than 40 years ago39. No attempt has ever been made to examine it on a global scale. This early report suggests a
lower sulphur amino acid intake in humid parts of Africa, a pattern which is consistent with the geographical distribution
of kwashiorkor presented in this report. This early study on sulphur amino acid availability was based on agriculture and
food consumption data published in the previous 25 years. This information needs updating.
The role of a possible limiting nutrient should also be examined with modern techniques of diet modelling, using linear
programming, which are powerful tools to investigate limiting nutrients based on food available in the community and semiquantitative data on food portions 47, 48. The geographical distribution of dietary patterns and of the possible association
PUTTING
CHILD KWASHIORKOR
ON THE MAP
41
with kwashiorkor could also be studied at the county level, at the regional level or even in smaller geographical units. In
Malawi, for example, kwashiorkor prevalence seems to vary within regions with no clear explanation49. Comparing maps
of diet quality or dietary diversity and distribution of kwashiorkor could show possible linkages.
The limits of the examination of the geographical distribution of kwashiorkor to unravel its causes should be
acknowledged. This approach provides a very low level of evidence to assess causality, as this association can be
produced by unknown confounding factors or by untested causal factors. For a disease such as kwashiorkor, which is
most likely multifactorial, it has the advantage, however, of avoiding the problem of universal exposure to a risk factor. For
instance, if kwashiorkor is caused by the conjunction of two different risk factors, a case control study will fail to identify
one of these if present among all children in the study area. So this type of exploratory analysis is worth undertaking
despite its limitations. Other factors proposed to explore are: diet (including but not limited to deficient sulphur amino
acid), gut microbiota, metabolism alterations, and infections (as a potential precipitating factor) and factors that tend to
coincide with geography, such as temperature, altitude, harvest patterns, infections and waterborne diseases.
Investigate why cross-sectional surveys poorly reflect the type of SAM seen in treatment facilities
Many factors can explain the weak relationship observed between survey and clinical data. First, there may be a problem
in data collection of oedema. Research should try to improve standardisation of the oedema assessment and test its
reproducibility. The use of physical models or of photographs showing different degrees of oedema should be tested.
Measure incidence rather than prevalence
Incidence studies may be better suited to more accurately describe the burden of oedema in countries. Actual incidence
of kwashiorkor was measured in Malawi, with 2.6% of children developing kwashiorkor during 20 weeks follow up (50).
Studies such as this may be a better source of data on kwashiorkor than cross-sectional studies, so they could potentially
supplement similar databases in the future.
The duration of each episode of oedema and different treatment-seeking behaviours deserve further investigation
Special attention should be given to variations of these different factors in relation to the degree of associated wasting.
It is quite plausible that when this level of associated malnutrition is low, oedema resolves more rapidly and is not
perceived as worrisome by carers, as it would be in situations with high levels of associated malnutrition. On the other
hand, in situations where oedema is associated with a high mortality, the early death of these children may explain their
low prevalence in surveys but a high level of treatment-seeking behaviour. Clinical studies examining the relationship
between MUAC, mortality, response to treatment and duration of oedema under treatment could shed light on these
possible factors and allow practitioners to identify areas of treatment in need of improvement.
Examine the effect of the degree of oedema and of background malnutrition on the prognosis of kwashiorkor
There is little data on the relationship between the degree of oedema, the associated malnutrition and the associated
risk of mortality. Improved knowledge of risk associated with the degree of oedema and of background malnutrition, as
assessed by MUAC (as it is less influenced by oedema), would help to better choose between different treatment options.
Other important aspects to try to gather data on are:
1
2
3
Comorbidities associated with kwashiorkor and marasmus (e.g. diarrhoea, fever, HIV status, etc.).
Degree of chronic malnutrition (stunting, underweight) in each of these cases. Little information was
found on kwashiorkor’s relationship with stunting, and this should be further explored.
Response to treatments, stratified by malnutrition type, as well as co-morbidities.
42
PUTTING
CHILD KWASHIORKOR
ON THE MAP
8 CONCLUSION
As highlighted above, the data we have on kwashiorkor are limited, of poor quality and sporadic, despite this being an
important disease with high mortality. The dearth of clear data suggests that there really needs to be a more systematic
and standardized approach in collecting better data that is of immediate use for practitioners as well as researchers in
the future.
There are likely more children with kwashiorkor than are reported in surveys, as the admission data indicates, and these
children pose a huge resource burden to the health care system in many countries. Appropriate resources need to be
devoted to screening, referring and treating children with kwashiorkor.
The authors would like to urge more studies to include kwashiorkor and for the health and nutrition sector to improve
collection and availability of data on oedema within the SAM management programmes, community work and health
systems and to identify the extent of kwashiorkor in order to generate donor and Government interest and leverage
more resources for preventing kwashiorkor.
PUTTING
CHILD KWASHIORKOR
ON THE MAP
43
REFERENCES
1 ICD-10 Version: 2010 [Internet]. ICD-10. Available from: http://apps.who.
int/classifications/icd10/browse/2010/en#/E40-E46
2 WHO | International Classification of Diseases (ICD) Revision [Internet].
WHO. [cited 2015 Nov 24]. Available from: http://www.who.int/
classifications/icd/revision/icd11faq/en/
3 Heikens GT, Manary M. 75 years of Kwashiorkor if Africa. Malawi Med J.
2009 Sep;21(3):96–100.
4 Classification of infantile malnutrition. Lancet Lond Engl. 1970 Aug
8;2(7667):302–3.
5 WHO | Training course on the management of severe malnutrition
[Internet]. WHO. [cited 2015 Nov 24]. Available from: http://www.who.int/
nutrition/publications/severemalnutrition/training_inpatient_MSM/en/
6 Ahmed T, Rahman S, Cravioto A. Oedematous malnutrition. Indian J Med
Res. 2009 Nov;130(5):651–4.
7 Jackson AA. Severe undernutrition in Jamaica. Kwashiorkor and
marasmus: the disease of the weanling. Acta Paediatr Scand Suppl.
1986;323:43–51.
8 The Lancet. Maternal and Child Nutrition series [Internet]. 2013. Available
from: http://www.thelancet.com/series/maternal-and-child-nutrition
9 Global nutrition report 2014: Actions and accountability to accelerate the
world’s progress on nutrition. S.l; S.l: Washington, D.C; 2014.
10 Brock JF. Survey of the world situation on kwashiorkor. Ann N Y Acad Sci.
1954 May 10;57(6):696–713.
11 Joint Child Malnutrition Estimates (UNICEF/WHO/World Bank) | Data
[Internet]. 2014 [cited 2015 Nov 24]. Available from: http://data.worldbank.
org/child-malnutrition
12 Briend A, Collins S, Golden M, Manary M, Myatt M. Maternal and child
nutrition. Lancet Lond Engl. 2013 Nov 9;382(9904):1549.
13 Frison S, Checchi F, Kerac M. Omitting edema measurement: how much
acute malnutrition are we missing? Am J Clin Nutr. 2015 Sep
16;ajcn108282.
14 Williams CD. Protein-calorie malnutrition. Lancet Lond Engl.
1972 Jun 17;1(7764):1333.
15 Bhattacharyya AK, Chattopadhyay PS, Paladhi PK, Ganguli S,
Bhattacharyya N. Kwashiorkor and marasmus: changing hospital incidence
of syndromic presentation (1957-88). Indian Pediatr. 1990 Nov;27(11):1191–8.
27 World Health Organization, Nutrition for Health Deveand lopment.
Guideline. [Internet]. 2013 [cited 2016 Mar 8].
Available from: http://www.ncbi.nlm.nih.gov/books/NBK190328/
28 Andre A, Myatt M, Dent N, Brown R. Putting kwashiorkor on the map
[Internet]. CMAM Forum; 2013. Available from: http://www.cmamforum.
org/Pool/Resources/ Putting-kwashiorkor-on-the-map-CMAM-Forum-2013.pdf
29 Global SAM Management Update - Summary of Findings, September
2013 [Internet]. ReliefWeb. 2013 [cited 2015 Nov 24]. Available from:
http://reliefweb.int/report/world/global-sam-management-updatesummary-findings-september-2013
30 NUTRIDASH 2013 Global Report on the Pilot Year [Internet]. 2013.
Available from: http://www.cmamforum.org/Pool/Resources/
Global-NutriDash-2013-report-UNICEF-2014.pdf
31 Annegers JF. The protein-calorie ratio of West African diets and their
relationship to protein calorie malnutrition. 1973;2:225–35.
32. Golden MH, Ramdath D. Free radicals in the pathogenesis of kwashiorkor.
Proc Nutr Soc. 1987 Feb;46(1):53–68.
33
Lin CA, Boslaugh S, Ciliberto HM, Maleta K, Ashorn P, Briend A, et al.
A prospective assessment of food and nutrient intake in a population of
Malawian children at risk for kwashiorkor. J Pediatr Gastroenterol Nutr.
2007 Apr;44(4):487–93.
34 Kismul H, Van den Broeck J, Lunde TM. Diet and kwashiorkor:
a prospective study from rural DR Congo. PeerJ. 2014;2:e350.
35 Roediger WE. New views on the pathogenesis of kwashiorkor: methionine
and other amino acids. J Pediatr Gastroenterol Nutr. 1995 Aug;21(2):130–6.
36 Jahoor F. Effects of decreased availability of sulfur amino acids in severe
childhood undernutrition. Nutr Rev. 2012 Mar;70(3):176–87.
37
Amadi B, Fagbemi AO, Kelly P, Mwiya M, Torrente F, Salvestrini C, et al.
Reduced production of sulfated glycosaminoglycans occurs in
Zambian children with kwashiorkor but not marasmus. Am J Clin Nutr.
2009 Feb;89(2):592–600.
38
Golden MH. Nutritional and other types of oedema, albumin,
complex carbohydrates and the interstitium - a response to Malcolm
Coulthard’s hypothesis: Oedema in kwashiorkor is caused by hypo-albumi
naemia. Paediatr Int Child Health. 2015 Apr 6;2046905515Y0000000010.
39 Annegers JF. Protein quality of West African foods. Ecol Food Nutr.
1974;3:125–30.
16 Oyelami OA, Ogunlesi TA. Kwashiorkor--is it a dying disease? South Afr
Med J Suid-Afr Tydskr Vir Geneeskd. 2007 Jan;97(1):65–8.
40 Tremaroli V, Bäckhed F. Functional interactions between the gut
microbiota and host metabolism. Nature. 2012 Sep 13;489(7415):242–9.
17 Black RE, Allen LH, Bhutta ZA, Caulfield LE, de Onis M, Ezzati M, et al.
Maternal and child undernutrition: global and regional exposures and
health consequences. Lancet Lond Engl. 2008 Jan 19;371(9608):243–60.
41 Crane RJ, Jones KD, Berkley JA. Environmental enteric dysfunction:
An overview. Food Nutr Bull. 2015;36(Supplement 1):76S – 87S.
18 Ndekha MJ. Kwashiorkor and severe acute malnutrition in childhood.
Lancet Lond Engl. 2008 May 24;371(9626):1748; author reply 1749.
42 Mwangome MK, Fegan G, Prentice AM, Berkley JA. Are diagnostic criteria
for acute malnutrition affected by hydration status in hospitalized children?
A repeated measures study. Nutr J. 2011;10:92.
19 Black RE, Victora CG, Walker SP, Bhutta ZA, Christian P, de Onis M, et al.
Maternal and child undernutrition and overweight in low-income and
middle-income countries. Lancet Lond Engl. 2013 Aug 3;382(9890):427–51.
43 Sandiford P, Paulin FH. Use of mid-upper-arm circumference for nutritional
screening of refugees. Lancet Lond Engl. 1995 Apr 29;345(8957):1120.
20 Briend A. Kwashiorkor: still an enigma–the search must go on. In: CMAM
Forum Technical Brief [Internet]. 2014 [cited 2015 Nov 24]. Available from:
http://fr.cmamforum.org/Pool/Resources/Kwashiorkor,-still-an-enigmaCMAM-Forum-Dec-2014.pdf
21 Smith MI, Yatsunenko T, Manary MJ, Trehan I, Mkakosya R, Cheng J, et al.
Gut microbiomes of Malawian twin pairs discordant for kwashiorkor.
Science. 2013 Feb 1;339(6119):548–54.
22 Briend A, Wojtyniak B, Rowland MG. Arm circumference and other factors
in children at high risk of death in rural Bangladesh. Lancet Lond Engl.
1987 Sep 26;2(8561):725–8.
44
Myatt M, Khara T, Collins S. A review of methods to detect cases of
severely malnourished children in the community for their admission into
community-based therapeutic care programs. Food Nutr Bull. 2006
Sep;27(3 Suppl):S7–23.
45
Blackwell N, Myatt M, Allafort-Duverger T, Balogoun A, Ibrahim A,
Briend A. Mothers Understand And Can do it (MUAC): a comparison of
mothers and community health workers determining mid-upper arm
circumference in 103 children aged from 6 months to 5 years. Arch Public
Health Arch Belg Santé Publique. 2015;73(1):26.
46 Roediger WE. New views on the pathogenesis of kwashiorkor: methionine
and other amino acids. J Pediatr Gastroenterol Nutr. 1995 Aug;21(2):130–6.
23
Dramaix M, Hennart P, Brasseur D, Bahwere P, Mudjene O, Tonglet R,
et al. Serum albumin concentration, arm circumference, and oedema
and subsequent risk of dying in children in central Africa. BMJ. 1993
Sep 18;307(6906):710–3.
47
Ferguson EL, Darmon N, Fahmida U, Fitriyanti S, Harper TB, Premachandra
IM. Design of optimal food-based complementary feeding
recommendations and identification of key “problem nutrients” using goal
programming. J Nutr. 2006 Sep;136(9):2399–404.
24
Prudhon C, Golden MH, Briend A, Mary JY. A model to standardise
mortality of severely malnourished children using nutritional status on
admission to therapeutic feeding centres. Eur J Clin Nutr.
1997 Nov;51(11):771–7.
48
Fahmida U, Kolopaking R, Santika O, Sriani S, Umar J, Htet MK, et al.
Effectiveness in improving knowledge, practices, and intakes of “key
problem nutrients” of a complementary feeding intervention developed by
using linear programming: experience in Lombok, Indonesia. Am J Clin
Nutr. 2015 Mar;101(3):455–61.
25 Talbert A, Thuo N, Karisa J, Chesaro C, Ohuma E, Ignas J, et al. Diarrhoea
complicating severe acute malnutrition in Kenyan children: a prospective
descriptive study of risk factors and outcome. PloS One. 2012;7(6):e38321.
26 Trehan I, Goldbach HS, LaGrone LN, Meuli GJ, Wang RJ, Maleta KM, et al.
Antibiotics as part of the management of severe acute malnutrition.
N Engl J Med. 2013 Jan 31;368(5):425–35.
49 Courtright P, Canner J. The distribution of kwashiorkor in the southern
region of Malawi. Ann Trop Paediatr. 1995 Sep;15(3):221–6.
50 Ciliberto H, Ciliberto M, Briend A, Ashorn P, Bier D, Manary M. Antioxidant
supplementation for the prevention of kwashiorkor in Malawian
44
PUTTING
CHILD KWASHIORKOR
ON THE MAP
ANNEXES
ANNEXES
Annex 1: Project information sheet shared with partners
Annex 1: Project information sheet shared with partners
47
PUTTING
CHILD KWASHIORKOR
ON THE MAP
48
45
46
PUTTING
CHILD KWASHIORKOR
ON THE MAP
Annex 2: Letter of agreement
Agency Request Letter – Nov 2014
Dear XXXX
ACF-UK is writing on behalf of the Kwashiorkor Mapping Core Group (comprised of ACF-UK, the CMAM
Forum, UNICEF, WHO), to request your organisation to share with us raw anthropometric survey datasets, in
particular from those surveys conducted within the last 5 years.
This work forms part of Phase Two of “Putting Kwashiorkor on the Map” project, which aims to develop a
more comprehensive global map to help provide an estimate of the burden and numbers affected by oedematous
malnutrition. This work will be conducted under the guidance of a Technical Advisory Committee and in
consultation with the agencies sharing data. This work follows on from the preliminary map and request for data
sent out by the CMAM Forum in October 20131.
The aim is to collate these data in order to produce an aggregated database, which will enable us to gain a more
accurate idea of actual numbers of cases of nutritional oedema and the estimated global burden. The dataset
will primarily be used to answer the questions outlined in the attached document which explains the project in
more detail. The data will not be used for any other use than stated in the attached sheet without permission and
it will not be shared to any third party without permission. These data are anonymous and there is no risk of
identification of individuals (ie no personal identifiers). The future of the combined database will be re-discussed
with all agencies at the close of this project.
Any data shared by your organisation will be rigorously checked, cleaned and verified to ensure that there is no
duplication with other or existing data sets. In return for the data shared, we will acknowledge your organisation
in any maps or reports produced. We would also be able to return the files you provided in cleaned format if
required.
We attach a standard project agreement letter that would be signed between your agency and Action Against
Hunger UK (representing the Kwashiorkor Mapping Core Group) and an information sheet giving more details
about this project.
If you require further information, please contact Nicky Dent at [email protected]
We look forward to future collaboration.
Yours sincerely,
Saul Guerrero
Director of Operations
Action Against Hunger UK
On behalf of the Kwashiorkor Mapping Core Group
http://www.cmamforum.org/Pool/Resources/Putting-kwashiorkor-on-the-map-CMAM-Forum-2013.pdf
http://www.cmamforum.org/Pool/Resources/Donner-sa-place-au-kwashiorkor-CMAM-Forum-2013.pdf
1
PUTTING
CHILD KWASHIORKOR
ON THE MAP
47
Annex 3: Surveys by contributing agency
Country
# surveys # children
Contributing agencies
Afghanistan
43*
48878
ACF, MSF, and UNICEF/MOH
Albania
1
906ACF
Angola
22
17361
ACF and MSF
Bangladesh
26
13480
ACF, Tdh, and UNHCR
Benin
7*
7930UNICEF/MOH
Bolivia
3
1775
ACF and UNICEF/MOH
Botswana
1
164UNHCR
Burkina Faso
50
40446
ACF, Tdh, UNHCR, UNICEF/MOH
Burundi
25
14742
ACF, Concern, MSF, UNHCR, and World Vision
Cameroon
9
5642
ACF, MSF, UNICEF/MOH, and Zerca y Lejos
Central African Republic
58
36443
ACF and UNICEF/MOH
Chad
201
124096
ACF, Concern, IRC, MSF, UNHCR, and UNICEF/MOH
Congo
264
227390
ACF, Concern, MSF, Save, UNHCR, and UNICEF/MOH
Cote d’Ivoire
49
24233
ACF and UNICEF/MOH
Djibouti
7
2516UNHCR
Eritrea
3
1969
Save and UNHCR
Ethiopia
233
155494
ACF, Concern, GOAL, MSF, Save, and UNHCR
The Gambia
8
6769
UNICEF/MOH
Guatemala
2
625
ACF and FEWS NET
Guinea
12
9603
ACF and UNICEF/MOH
Guinea-Bissau
13
216UNICEF/MOH
Haiti
49
39764ACF
India
8
5182
ACF, MSF, and Tdh
Indonesia
3
1749
ACF and MSF
Jordan
2
802UNHCR
Kenya
107
71475
ACF, Concern, IMC, Save, UNHCR, UNICEF/MOH, & World Vision
Liberia
52
31230
ACF, UNHCR, and UNICEF/MOH
Macedonia
1
865ACF
Madagascar
4
3180
ACF and MSF
Malawi
16
16277
Concern, GOAL, MSF, Save, and UNHCR
Mali
14
10968
ACF and UNICEF/MOH
Mauritania
56
36432
ACF, UNHCR, and UNICEF/MOH
Mozambique
11
3867
ACF, MSF, and UNHCR
Myanmar
22
14391
ACF, Plan, and Save
Nepal
12
7650
ACF, Concern, and UNHCR
Nicaragua
2
1017ACF
Niger
38
49411
ACF, Concern, GOAL, and MSF
Nigeria
107
66398
IMC, MSF, Save, and UNICEF/MOH
Pakistan
18
14200
ACF and MSF
Philippines
12
6220
ACF, UNICEF/MOH, Phillipine National Nutrition Cluster,
and National Nutrition Council
Rwanda
21
13534
ACF, Concern, MSF, and UNHCR
Senegal
7
8445UNICEF/MOH
Sierra Leone
58*
64028
ACF, GOAL, MSF, and UNICEF/MOH
Somalia
227
237498
ACF, FSNAU, and MSF
South Sudan
140
96959
ACF, Concern, GOAL, IRC, MSF, UNHCR, and World Vision
Sri Lanka
3
2586
ACF and MSF
Sudan
136
109099
ACF, Concern, GOAL, MSF, Save, and UNHCR
Tajikistan
5
4337ACF
Tanzania
7
4903
ACF, MSF, and UNHCR
Thailand
2
1812MSF
Togo
19*
11976UNICEF/MOH
Uganda
74
48503
ACF, GOAL, MSF, and UNHCR
Yemen
2
816UNHCR
Zambia
5
2095
MSF, UNHCR, and World Vision
Zimbabwe
1
700
World Vision
48
PUTTING
CHILD KWASHIORKOR
ON THE MAP
Annex 4: Additional maps
PROPORTION OF SAM CASES DEFINED BY WHZ <-3 OR OEDEMA WITH KWASHIORKOR (2006-2015) (EXCLUDING
COUNTRIES WITH <20 CASES OF SAM DEFINED BY WHZ AND OR OEDEMA)
PROPORTION OF SAM CASES DEFINED BY MUAC <115mm OR WFZ<-3 OR OEDEMA WITH KWASHIORKOR
(2006-2015) FROM UNHCR AND OTHER REFUGEES DATA (EXCLUDING COUNTRIES WITH <20 CASES OF SAM)
PUTTING
CHILD KWASHIORKOR
ON THE MAP
49
Annex 5: Description of data sources
Contains
Type
id
Unique identifier Filename of linked dataset
Text
1
unsr
UN sub-region
Text
2
Column name
countryCountry
Notes Required?
Text
✓
✓
2
✓
Text
3
Text
3
regionRegion (highest level administrative division)
within country
districtDistrict (next highest level administrative division)
within region
popType
1 = rural, 2 = urban,
3 = mixed rural/urban, 4 = refugee, 5 = IDP,
6 = other, 7 = unknown
Number (coded)3;4
month
Month of survey
Number
year
Year of survey
Number
agency
Name of agency
Text
reportFile
Name of report file
Text
rebuild
Rebuild database flag
Number
✓
✓
✓
1;3
4
1. These are file names 2. Could be coded 3. Code as NA if data are not available 4.Binary categorical variables (all allowed) 5. This is a binary flag that causes the database
system to recalculate nutritional indices and associated flags (used to indicate recently added datasets and clears when indices have been added).
Annex 6: Survey dataset
Column name
Contains
Type
Notes Required?
✓
age
Age in months
Number
2
✓
sexSex (1=Male 2=Female)
Number2
✓
weight
Weight in kg (##.#)
Number2
✓
height
Height in cm (###.#)
Number2
✓
muac
MUAC in mm (###)
Number2
✓
oedema
Bilateral pitting oedema (1=Present 2=Absent)Number
2
✓
psu
Cluster identifier
1. Record as 9999 if missing in all records 2. Record as NA if data not available
Number
1
50
PUTTING
CHILD KWASHIORKOR
ON THE MAP
Annex 7: UNHCR admission data
I = Inpatient O = Outpatient
UNHCR KWASHIORKOR PERCENTAGES (TOTAL NUMBER OF KWASHIORKOR ADMISSIONS/TOTAL NUMBER
OF SAM ADMISSIONS) FROM 2008-2014
Country
2008
2009
2010
2011
2012
2013
2014
2015
BangladeshI
12.5% (1/8)10.0% (2/20)2.4% (2/83)0% (0/108)1.1% (1/87)1.6% (1/62)0% (0/52)
3.4% (2/58)0% (0/58)0.3% (1/303)0% (0/351)0% (0/238)0% (0/263)0.4% (1/247)0.0% (0/6)
O
CameroonI
0% (0/4)13.3% (2/15)0% (0/17)
O1.4% (3/214)2.9% (12/410)2.6% (37/1,445)
CAR
O 1.5% (1/66)
Chad
I7.3% (4/55)10.5% (31/296)15.5% (15/296)29.6% (92/311)14.0% (43/308)20.5% (20/292)12.5% (67/536)
O
21.7% (5/23) 10.6% (62/587) 8.3% (77/929) 2.0% (39/1,938) 1.2% (76/6,144) 2.5% (131/5,253) 2.4% (106/4442)
Congo (DR)
I
21.6% (8/37)80.0% (20/25)
O 17.0% (37/218)19.9% (99/498)
Djibouti
I20.0% (10/50)23.3% (10/43)3.0% (2/67)17.9% (5/28)7.9% (7/38)7.7% (3/39)
O 2.0% (2/101) 15.6% (23/147) 3.0% (4/133) 4.8% (6/125) 0% (0/28) 0.0% (0/124)
Eritrea I0.0% (0/2)
O 6.7% (2/30) 0.0% (0/35) 0.0% (0/1)
Ethiopia I87.5% (7/8)25.9% (14/54)8.2% (8/61)17.2% (258/1,497)15.0% (66/140)14.2% (64/4,52)18.8% (145/771)
O7.1% (2/28)12.0% (26/217)1.3% (8/600) 1.1% (136/12,725) 2.4% (85/3,567) 0.6% (15/2,466)2.3% (87/3,834)
Kenya
I22.2% (103/463)20.3% (218/1,073)5.4% (73/1,353)5.7% (400/6,978)10.1% (291/2,887)6.4% (131/2,062)8.2% (213/2,612)
Liberia O 12.2% (22/181)4.3% (3/70)
NamibiaI0.0% (0/1)
O20.0% (1/5)
Nepal
I0.0% (0/2)0.0% (0/10)0.0% (0/2)0.0% (0/1)
O Rwanda
I34.1% (15/44)0.0% (0/14)
O South Sudan
I
O0.0% (0/897)2.7% (91/3,387)4.9% (68/1,395)1.3% (2/149)
Sudan
I1.1% (4/372)0.0% (0/292)0% (0/366)1.9% (1/52)0.6% (1/169)
O0.2% (1/480)
Tanzania
I40.0% (50/125)53.2% (99/186)52.2% (36/69)62.0% (31/50)23.1% (3/13)44.4% (8/18)75.0% (3/4)
O33.3% (27/81)17.2% (22/128)13.8% (15/109)22.7% (44/194)21.7% (13/60)6.1% (2/33)28.6% (2/7)
33.3% (8/24) 39.3% (24/61) 34.8% (8/23) 0.0% (0/13)
0.0% (0/19) 0.0% (0/7)0.0% (0/1)
1.6% (6/376)
0.0% (0/29)0.2% (2/9,11)0.4% (1/266)
0.0% (0/21)
0.2% (1/653)0.2% (1/646)0.0% (0/90)0.2% (1/549)
Uganda
I9.1% (1/11)33.3% (14/42)32.0% (63/197)49.7% (150/302)44.2% (125/283)44.4% (8/18)
O2.0% (1/51)19.6% (31/158)15.1% (285/1,887)29.5% (338/1,145)
34.9% (236/677 29.7% (11/37)
Yemen I0.0% (0/1)
O 3.2% (1/31)2.1% (3/141)1.4% (4/288)1.9% (5/268)1.0% (3/289)0.0% (0/28)11.1% (2/18)
Zambia I7.1% (2/28)27.3% (3/11)
O25.0% (1/4)18.5% (20/108)46.2% (6/13)25.0% (4/16)
TOTALS
Kwash Admissions
SAM Admissions
Country
Bangladesh
Cameroon
CAR Chad Congo
Djibouti Eritrea # of Camps
3
7
1
20
4
2
1
367
864
662
1,836
1,362
1,250
1,530
176
1,274
4,894
8,156
42,278
29,167
26,570
27,858
1,767
Gender Ratio
87.5% (n=7:1)
39.2% (n=20:31)
0% (n=0:1)
49.1% (n=412:427)
43.9% (n=72:92)
63.2% (n=43:25)
0% (n=0:2)
Country
Ethiopia Kenya
Liberia
Namibia
Nepal
Rwanda South Sudan
# of Camps
19
6
4
1
2
3
7
Gender Ratio
60.5% (n=555:363)
51.0% (n=2,049:1,971)
28.0% (n=7:18)
100% (n=1:0)
63.5% (n=40:23)
49.4% (n=81:83)
Country
Sudan
Tanzania
Uganda
Yemen
Zambia
# of Camps
7
1
8
3
3
Gender Ratio
100% (n=10:0)
54.2% (n=207:175)
50.9% (n=643:620)
43.8% (n=7:9)
47.4% (n=18:20)
PUTTING
CHILD KWASHIORKOR
ON THE MAP
51
Annex 8: Other tables
PROPORTION OF SAM CASES (DEFINED BY MUAC <115MM, WHZ<-3 OR OEDEMA) WITH KWASHIORKOR (2006-2015)
Country
# SAM cases
Afghanistan
Bangladesh
Benin
Bolivia
Botswana
Burkina Faso
Burundi
Cameroon
Central African Republic
Chad
Congo (Kinshasa)
Cote d’Ivoire
Djibouti
Eritrea
Ethiopia
Gambia
Guatemala
Guinea
Guinea-Bissau
Haiti
India
Jordan
Kenya
Liberia
Madagascar
Malawi
Mali
Mauritania
Mozambique
Myanmar
Nepal
Nicaragua
Niger
Nigeria
Pakistan
Philippines
Rwanda
Senegal
Sierra Leone
Somalia
South Sudan
Sri Lanka
Sudan
Tanzania
Togo
Uganda
Yemen
Zambia
Zimbabwe
# Oedema cases % SAM cases
with oedema
967
272
149
6
1
1076
27
134
968
3038
8236
383
101
21
4637
139
5
98
85
197
205
3
2326
128
53
152
322
655
10
256
340
12
1149
2213
388
104
47
181
863
11608
2344
15
1544
17
172
764
17
53
22
71
3
4
3
0
45
0
30
129
179
1876
72
7
0
218
2
3
3
1
51
5
0
100
16
0
42
1
7
2
6
28
8
14
113
4
0
16
2
112
666
106
1
112
4
40
49
9
17
12
Lower 95% CI Upper 95% CI
7.34 5.789.17
1.10 0.233.19
2.68 0.746.73
50.00 11.8188.19
0.00 0.0097.50
4.18
3.07
5.56
0.00 0.0012.77
22.39 15.6430.39
13.33
11.25
15.63
5.89 5.086.79
22.78
21.88
23.70
18.80
15.01
23.08
6.93 2.8313.76
0.00 0.0016.11
4.70 4.115.35
1.44 0.175.10
60.00 14.6694.73
3.06 0.648.69
1.18 0.036.38
25.89 19.9232.59
2.44 0.805.60
0.00 0.0070.76
4.30 3.515.20
12.50 7.3219.50
0.00 0.006.72
27.63 20.7035.46
0.31 0.011.72
1.07 0.432.19
20.00 2.5255.61
2.34 0.865.03
8.24 5.5411.68
66.67 34.8990.08
1.22 0.672.04
5.11 4.236.11
1.03 0.282.62
0.00 0.003.48
34.04 20.8649.31
1.10 0.133.93
12.98
10.81
15.40
5.74 5.326.18
4.52
3.72
5.44
6.67
0.17
31.95
7.25 6.018.66
23.53 6.8149.90
23.26 17.1630.29
6.41 4.788.39
52.94 27.8177.02
32.08 19.9246.32
54.55 32.2175.61
52
PUTTING
CHILD KWASHIORKOR
ON THE MAP
PROPORTION OF SAM CASES (DEFINED BY WHZ OR OEDEMA) WITH KWASHIORKOR (2006-2015)
Country
# SAM cases
Bolivia
Guatemala
Yemen
Nicaragua
Zimbabwe
Rwanda
Zambia
Congo (Kinshasa)
Haiti
Malawi
Tanzania
Togo
Cote d’Ivoire
Cameroon
Central African Republic
Mozambique
Sierra Leone
Liberia
Nepal
Afghanistan
Uganda
Sudan
Djibouti
Somalia
Chad
Ethiopia
Nigeria
Sri Lanka
Burkina Faso
South Sudan
Kenya
Guinea
Myanmar
Benin
India
Guinea-Bissau
Niger
Mauritania
Gambia
Pakistan
Bangladesh
Senegal
Mali
Burundi
Eritrea
Jordan
Madagascar
Philippines
# Oedema cases % SAM cases
with oedema
4
3
4
3
13
9
12
8
18 12
35 17
36 17
4928
1951
146 57
124 43
12
4
133 41
272
73
116 31
597
140
10
2
621
113
104 17
230 28
603 73
469 52
1228 112
81
7
8835 679
2443 185
3147 234
1690 118
15
1
859
45
2027
106
1951 100
75
3
178 7
121 4
164 5
66
2
785 14
503 8
128 2
263 4
214 3
172 2
267 1
11
0
19
0
2
0
20
0
93 0
Lower 95% CI Upper 95% CI
75.0 19.499.4
75.0 19.499.4
69.2 38.690.9
66.7 34.990.1
66.7 41.086.7
48.6 31.466.0
47.2 30.464.5
39.6
38.2
41.0
39.0 31.147.5
34.7 26.443.7
33.3 9.965.1
30.8 23.139.4
26.8
21.7
32.5
26.7 18.935.7
23.5
20.1
27.1
20.0 2.555.6
18.2
15.2
21.5
16.3 9.824.9
12.2 8.217.1
12.1 9.615.0
11.1 8.414.3
9.1 7.610.9
8.6 3.517.0
7.7 7.18.3
7.6 6.68.7
7.4 6.58.4
7.0 5.88.3
6.7
0.2
31.9
5.2
3.8
6.9
5.2
4.3
6.3
5.1 4.26.2
4.0 0.811.2
3.9 1.67.9
3.3 0.98.2
3.0 1.07.0
3.0 0.410.5
1.8 1.03.0
1.6 0.73.1
1.6 0.25.5
1.5 0.43.8
1.4 0.34.0
1.2 0.14.1
0.4 0.02.1
0.0 0.028.5
0.0 0.017.6
0.0 0.084.2
0.0 0.016.8
0.0 0.03.9
PUTTING
CHILD KWASHIORKOR
ON THE MAP
53
MALAWI NATIONAL DATA
Month
2010
2011
2012
2013
January I66.2% (1,103/1,667) 65.5% (1,186/1,810)73.1% (1,198/1,639) 60.3% (1,914/3,175) O
58.0% (1731/2,983) 66.0% (1,859/2,815) 2014
73.1% (789/1,080)48.8% (522/1,070)
48.4% (1,113/2,301)
48.6% (1,468/3,023)
FebruaryI 69.8% (952/1,363)69.7% (1,163/1,669)73.9% (1,098/1,486)72.4% (1,112/1,535)60.0% (644/1,079)
O58.9% (1,599/2,717)60.7% (1,953/3,219)63.7% (1,800/2,825)58.0% (1,591/2,743)49.4% (1,976/4,002)
March I66.1% (906/1,371)68.0% (785/1,154)69.3% (645/931)70.3% (727/1,034)56.9% (583/1,025)
O
59.7% (1,611/2,697)58.8% (1,615/2,748)59.1% (1,096/1,855)50.1% (1,087/2,171)46.6% (1,404/3,015)
AprilI 65.2% (758/1,163)
63.5% (512/806)
74.0% (605/818)
59.9% (419/700)
62.6% (428/684)
O54.7% (1,282/2,342)54.6% (1,166/2,136)62.0% (717/1,157)51.5% (843/1,637)45.7% (1,111/2,431)
May
I65.5% (685/1,046)63.6% (462/726)69.1% (403/583)46.0% (394/857)54.4% (277/509)
O53.9% (1,047/1,944)53.6% (1,052/1,964)64.8% (908/1,401)53.5% (925/1,729)41.1% (733/1,784)
June I63.8% (561/879)64.0% (361/564)73.4% (448/610)43.6% (347/795)54.4% (278/511)
O59.3% (1,063/1,792)53.7% (866/1,614)61.4% (790/1,286)49.6% (871/1,755)
47.7% (906/1,901)
July I65.2% (453/695)66.3% (395/596)63.6% (288/453)60.0% (250/417)59.1% (234/396)
O 59.1% (1,023/1,732)
57.4% (826/1,438)
64.7% (905/1,398)
49.2% (751/1,525)
43.4% (771/1,778)
August
I
68.5% (567/828)65.2% (306/469)62.5% (326/522)58.3% (217/372)56.9% (272/478)
O57.8% (772/1,336)59.2% (804/1,359)58.2% (896/1,539)43.5% (612/1,408)40.1% (711/1,774)
SeptemberI
62.6% (503/803)60.3% (418/693)64.3% (296/460)53.2% (192/361)60.1% (264/439)
O57.4% (831/1,448)58.6% (953/1,626)58.2% (820/1,409)47.7% (765/1,604)37.3% (588/1,575)
OctoberI
58.0% (531/915)69.5% (439/632)63.7% (445/699)48.2% (290/602)52.7% (227/431)
O55.8% (844/1,513)57.9% (998/1,723)60.9% (984/1,615)44.6% (718/1611)
NovemberI
40.9% (708/1,733)
55.6% (508/913)67.4% (538/798)62.7% (458/731)54.5% (381/699)50.0% (256/512)
O53.8% (1,143/2,124)52.3% (929/1,775)61.5% (929/1,511)44.5% (874/1,964)41.5% (617/1,488)
December
I
62.9% (673/1,070)65.4% (585/895)66.2% (437/660)53.6% (287/535)51.8% (268/517)
O
54.0% (1,060/1,963)56.9% (964/1,694)56.3% (850/1,511)41.5% (685/1,651)45.7% (875/1,913)
I = Inpatient O = Outpatient
54
PUTTING
CHILD KWASHIORKOR
ON THE MAP
INPATIENT ADMISSIONS ONLY: UGANDA MULAGO HOSPITAL (2009-2014))
Month
2009
2010
January
February 2011
2012
73.3% (88/120)46.5% (33/71) 2013
2014
42.3% (30/71)54.3% (44/81)
53.3% (48/90)
52.4% (33/63)50.0% (32/64)42.4% (36/85)41.9% (31/74)53.8% (35/65)
March68.4% (54/79)50.7% (38/75)33.1% (39/118)48.6% (34/70)61.8% (42/68)
April63.8% (51/80)55.4% (51/92)50.5% (55/109)37.2% (48/129)37.6% (32/85)45.9% (45/98)
May72.1% (44/61)66.7% (60/90)59.4% (79/133)42.7% (50/117)46.3% (37/80)56.1% (74/132)
June58.1% (43/74)56.5% (48/85)59.9% (97/162)39.8% (64/161)53.1% (52/98)55.1% (76/138)
July66.2% (51/77)61.0% (61/100)47.7% (73/153)44.9% (75/167)51.0% (51/100)51.4% (75/146)
August63.9% (39/61)57.6% (34/59)49.3% (37/75)43.9% (43/98)48.2% (40/83)70.9% (78/110)
September73.1% (49/67)56.4% (31/55)46.3% (44/95)44.4% (36/81)45.6% (36/79)61.5% (48/78)
October66.7% (62/93)65.8% (50/76)48.0% (48/100)57.7% (45/78)51.6% (63/122)54.2% (39/72)
November65.1% (69/106)49.5% (49/99)54.3% (38/70)54.3% (38/70)51.6% (63/122)54.2% (39/72)
December63.0% (58/92)54.2% (39/72)41.5% (34/82)62.3% (43/69)54.4% (49/90)51.4% (36/70)
Total Per Year
Total K <6mths
65.5% (466/711)60.4% (598/990)51.2% (608/1,188)44.0% (547/1,244)50.0% (532/1,063)55.8% (651/1,166)
9
17
18
15
16
17
Average Ratio
Male:Female 297:178 341:274367:259319:243 324:224387:281