ACUTE MALNUTRITION BENCHMARKING SYSTEM FOR GLOBAL HUMANITARIAN RESPONSE

Page created by Sherry Norris
 
CONTINUE READING
ACUTE MALNUTRITION
    BENCHMARKING SYSTEM FOR GLOBAL HUMANITARIAN RESPONSE

Helen Young1, Susanne Jaspars2, Tanya Khara and Steve Collins3
25 November 2005

INTRODUCTION

Acute malnutrition among young children is one of the most widely used indicators of the
extent and severity of a humanitarian crisis. There is a long history of its application in
famine situations, refugee crises and complex emergencies dating back to the 1960s4.
Since that time approaches and methods have been developed and standardized to such an
extent that acute malnutrition has become one of the most standardized and reliable
indicators used in emergencies.

Purpose of the paper
The purpose of this paper is threefold;
   • first, to review the role of nutritional information for assessing the nature of
       severity of the crisis, and as benchmarks for delivery and performance of
       humanitarian assistance;
   • second, to identify the areas of consensus and outstanding technical and
       institutional issues, and;
   • third to make recommendations on establishing an international system for
       assessing the nature and severity of the crisis, and as benchmarks for delivery and
       performance of humanitarian assistance.

The paper first provides background on acute malnutrition and its use. This is followed
by a review of the areas of consensus and outstanding technical issues. Next is a review
of the institutional issues relating to the use of acute malnutrition as benchmarks,
followed by conclusions and recommendations.

Background
The prevalence of wasting and nutritional oedema among children under 5 years of age is
used as an indicator for the prevalence of acute malnutrition. In humanitarian

1
  Feinstein International Famine Center, Friedman School of Nutrition Science and Policy, Tufts
University.
2
  Independent Consultant
3
  Valid International
4
  Early efforts to estimate malnutrition occurred during the Nigerian civil war in Biafra, the famines in
   Ethiopia and among Cambodian refugees in Thailand. Black M. A Cause For Our Times Oxfam the First
   50 Years. Oxford: Oxfam, 1992; Davis LE. Epidemiology of famine in the Nigerian crisis: rapid
   evaluation of malnutrition by height and arm circumference in large populations. Am J Clin Nutr
   1971;24(3):358-64; Rivers. JPW, Holt JFJ, Seaman JA, Bowden MR. Lessons for epidemiology from the
   Ethiopian Famines. Annales de la Societe Belge de Medecine Tropicale 1976;56(4-5):345-357; 4.
   Seaman J, Holt J. The Ethiopian Famine of 1973-74, Wollo Province. Proceedings of the Nutritional
   Society 1975;34(114A).

                                                                                                       1
emergencies, nutritional status is assessed using the Weight-for-Height/ Length (WFH/L)
nutritional index and is based on body measurements (anthropometry). Cut-off’s are
established for classifying moderate or severe acute malnutrition. INGOs and others
usually gather this data using standardized survey techniques and also obtain data on
possible underlying causes of malnutrition. Hence the nutritional index weight-for-height
is used as a population indicator. Many agencies have developed thresholds to classify
the severity of malnutrition in the sampled population. Information on prevalence and
underlying causes is used to identify relief needs, to prioritise affected groups or
geographical areas, to plan nutritional interventions, to target scarce resources, and also to
monitor the effectiveness of aid programmes5.

Over the past two decades the nutrition sector within the humanitarian system has
demonstrated a remarkable degree of collaboration and cohesion in a range of joint
initiatives aimed at addressing some of the more technical issues. These in turn have
greatly contributed to improved practice, in terms of the application of standard
procedures and protocols and the development of a collective process of institutional
learning.     Part of this process has been the regular, although ad hoc, meetings of
representatives of this sector who come together to share and discuss common concerns6.
Two recent global initiatives, Sphere and SMART, have used this existing emergency
nutrition community for consensus building around standards, indicators and methods for
nutrition surveys and programming.

Earlier processes of consensus building, such as Sphere, have recognized the importance
of ensuring a broad representation including national governments and civil society
groups. While this may not be always possible it is nevertheless essential to consider
their role in developing benchmarks and standards in their own countries, how this has
influenced international practice, and subsequently the implications of new
recommendations for promoting good practice locally.

What do we mean by standards and benchmarks?
Within the humanitarian field the term ‘standards’ is used differently depending on the
agency and context. For example in the operational context of humanitarian
programmes, a standard may refer to the particular level of an indicator such as the
provision of 2100 kcal per person per day to food aid dependent populations, i.e. 2,100
kcal is the operational standard. But this same standard corresponds to a ‘key indicator’
in the language of Sphere, whose minimum standards have a rather different meaning.

According to the 2000 edition of the handbook, a ‘minimum standard’ is qualitative in
nature and specifies the minimum levels to be attained in the provision of food security,
nutrition and food aid response. However, review of the 2000 standards shows that some
existing minimum standards allude directly to people’s rights, as expressed in legal
instruments (for example the right to food), while others are more operationally focused,

5
    Young, H., A. Borrel, et al. (2004). "Public nutrition in complex emergencies." The Lancet 365(1909): 1899.
6
 Young, H. (1999). "Public Nutrition in Emergencies: An Overview of Debates, Dilemmas and Decision-
making." Disasters 23(4): 277-292.

                                                                                                              2
which are much more a matter of technically applied good practice (as reflected in
agency policies and guidelines)7. Standards can thus be understood as a unit of
measure, or as an aspiration as for example the standards which are based directly on
human rights8. Clearly there is an opportunity now for addressing some of these
anomalies and reaching consensus on the use of important terms, such as standard,
indicator and benchmark, in such a way that reflect operational practice and the more
general use of these terms.

Given the extensive consensus building that took place as part of the development of the
Sphere minimum standards and indicators, any benchmarking system needs to build on
this. For example, general nutrition support standard 1; “the nutritional needs of the
population are met” has as one of its indicators “levels of moderate and severe
malnutrition are stable at, or declining to, acceptable levels”. The latter would be more
appropriate as a benchmark for monitoring the performance of the humanitarian system
overall. Similarly, the standards and indicators for the correction of severe and
moderate malnutrition have generally been accepted by the international nutrition
community, and can form the basis of a benchmarking system. Where relevant, Sphere
standards, indicators and guidance notes are referred to in this paper.

Another source of confusion relates to ‘standards’ versus ‘standardized’.
Standardization implies that the same template is used in every context9. Methodological
approaches to nutritional surveys have been standardized over the years and there is now
consensus on a wide range of technical details (see discussion below). This
‘standardization’ of humanitarian practice has directly contributed to an improvement in
the overall availability of reliable data10. A range of actors have contributed to this
improvement in practice through standardization, which is reflected in the development
of policies, technical papers, good practice guidelines and a range of capacity
development initiatives including training. Annex 1 shows examples of the range of
policies, good practice guidelines and training activities by the different groups of
stakeholders.

7
  Young, H., A. Taylor, et al. (2004). "Linking Rights and Standards: The Process of Developing 'Rights-
based' Minimum Standards on Food Security, Nutrition and Food Aid." Disasters 28(2): 142-159.
8
  Lowrie, S (2005) Annex to ECHO guidelines on water and sanitation. Lowrie describes the origins of the
use of the word standard in the English language as first: scientific: a unit of measure, and in the military:
as a rallying symbol behind which the army fights.
9
  Lowrie, S (2005). Annex to ECHO guidelines on water and sanitation
10
   Young, H., A. Borrel, et al. (2004). "Public nutrition in complex emergencies." The Lancet 365(1909):
1899.

                                                                                                            3
THE ROLE OF NUTRITION WITHIN HUMANITARIAN INFORMATION
SYSTEMS

The role of nutrition information within the broader humanitarian information system is
generally agreed, which includes:

Early Warning & other food security information systems. Acute malnutrition can be
used as an indicator of food insecurity (if health and care as possible causes are also taken
into account), and together with information on food production, market prices, coping
strategies and population migration patterns can be an early indicator of food crisis11.
The relationship between food security indicators and nutritional status varies between
different population groups, however, as was found for example in an analysis of SC-
UK’s nutritional surveillance data in Ethiopia.12

Nutrition surveillance. Nutrition surveillance in emergencies can be part of a famine
Early Warning System (as for example in the Kenya Arid Lands Drought Monitoring
Programme, or Somalia’s Food Security Assessment Unit - FSAU) or a separate initiative
(for example SC-UK’s system in Ethiopia). The aim of such systems is to monitor trends
in nutritional status, to detect a deteriorating nutritional situation and predict future
change. The focus is therefore on trends rather than absolute levels of malnutrition.
On-going surveillance also allows for the interpretation of prevalences of malnutrition
compared to what is normal for that time of the year, for the population in question.
Both the FSAU in Somalia and SC-UK in Ethiopia are developing baselines which
document seasonal and historical trends in the prevalence of malnutrition.

Initial assessment – rapid needs assessments. Rapid initial assessments may make use of
existing nutritional information, for example trends in the number of malnourished
children coming to MCH clinics, or recent nutritional surveys. Alternatively, such
assessments may include the measurement of MUAC of children amongst the affected
population13, or purposive sampling using WFH amongst some of the worst affected
population14. They key role of nutritional information is that is it one of a number of
indicators to use to assess the severity of crisis. One of the Sphere indicators for
nutrition assessment and analysis states that the underlying causes of malnutrition need to
be investigated before conducting a nutrition anthropometric survey, and thus an
investigation of underlying causes should be part of an initial assessment to determine
risks to nutrition.

11
   Young, H. and Jaspars, S. (1995). Nutrition Matters: People, Food and Famine. IT publications; Kelly,
M. (1992) Entitlements, coping mechanisms, and indicators of access to food: Wollo region, Ethiopia,
1987-88. Disasters, 16, (4): 322-338.
12
   Duffield, A. and Myatt, M. (2004, March). An analysis of Save the Children UK’s and the Disaster
Preparedness and Prevention Commission’s Nutritional Surveillance Programme dataset in some of the
most drought prone areas of Ethiopia, 1995-2001. Draft.
13
   MSF (2004, July). Nutrition Guidelines. Draft.
14
   Jaspars, S and Khogali, H. (2001, May). Oxfam’s approach to nutritional surveys. SC-UK (2004).
Emergency Nutrition Assessment.

                                                                                                           4
Emergency needs assessment. These may be multi-sectoral, or sectoral in focus, e.g.
nutrition/ mortality surveys, food security/livelihoods, public health/ watsan assessments,
livelihoods. Nutritional/mortality surveys are most commonly carried out to estimate the
severity of crisis. The specific objectives will vary according to the agency which carries
them out. So for example, for MSF and ACF this may be to determine the need for
selective feeding programmes, for UNHCR and WFP to also determine the need and type
of general rations required, or for Oxfam or SC-UK to identify appropriate interventions
to address the underlying causes of malnutrition.

It is rare for surveys to estimate the overall extent of the crisis, i.e. to cover the entire
emergency affected population in a region. However, there have recently been two
region wide nutrition and food security assessments in Darfur, the first in September
2004, and the second in September 200515.

Monitoring and evaluation. If a project has nutritional objectives, nutritional surveys are
often recommended as part of the project, usually at 3 month intervals, to monitor the
wider nutritional situation and progress towards objectives, in other words to monitor
performance. Nutritional surveys are also carried out to assess the coverage of assistance
programmes, in particular feeding programmes. Given the long causal pathways
affecting nutritional status, to monitor impact, anthropometric data needs to be combined
with other information in order to better understand programme impact. In practice,
impact of nutrition programmes in emergencies is rarely systematically evaluated or
documented in an accessible manner.16

AREAS OF CONSENSUS

Amongst the international emergency nutrition community there is broad agreement on
the measurement of malnutrition, nutrition survey methods, the use of a conceptual
framework to assess underlying causes of malnutrition, and the information that is
necessary to interpret nutritional status data. There is also agreement on the indicators
for monitoring the quality and performance of selective feeding programmes. This broad
agreement is reflected in the Sphere handbook and SMART guidelines, which included
consultation with a wide range of actors from the nutrition community, and also the broad
array of good practice guidelines (see Annex 1).

Measurement of malnutrition
There is general agreement about the nutritional indices and cut off points to use to
measure acute malnutrition. The recommended index is weight for height, which is then
as compared to an international reference (NCHS/WHO/CDC) to create the indicator of
weight for height. Cut-off points for determining whether a child is malnourished are

15
   Powerpoint Presentation, Emergency Food Security and Nutrition Assessment, Darfur, September 2005.
Presented by WFP ODAN, November 2005, World Food Programme, Rome
16
   Duffield, A., Reid, G., Walker, D., Shoham, J. (2004, December). Review of published literature for the
impact and cost-effectiveness of six nutrition related emergency interventions. Emergency Nutrition
Network.

                                                                                                         5
either 2 SD below the median of the reference population, or 80% of the median of the
reference population for moderate malnutrition, and 3 SD, or 70% of the median for
severe malnutrition.

The mid upper arm circumference is a useful anthropometric measure, especially for
nutritional screening. Many guidelines recommend the collection of MUAC data along
with weight for height. MUAC is useful for nutritional screening of all children in a
population as it requires little in the way of equipment and training, and screening teams
can cover a large number of children quickly. MUAC has been shown to have a stronger
association with risk of mortality than weight-for-height/length among children between
1 and 5 years of age.17 In the past few years there has been growing pressure for MUAC
to be used as an independent indicator for admission into Outpatient Therapeutic
programs18 and this recommendation has recently been endorsed by an informal WHO
committee.19

Nutrition/anthropometric surveys
Most guidelines recommend the use of two stage 30x30 cluster surveys to for
emergencies. Nutritional status, weight-for-height, is commonly measured on children
between 6 and 59 months (65-110 cm), which are taken to reflect the nutritional status of
the population as a whole. Some guidelines indicate when it may be appropriate to
measure the nutritional status of adults.20

In addition to collection information about weight-for-height, all guidelines recommend
that oedema is collected as another indicator of severe malnutrition, sex and age of the
child, vaccination coverage (at least for measles), and if appropriate, the coverage of
feeding programmes. Most guidelines also recommend the collection of retrospective
mortality data along with nutritional status data, if accurate mortality surveillance
systems do not already exist.

The prevalence of malnutrition is the percentage of the sampled population below the
agreed cut-off points; i.e. below –2 Z-scores for the prevalence of moderate malnutrition
and below -3 Z-scores for severe malnutrition. The value of frequency distribution
curves, and mean nutritional status, for the entire sampled population is highlighted in

17
   - Alam, N., B. Wojtyniak, and M. M. Rahaman, "Anthropometric indicator and risk of death," American
Journal of Clinical Nutrition 49: 884-888 (1989).
- Briend, A., B. Wojtyniak, and M. G. M. Rowland, "Arm circumference and other Factors in children at
high risk of death in rural Bangladesh," Lancet Sept. 26: 725-727 (1987).
- Briend, A. and et al., "Usefulness of nutritional indices and classification in predicting death of
malnourished children," BMJ 293: 373-376 (1986).
- Vella, V. et al., "Anthropometry as a predictor for mortality among Ugandan children, allowing for socio-
economic variables," Eur.J.Clin.Nutr. 48 (1994).
18
   Myatt, M, Khara, T and 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” Draft
background paper for the WHO UNICEF and SCN Informal Global Consultation on Community Based
Management of Severe Malnutrition in children 21-23 November 2005
19
   WHO, UNICEF and SCN Informal Global Consultation on Community Based Management of Severe
Malnutrition in children 21-23 November 2005
20
   E.g. SMART, SC-UK, Oxfam.

                                                                                                          6
many guidelines. Changes in distribution curves, between repeat surveys, show that
food insecurity and famine affects all individuals within a defined population.21

Conceptual framework for analyzing the causes of malnutrition
The conceptual framework for analyzing the causes of malnutrition, as first developed by
UNICEF22 has generally been adopted by international agencies to form the basis of
nutrition assessments in emergencies23. The framework has also been adopted in an
increasing number of national information systems24. The framework gives inadequate
food intake and disease as the immediate causes of malnutrition. There are three clusters
of underlying causes: inadequate household food security, inadequate maternal and child
care, insufficient services and an unhealthy environment. There is a third level of basic
causes which includes formal and informal institutions, political, economic and
ideological superstructure, and the potential resources (See Annex 2).

The framework can be used to develop a local framework on the underlying causes for
the specific emergency, which may form the basis for the nutritional assessment25 . This
in turn leads to the identification of appropriate interventions to address the causes of
malnutrition. ACF, for example, often carries out a causal analysis (using the conceptual
framework) before carrying out a anthropometric survey. In fact, in the Sphere
handbook, one of the indicators for the nutrition assessment and analysis standard states
that “before conducting an anthropometric survey, information on the underlying causes
of malnutrition is analysed and reported, highlighting the nature and severity of the
problem and those groups with greatest nutritional support needs26. The conceptual
framework is also recommended to determine the relative importance of the different
underlying causes in causing malnutrition and mortality, and thereby prioritise
interventions, and for coordination of different sectoral responses.

The interpretation of nutritional information
There is consensus that it is not possible to use nutrition data alone for decision making,
whether it is as part of a nutritional surveillance system, needs assessment or monitoring
and evaluation. Additional information on the underlying causes of malnutrition, and
the risks associated with malnutrition, is necessary.

21
   Young, H. and Jaspars, S. (1995). Nutrition Matters; People, food and famine. IT publications.
Golden, M. and Grellety, Y(2002). Population nutritional status during famine.
22
   UNICEF (1990). Strategy for improved nutrition of children and women in developing countries. A
UNICEF Policy Review. New York, UNICEF.
23
   WHO (2000), The Sphere Project (2004), Save the Children (2004), Nutrition Information in Crisis
Situations (NICS), UN System Standing Committee on Nutrition, WFP (2000). Food and Nutrition
Handbook. Rome, World Food Programme, Oxfam GB (2001, May), MSF (2001). Presentation by Saskia
van der Kam at an inter-agency workshop on Minimum Standards for Disaster Response, Oxford. July 2-3.
Note that the framework has not been adopted by the SMART project, and Mike Golden has proposed an
alternative conceptual framework on causes which gives both disease and malnutrition as direct causes of
death, but a lesser role for disease as a cause of malnutrition.
24
   ENCU led Nutrition Working Group (2002). National Nutrition Survey Guidelines. Addis Ababa,
Disaster Prevention and Preparedness Commission.
Also nutrition guidelines in Afghanistan, Sudan, Malawi
25
   For example in ACF, Oxfam, SC-UK guidelines
26
   Sphere (2004). Humanitarian Charter and Minimum Standards in Disaster Response. P.115.

                                                                                                       7
There is agreement that the following factors need to be considered in the interpretation
of the prevalence of acute malnutrition;

     •   The underlying causes of malnutrition; food insecurity, inadequate care, or poor
         health environment.
     •   Morbidity. Infectious disease can be a cause of malnutrition, and malnutrition can
         increase vulnerability to disease.
     •   Mortality. Whilst most guidelines recommend the collection of mortality data
         (either from secondary information or as part of the nutritional survey), as an
         additional indicator of severity of crisis, few guidelines explain that the
         relationship between malnutrition and mortality depends on the health
         environment or prevailing disease patterns or how the two should be analysed
         together27.
     •   Seasonality. Many rural populations, in particular those who have an annual
         rainy season (associated with harvest, better pasture, reduction in food prices),
         show large fluctuations in the prevalence of malnutrition over a year Any
         prevalence of malnutrition must therefore be interpreted in relation to what can be
         expected at that time of the year.

Sphere recommends that: determining whether levels of malnutrition are acceptable
requires an analysis of the situation in light of the reference population, morbidity and
mortality rates, seasonal fluctuations, pre-emergency levels of malnutrition, and the
underlying causes of malnutrition28. Guidelines reflect that there is broad consensus on
this. Putting this into practice is more difficult, as pre-emergency levels and seasonal
changes in malnutrition are not always available, and as mentioned above, there is little
guidance on how to analyse malnutrition and mortality data together.

Performance of feeding programmes.
The Sphere guidelines provide the standards and indicators for programme quality for
most UN, international organisations and NGOs. They include both process indicators
(e.g. number of staff required in a TFC) and outcome indicators (e.g. % of children
recovered). Amongst the organisations that do not subscribe to Sphere, the MSF
movement in particular, there appears to be little disagreement over the standards and
indicators for assessing programme quality. Rather their reservations about Sphere
concern the interpretation of indicators29 and the use of the standards30. More recently,
reservations have been expressed over the applicability of some of Therapeutic Feeding
Centre (TFC) indicators. In particular, those for weight gain and length of stay are
considered to be inappropriate for community-based approaches to the treatment of acute
27
   The only guidelines which explain how malnutrition and mortality rates should be analysed together are
those of SC-UK. SC-UK used the research carried out by Helen Young as the basis for this. This research
is covered in the section on “the relationship between malnutrition and mortality”.
28
   Sphere (2004). Humanitarian Charter and Minimum Standards in Disaster Response. P. 139.
29
   Griekspoor A and Collins S, ‘Raising standards in emergency relief: how useful are Sphere minimum
standards for humanitarian assistance?,’ BMJ 323 (7315): 740-742 (2001).
30
   Tong J, Questionable Accountability: MSF and Sphere in 2003. Disasters, Vol. 28 Issue 2 Page 176, June
2004.

                                                                                                       8
malnutrition.31 In addition to the Sphere indicators, WFP/UNHCR identify indicators to
trigger alarm i.e. unacceptable levels (see Table 1 below). These are sometimes used as a
tool for internal review and action within programmes, but do not appear to be used often
by donors.

Table 1: Indicators for Monitoring Feeding Programmes (WFP and UNHCR 1999)
SFP Indicators              Alarming               TFP Indicators        Alarming
                            (%)                                          (%)
Recovery rate               10                    Death rate            > 15
Defaulting rate             >30                    Defaulter rate        > 25
                                                   Weight         gain
the data, which in turn depends on who carries out the survey and with what objectives.
The usual practice for many NGOs a nutritional survey covers a small section of the
entire affected population, such as a district or an IDP or refugee camp. A recent trend
initiated by the World Food Programme is to increase survey coverage to include entire
disaster affected populations. For example, recent surveys in 2004 and 2005 carried out
by WFP in Darfur aimed to cover the known crisis affected (and accessible) population in
the entire region32.

NGO survey findings of specific geographical areas or populations cannot be
extrapolated to the wider affected population. Survey results that cover a wide area or
large population may mask pockets of high levels of malnutrition or mortality within the
survey population. Wide variations in nutritional status within the survey population are
reflected in larger confidence intervals and design effects, which suggest that there are
pockets of higher levels of malnutrition within the survey population. However, the data
cannot be disaggregated to provide estimates of malnutrition for sub-samples.
Nutritional surveys are extremely resource intensive, and in non-camp populations can
take up to a month to carry out. The value of the information gained must therefore be
carefully balanced against the time and resources needed to implement surveys.

Pastoralist populations and regions are particularly challenging to sample adequately, as
large parts of the population live in mobile units which cannot be traced easily. In
addition there may be large urban rural differences which are masked in nutrition
surveys. For example in pastoral areas most population data is from the more urbanized
settlements rather than rurally based mobile groups, which therefore may over-represent
the poor and destitute who have had to settle in larger villages and towns. Because
pastoral populations frequently live in small groups, with less than 30 children, they
cannot be satisfactorily sampled using the standard 30 x 30 cluster approach. In such
contexts increasing the number of clusters and decreasing the number of children per
cluster or sentinel site monitoring33 could be considered. A further problem of bias may
be introduced by migration which generates major demographic changes in the
population, the direction of which cannot be assumed and depends on who is migrating
and why.

A further sampling challenge is the compatibility of the cluster survey design with food
security and or livelihood assessment methods. Nutrition survey population estimates
are not necessarily compatible with either the population coverage or the unit of analysis
in food security surveys, which tend to focus on household economy or livelihood groups
which are purposively sampled. This makes drawing direct comparisons difficult.

32
     WFP (2004, October). Emergency Food Security and Nutrition Assessment in Darfur, Sudan.
33
     Young, H. and S. Jaspars (1995). "Nutritional assessments, food security and famine." Disasters 19(1): 26-36.

                                                                                                                     10
Assessment of the underlying causes of malnutrition
There is considerable diversity in methodological approaches for assessing underlying
causes and for inclusion of other indicators of nutritional status. There is long
experience of assessing food security, and more recently livelihoods approaches, yet little
headway in standardizing approaches to the same degree that nutrition surveys have been
standardized. This is in part because most food security assessment approaches are
based on qualitative methods (key informant and focus group interviews, PRA
techniques), for which tried and tested principles for ensuring good practice have been
specified but not adequately incorporated into standardized procedures (e.g. triangulation,
optimal ignorance, iterative analysis, team self-awareness and identity). Common
quantitative indicators of food security include rainfall data, crop assessments, and
market prices. However, statistical correlations of these indicators vary between and
within population, and cannot make conclusions about causation34, which is why many
guidelines recommend a combination of; review of secondary sources and for primary
data collection a qualitative review of underlying causes, combined with quantitative
estimates of acute malnutrition.

Based on the region wide nutrition survey in Darfur, WFP emergency food security
assessment guidelines recommend the use of a dietary diversity as a proxy measure, and
of food insecurity35. This builds on research carried out by IFPRI. Analysis focuses on
three main variables:
    - dietary diversity, defined as a number of unique foods
    - weekly consumption frequency for the selected foods
    - main two sources used by the household to acquire selected foods
This is compared with a reference food consumption indicator to estimate food gaps as a
benchmark for household food insecurity. However, this essentially measures current
food intake rather than food insecurity and says little about the nature of food insecurity
its causes and prognosis. This is particularly important in conflict related or political
emergencies, where the causes of malnutrition and food insecurity are likely to be closely
related to long term process of economic and political marginalization, the direct and
indirect impact of the conflict and violence, and the wider policies and war strategies
adopted by the combatants.

The relationship between malnutrition and mortality36
The interactions between malnutrition and mortality and their underlying causes are
complex, and not well understood. While the quality and analysis of nutritional data has
improved greatly in the past two decades, interpretation has lagged behind. For example
in every food crisis the debates persist about the severity of the situation, for example,
whether it corresponds to a famine or a food crisis. At a political level, famine
undoubtedly evokes a different level of response.

34
   Young, H. and Jaspars, S (1995). Nutrition Matters; People, Food and Famine. IT publications.
35
   WFP (2005, June). Emergency Food Security Handbook. First edition.
36
   This section on the relationshiop between malnutrition and mortality is based on Young (2003)
Nutritional Assessment in Emergencies: progress and remaining challenges. Unpublished paper.

                                                                                                   11
There are three core issues in relation to malnutrition and mortality, which directly affect
the interpretation and subsequent use of data for decision-making.
    • Increases in the rates of acute malnutrition and mortality over time are more likely
        to be exponential than linear. This has implications for the speed with which
        food insecurity progresses to a famine that kills.
    • Rates of malnutrition and mortality do not necessarily increase in parallel, which
        means that malnutrition cannot be used to predict mortality. Situations of high
        malnutrition but low mortality, and vice versa, are qualitatively different and thus
        require different responses.
    • Survivor bias and replacement malnutrition – at what point does excess mortality
        have an impact on rates of malnutrition?

These points are explained more fully below.

    -Is the increase over time between malnutrition and mortality linear or exponential?
A body of influential epidemiological reviews and studies of mortality and malnutrition
in the early nineties demonstrated the strong and critically important association between
malnutrition and mortality among refugee populations. These reviews suggested that the
relationship between acute malnutrition prevalence and crude mortality rates were linear
(ibid), and that ‘mortality rates in refugee groups could be roughly predicted - or
assumed - based on their prevailing malnutrition rates.' 37

However, studies among non-emergency affected populations indicate that the
relationship between mortality and malnutrition is not linear, and that mortality increases
exponentially with declining nutritional status in any population, which is a result of the
synergism between malnutrition and morbidity38. Levels of exposure to disease
obviously change in different contexts, which accounts for varying mortality associated
with a given level of acute malnutrition in different contexts. This has been termed the
potentiating effect of malnutrition on mortality (ibid). Malnutrition and morbidity are
themselves influenced by a range of conditions, including the underlying causes of
malnutrition; food, health and care 39. It is likely that the synergism that occurs between
malnutrition and morbidity also exists between these underlying conditions. This would
mean the combined effects (multiplicative model) of a failure in all three groups of
underlying causes of malnutrition (food, health and care) is far greater than the sum of
their individual effects (additive model), which would account for the exponential
increase in mortality with declining nutritional status in any population. Exposure to
disease varies in different emergency contexts, which can explain in part the varying
mortality associated with a given level of malnutrition. In emergency contexts where
there is displacement and a concomitant deterioration in the public health and care
environment, declining nutritional status is likely to be associated with an exponential

37
   Nieburg, P., B. Person-Karell, et al. (1992). "Malnutrition-mortality relationships among refugees."
          Journal of Refugee Studies 5(3/4): 247-256. p. 251.
38
   Pelletier, D. L., E. A. Frongillo, et al. (1994). "A methodology for estimating the contribution of
malnutrition to child mortality in developing countries." Journal of Nutrition 124(10S): 2106S-2122S
39
   These three groups are taken from the well-known UNICEF conceptual framework of underlying causes
of malnutrition; food security, maternal and childcare, and public health.

                                                                                                     12
increase in mortality. This may explain the strong relationship found between
malnutrition and mortality in refugee contexts.

The multiplicative effect between underlying causes described above may partly account
for the profound difference in malnutrition and mortality rates found in situations of
extreme food insecurity versus situations of outright famine, described by one group of
famine scholars as ‘the difference between freezing water and ice’40. When food
insecurity reaches the stage of destitution, this may prompt distress migration and
subsequent localised public health crises wherever the displaced are forced to settle. At
this point not only has acute malnutrition increased because of the food insecurity, but
exposure to disease has simultaneously increased, thus ratcheting up (multiplying) the
combined impact of malnutrition and morbidity on mortality.

- How can malnutrition and mortality be used diagnose different types of crises?
Confusion arises because malnutrition and mortality do not always increase in parallel,
high levels of malnutrition are not always associated with high levels of mortality, and
vice versa. For example, Figure 1 shows the results of 15 nutrition and mortality
surveys completed in Ethiopia in late 2002, which shows situations of high malnutrition
and low mortality and vice versa. These surveys were vetted by the Emergency Nutrition
Coordination Unit to ensure their rigour and reliability.

Situations of elevated mortality but lower prevalences of malnutrition are relatively easy
to explain, as the mortality is probably caused by factors not related to malnutrition, for
example disease epidemics or “health crises”. An example of high mortality but low
malnutrition is the 1991 refugee crisis in Northern Iraq, where a survey of Kurdish
refugees found a prevalence of acute malnutrition among children under five years of
4.3%, and CMR was 8.9/1000/month (equivalent to 3 per 10,000 per day)41. Two thirds
of the deaths occurred among children aged 5 years or younger, and half among infants
younger than 1 year. Most deaths were due to diarrhoea and dehydration.

Studies during periods of severe food insecurity and famine among more settled or home
based populations have shown no obvious relationship between mortality rates and the
prevalence of malnutrition42. This is most likely due to a more stable public health
environment, with functioning health services, including immunization and stable home
environment i.e. not displaced.

40
   Rivers, J. e. a. (1976). "Lessons for epidemiology from the Ethiopian Famines." Ann Soc Belg Med Trop
         56(4-5): 345-357.
41
   Yip, R. and T. W. Sharp (1993). "Acute malnutrition and high childhood mortality related to diarrhea.
         Lessons from the 1991 Kurdish refugee crisis." JAMA 270(5): 587-90.
42
   Young, H. and S. Jaspars (1995). "Nutrition, disease and death in times of famine." Disasters 19(2): 94
109. Young, H (forthcoming). Nutritional Assessment in Emergencies: progress and remaining
challenges

                                                                                                       13
-        Survivor bias and replacement malnutrition
It is a widely held belief that high mortality in a population can mask deteriorating
nutritional status43. The concept of “replacement malnutrition” and the associated
“survivor bias” has become a widely used explanation in subsequent nutritional survey
reports and refereed academic papers to argue that high mortality is masking a
deteriorating nutritional situation44. However, in an emergency context infant and child
deaths are not limited to the severely or moderately malnourished; deaths occur among
the malnourished and those who are not malnourished. If the data are examined, it has
been found that this only holds true if under five mortality rates are extremely high (in
excess of 15 or 20 per 10,000 per day). It thus seems unlikely that even under conditions
of ‘Famine: Out of Control’ (U5MR>10/10000/day) there would be a significant affect
on prevalence of acute malnutrition.

In conclusion, understanding this relationship between malnutrition and mortality is
important for understanding the exponential progression from acute food insecurity to
famine, for differentiating between qualitatively different types of emergencies and
famines, and for determining programme priorities.

SC-UK has used these research findings in their nutrition assessment guidelines, and have
developed a model showing the different possible combinations of malnutrition and
mortality and the likely causes45. A fuller technical review of data sets with associated
analysis of underlying causes is necessary to see how indicators of malnutrition and
mortality may be used in combination to help characterize qualitatively different types of
famine situation (food crises, health crises, or combinations of both (‘emergencies out of
control). The further inclusion of care and protection factors into such a model would
help refine this model even further.

The use of thresholds and classification systems for nutritional risk
Thresholds for the prevalence of malnutrition are used in decision-making frameworks
for selective feeding programmes, and in systems to classify the severity of food
insecurity. Early decision-making frameworks are response driven because they were
developed by operational humanitarian agencies as a tool to determine the need for

43
   This originates from a paper by Nieburg, Berry et al, 1986, who compared anthropometric data from two cross-
sectional surveys of nutritional status among refugees in eastern Sudan. They found that nutritional status appeared
serious but relatively stable between the two surveys performed over a two month interval. But during this time other
data indicated high childhood mortality in the camp. The authors argued that the deceptive appearance of stability in
nutritional status in the face of high mortality may be explained by ongoing nutritional deterioriation (“replacement
malnutrition”) among surviving children.
44
   Smith, M. C. and S. Zaidi (1993). "Malnutrition in Iraqi children following the Gulf war: results of a national
           survey." Nutrition Reviews 51(3): 74-78.
Salama, P., F. Assefa, et al. (2001). "Malnutrition, measles, mortality, and the humanitarian response during a famine in
           Ethiopia." American Medical Association 286(5): 563-571.
Woodruff, B. A., M. Reynolds, et al. (2002). Summary of Nutrition and Health Survey Badghis Province, Afghanistan.
           February – March 2002
45
     SC-UK (2004). Emergency Nutrition Assessment. Guidelines for Field Workers. p. 193.

                                                                                                                     14
selective feeding programmes46. Since then they were adopted by other INGOs47, and
then incorporated within the UN guidelines48. Only now twenty years on, are there plans
to undertake proper evaluative research into the efficacy of supplementary feeding
programmes in emergencies and the use of such frameworks49. The framework in the
WHO guidelines is given in Table 2.

Table 2: Decision making framework for the implementation of selective feeding
programmes (WHO 2000)

Finding                                                       Action required
Food availability at household level                          Unsatisfactory situation:
 1/10,000/day
          • Epidemic of measles or whooping cough

Decision making frameworks for selective feeding include a range of aggravating factors.
The range of aggravating factors varies between agencies, and includes: general ration
below minimum energy requirements, a crude mortality rate above 1/10,000/day,
epidemic of measles or whooping cough50 , MSF adds severe cold and inadequate shelter

46
   Lusty, T. and P. Diskett (1984). OXFAM's Practical Guide to Selective Feeding Programmes. Oxfam
Practical Guide No 1. Oxford, Oxfam Health Unit, Oxfam.
47
   MSF (1995). Nutrition Guidelines. Paris, France, Medecins Sans Frontieres.
48
   WHO (2000). The Management of Nutrition in Major Emergencies. Geneva, World Health Organization,
United Nations High Commissioner for Refugees, International Federation of Red Cross and Red Crescent
Societies, World Food Programme.
49
   ENN/SC-UK. 2004. Proposal for a multi-agency review of the impact of SFPs
50
   WHO (2002). The management of nutrition in major emergencies.

                                                                                                                      15
to the list, UNHCR/WFP add a high prevalence of respiratory and diarrhoeal disease, and
USAID add ‘Severe public health hazards exist’51. There is discrepancy between
guidelines on the responses required for each level of malnutrition. In the WHO
guideline above and the UNHCR/WFP guidelines for selective feeding programmes in
emergency situations (1999), the highest level of concern (a ‘serious’ situation) is >15%
and this should trigger a blanket distribution to vulnerable groups. By contrast, MSF
guidelines set an additional level requiring blanket supplementary feeding as well as
targeted SFP and TFC where malnutrition prevalence is >20%.52

There are several difficulties with using a decision-making framework like the one shown
in Table 2:
    1. The framework re-enforces the ‘food first’ culture of emergency response. The
       most common humanitarian response strategy has been free food relief even
       though malnutrition can have multiple causes. The food first culture has been
       reinforced by early epidemiological studies of malnutrition in refugee situations,
       where it was listed first among the principal causes of refugee mortality. It also is
       driven in part by donor oriented food aid policies and systems53. This food based
       approach remains the dominant humanitarian response paradigm despite efforts to
       broaden the analysis and response to take account of wider food security, public
       health and livelihood issues. Food is necessary but on its own insufficient to
       support nutrition, and prevent malnutrition.
    2. The use of two or three aggravating factors to interpret the prevalence of
       malnutrition is not consistent with the use of the conceptual framework of
       underlying causes of malnutrition, which in addition to disease and food intake,
       gives underlying and basic causes which contribute to malnutrition. Maternal
       and child care as an underlying cause of malnutrition is not covered at all by such
       decision making frameworks. Factors affecting care on a population level might
       be numbers of unaccompanied children, prevalence of HIV/AIDS, changes in
       infant feeding practices, violence against women etc.
    3. There are large regional differences in levels of acute malnutrition. Large
       differences in the pre-disaster prevalence of malnutrition exist between regions,
       countries, within countries. For example, at the time when thresholds were being
       developed in the mid nineties the national rate of acute malnutrition in
       Bangladesh was 17.8%54, thus the entire country of more than 140 million would
       have been classified as ‘critical’ according to the WHO (2002) classification.
    4. Many populations experience normal seasonal changes in nutritional status.
       Normal seasonal changes can see a drop in the prevalence of malnutrition as large
       as 20% within the space of a three month period.55

51
   USAID 2000. Field Operations Guide for disaster assessment and response. USAID Bureau for
Humanitarian Response, Office of Foreign Disaster Assistance. Version 3.
52
   Boelaert M. et al. 1995. Nutrition Guidelines.
53
  Barrett, C. B. and D. G. Maxwell (2005). Food Aid After Fifty Years. Recasting its role. London and New
York, Routledge.
54
   WHO (1997). WHO Global Database on Child Growth and Malnutrition. Geneva, Programme of
Nutrition, Family and Reproductive Health WHO/ NUT/ 97.4.
55
   Young, H. and Jaspars, S. Nutrition Matters; People, Food and Famine. IT publications. Duffield, A.
and Myatt, M. (2004, March). An analysis of Save the Children UK’s and the Disaster Preparedness and

                                                                                                            16
5. The relationship between malnutrition and mortality is complex, especially
       outside of the context of refugee camps, the severity of the complex emergency
       cannot be judged by such factors alone, particularly in conflict situations where
       protection is a critical factor.

MSF has now adapted the framework to take account of the underlying causes of
malnutrition.56 Others use a similar framework to distinguish different phases or levels of
food insecurity, or have rejected the use of the framework57 . The framework has not
been adopted by Sphere or SMART.

MSF’s new framework gives four stages of food insecurity (food insecurity, food crisis,
serious food crisis, and famine). For each of these phases there are threshold for levels of
malnutrition, mortality, and general information on food availability and access. Health,
food security and nutritional objectives are given for responses, but the focus is on
nutritional interventions. Other presentations of the framework also include caring
behaviours as indicators for the different stages of food insecurity58.

Darcy and Hoffman have proposed a classification of levels and types of food insecurity,
to determine whether a population is suffering chronic food insecurity, acute food crisis,
extended food crisis, or famine. The framework gives a range of responses.59 The
FSAU in Somalia has adopted a similar five phase food security classification, which
uses mortality, access to food, coping strategies, livelihood assets, probability of hazards,
and civil security as indicators in addition to the prevalence of acute malnutrition60. The
thresholds of malnutrition prevalence and mortality rates for famine are generally much
higher than those indicating a “serious situation” in previous frameworks, and ranges
from >25% for acute malnutrition and >2/10,000/day CMR in Darcy and Hoffman’s
paper to 40-50% acute malnutrition and CMR >5/10,000/day in the MSF framework.
The classifications of phases or levels of food insecurity are different for each system.
The differences between the classification systems reflect the serious definitional issues
on what constitutes a famine, food crisis, food insecurity, which were discussed above.
None of the classification systems includes a health crisis, although there are national
efforts to integrate public health concerns into national early warning system.

The UN has a global nutrition information system which classifies emergencies according
to the severity of nutritional risk. The Nutrition Information in Crisis Situations
information system was started in 1993 as the Refugee Nutrition Information System, and
is part of the UN Standing Committee on Nutrition, which is the focal point for
harmonizing nutrition policies in the UN system. The NICS approach classifies

Prevention Commission’s Nutritional Surveillance Programme dataset in some of the most drought prone
areas of Ethiopia, 1995-2001. Draft.
56
   MSF (2004) Nutrition Guidelines. Draft.
57
   SC-UK (2004) Emergency Nutrition Assessment. Guideline for field workers.
58
   MSF (2001). Presentation by Saskia van der Kam at an inter-agency workshop on Minimum Standards
for Disaster Response, Oxford. July 2-3.
59
   Darcy,J and Hoffman, C-A. (2003, September). According to need? Needs assessment and decision
making in the humanitarian sector. HPG report 15. ODI.
60
   FSAU (2005, September). 2005 post Gu analysis.

                                                                                                  17
emergency situations into five categories relating to prevalence of malnutrition and/ or
levels of nutritional risk. Situations may be classified in the absence of prevalence, but
when sufficient information on underlying causes is available to determine risk.

The NICS approach is unique because it is the only system which considers all
underlying causes of malnutrition, key constraints in the delivery of humanitarian
assistance, as well as the prevalence of malnutrition. It is also the only system that
allows for the possibility that malnutrition and mortality rates may not rise in parallel.
NICS uses 5-8% malnutrition as a worrying nutritional situation and 10% as a serious
situation. However, these levels are used with caution recognizing the importance of
contextual and trend analysis. Clearly, the better the analysis and interpretation of the
prevalence of malnutrition in survey reports, the easier it is to assign a nutritional risk
category. See annex 3 for an example of the table that NICS uses to classify
emergencies according to nutritional risk.

Taking account of seasonal or intra-regional differences in the prevalence of malnutrition
is often difficult, as pre-emergency surveys are not always available or comparable to
post emergency surveys. SC-UK in Ethiopia has gone furthest in taking account of
seasonal and inter-district differences in prevalences of malnutrition. They have
established thresholds for malnutrition prevalences for different emergency stages by
season and by district. For each stage, a checklist is given for additional information
needed to determine response61. With interpretation based on comparison with what is
“normal”, there is of course a question of when does normal become unacceptable. There
are an increasing number of populations, in particular in the Horn of Africa, which suffer
high prevalences of malnutrition on a continuous or regular basis.

INSTITUTIONAL ISSUES

The use of nutritional data is constrained not only by the technical challenges but also by
the institutional challenges presented by international humanitarian systems. In the
context of acute malnutrition in emergencies, the term institution refers to the social,
cultural and political structures that govern the collection, analysis, interpretation,
dissemination and use of nutritional data. Institutions have their own self-supporting
logic, laws, principles and technical practices. In nutrition in emergencies this is
reflected by a growing literature, a range of national and UN policies and plethora of
good practice guidelines, which are adhered to by a range of stakeholders and advocates.
Some of the principal institutional challenges to gathering and using nutritional data
effectively are considered below.

61
  Duffield, A. and Myatt, M. (2004, March). An analysis of Save the Children UK’s and the Disaster
Preparedness and Prevention Commission’s Nutritional Surveillance Programme dataset in some of the
most drought prone areas of Ethiopia, 1995-2001. Draft.

                                                                                                     18
Who are the stakeholders and how does the system work?
The range of actors with an interest in acute malnutrition in emergencies, includes a large
number of UN agencies (WHO, UNICEF, WFP, UNHCR, FAO), professional and
technical networks (SCN, Emergency Nutrition Network, Sphere project consultative
groups, NutritionWorks), donor led technical groups (SMART, CDC), national
governments, NGOs (e.g. ACF, SC-UK, Oxfam, MSF, Red Cross Movement) (See
Annex 1). Even within a single agency, there may be different units or teams working on
nutrition related issues, for example in WFP, there is a nutrition team in the policy unit,
but the VAM (vulnerability assessment and mapping) and ODAN (emergency needs
assessment) teams work also links with nutrition.
If there is to be consensus around the use of acute malnutrition as one of a set of
benchmarks, a first step must be to consider what the interests are of these different
stakeholders to analyse and respond to the humanitarian crises that involves acute
malnutrition? What are their respective roles in terms of determining whether or not a
survey is necessary, the processes of data collection & analysis, data coordination &
dissemination and the actual use of information for decision-making about advocacy
strategies, policy making and intervention strategies and programmes.

Furthermore, in some of the most acute crises (for example those associated with
displacement or acute shocks such as an earthquake or floods), it is not necessary to carry
out a nutritional survey to know that significant nutritional risks exist. Response should
start in the absence of nutritional surveys. Nutritional surveys may be useful at a later
stage to assess the severity of the crisis, and to monitor the performance of humanitarian
response. In other situations, it will not be possible to carry out surveys, due to
insecurity or lack of access. In slow onset emergencies, ideally humanitarian responses
should start early, to protect livelihoods, rather than wait until malnutrition and mortality
levels are unacceptable.

Judging the reliability of survey results can be problematic, in particular where the use of
nutritional data is highly politicized and linked with response by the key stakeholders. In
countries such as Ethiopia and Sudan, populations and local institutions are long familiar
with using regular nutritional surveys to assess the need for a humanitarian response. In
addition to some of the technical difficulties associated with surveying dispersed rural or
mobile populations, survey data may be purposefully manipulated to create high
malnutrition levels. Knowing that malnutrition levels are higher in settled destitute
populations, there may be an incentive to bias the survey in favour of these groups.
Recycling of malnourished children amongst households being surveyed is another
example of manipulation. This has been reported in Ethiopia and North Korea. In
situations of conflict, malnutrition may be deliberately created amongst displaced
populations to attract resources which are then diverted by the warring partly in control of
the area. Examples of this have been noted in South Sudan and Somalia.62 The wider use

62
  Scott-Villiers, A., Scott-Villiers, P., Dodge, C. Repatriation of 150,000 Sudanese refugees from Ethiopia.
The manipulation of civilians in a situation of conflict. Disasters 17 (3). 2002.

                                                                                                         19
You can also read