Understanding malnutrition in emergencies

1The forms of malnutrition

Malnutrition covers several realities. Acute malnutrition (wasting) reflects rapid weight loss, often linked to illness or a sudden lack of food: it is what kills fastest in emergencies. Chronic malnutrition (stunting) reflects long-term deprivation: the child is too short for their age. Micronutrient deficiencies (iron, vitamin A, iodine, zinc) weaken immunity and development.

Acute malnutrition is divided into moderate acute malnutrition (MAM) and severe acute malnutrition (SAM). A child with SAM has a risk of death about nine times higher than a well-nourished child. Global acute malnutrition (GAM) adds MAM and SAM together in a population.

2Causes: the UNICEF conceptual framework

The UNICEF framework distinguishes three levels of causes:

• Immediate causes: inadequate dietary intake and disease (diarrhoea, measles, malaria, respiratory infections), which reinforce each other.
• Underlying causes: household food insecurity, child care and feeding practices, access to water, sanitation and health services.
• Basic causes: poverty, conflict, displacement, gender inequality, governance.

Practical consequence: an effective nutrition response is multisectoral. Treating without acting on water, health and access to food means seeing the children come back.

3Measuring nutritional status

Three measurements are used for screening and diagnosis in children aged 6 to 59 months:

• Mid-upper arm circumference (MUAC), measured at the midpoint of the left upper arm with a colour-coded tape: below 115 mm = SAM; 115 to below 125 mm = MAM.
• Weight-for-height expressed as a z-score against WHO standards: below -3 = SAM; -3 to below -2 = MAM.
• Bilateral pitting oedema: press for three seconds on the top of both feet; if a dent remains, it is SAM, whatever the MUAC.

MUAC is simple, quick and can be used by community health workers and even by trained mothers (the “Family MUAC” approach). In pregnant or breastfeeding women, a low MUAC (often below 230 mm, depending on the national protocol) signals a risk.

4Thresholds and classification of the situation

Anthropometric surveys give the prevalence of GAM among children aged 6-59 months. Commonly used benchmarks (WHO): below 5% = acceptable; 5 to 9.9% = alert; 10 to 14.9% = serious; 15% or more = critical. The IPC Acute Malnutrition classification (IPC AMN) places each area from Phase 1 to Phase 5 and guides priorities.

Other indicators complete the analysis: crude mortality rate and under-five mortality, measles vaccination coverage, ongoing epidemics, breastfeeding practices. Always apply the country's national protocol: it takes precedence over general benchmarks.

5Field anthropometry: measuring accurately, not just measuring

The entire chain of nutrition decision-making rests on a handful of figures: a mid-upper arm circumference, a weight, a height, an age, and the presence or absence of oedema. An error of 5 mm on MUAC or 1 cm on height is enough to shift a child from one category to another, and a systematic error by a whole team distorts the prevalence of an entire survey, and therefore the classification of an area and the volume of funding. Measurement quality is not a technical detail: it is the first condition of accountability.

Minimum equipment and how to check it

The technique step by step

1. Prepare the child and the mother: explain what is going to be done, remove heavy clothing, shoes and jewellery. A child who is crying and struggling gives poor measurements: take your time.
2. MUAC: left arm relaxed along the body. Locate the midpoint between the acromion (tip of the shoulder) and the olecranon (tip of the elbow), with the elbow bent at 90 degrees, and mark this point. With the arm relaxed again, place the tape at this level, neither too tight (the skin must not wrinkle) nor too loose. Read to the nearest millimetre and call out the figure; the assistant repeats it and writes it down.
3. Weight: tare the scale with the mother alone, then weigh the mother holding the child; or use weighing pants with the hanging scale at the measurer's eye level. Read to the nearest 100 g, with the child still.
4. Height: two people. In the lying position, the assistant holds the head against the fixed headboard, gaze vertical (Frankfurt plane), while the measurer holds the knees and brings the footpiece against the soles of the feet. Read to the nearest millimetre. A standing height taken on a child who should have been measured lying down gives a result about 0.7 cm shorter: the WHO standards provide for this correction, but the actual position must be recorded.
5. Oedema: firm thumb pressure for three seconds on the top of both feet. If the pit persists on both sides, oedema is confirmed. It is then graded: + feet only; ++ feet and legs, or hands and forearms; +++ generalised oedema including the face. Grade +++ and the combination of oedema with wasting (often called marasmic kwashiorkor) always require inpatient care.
6. Immediate recording: never memorise several measurements. In tablet-based surveys, the form flags outliers in real time.

Standardisation of measurers

Before any survey, and ideally every year for programme teams, measurers take a standardisation test using the SMART method: each measurer measures a group of about ten children twice, at an interval, without seeing their first values. For each measurer, the technical error of measurement (precision: difference between their two measurements) and the bias (accuracy: difference from the supervisor's mean or that of the whole group) are calculated. The ENA for SMART software provides a results table with classification thresholds (good, acceptable, poor, reject). A measurer who fails the test is retrained, then reassigned to another role if they fail again.

Rule of thumb: measurement quality is not decided on the day of data collection, but on the day of practical training and standardisation. Surveys whose report does not present the results of the standardisation test are regularly rejected during validation by the technical working group.

Reading data quality after the fact

The ENA for SMART plausibility report automatically checks several criteria. You need to know how to read them, because an evaluator or a cluster coordinator will read them for you:

• Outliers ("flags"): weight-for-height z-scores more than 3 units from the observed sample mean are excluded from the analysis according to SMART criteria. More than 2.5% of excluded values signals a problem.
• Standard deviation of weight-for-height z-scores: expected to be between about 0.8 and 1.2. A wider standard deviation generally indicates measurement or age errors, and artificially inflates prevalence.
• Digit preference: if weights too often end in .0 or .5, or heights in .0, measurers are rounding. The digit preference score makes it possible to identify the team concerned.
• Sex ratio and age distribution: a shortfall of children aged 30 to 59 months suggests that older children were excluded or that age was poorly estimated.
• Overall plausibility score: a weighted summary of the criteria. A low score is good; beyond the "problematic" threshold, the survey cannot be used as is for classification.

The special case of age

In many crisis contexts, fewer than half of children have a document showing their date of birth. The local calendar of events (religious holidays, rainy seasons, harvests, elections, floods, epidemics, arrival at the site) is built with key informants, month by month over five years, then tested. Age is used to check eligibility (6 to 59 months) and to calculate the height-for-age and weight-for-age indices; weight-for-height, on the other hand, does not depend on age, which makes it robust for measuring acute malnutrition. In the absence of any reference point, a height of 65 to 110 cm is often used as a proxy inclusion criterion.

6Conducting a causal analysis of malnutrition in an area

The UNICEF conceptual framework (in its updated 2020 version, which emphasises the enabling determinants of good nutrition and not only the causes of malnutrition) shows where to look, but does not say what matters most in a given area. Yet a programme that treats malnutrition without knowing its dominant local causes will treat the same children year after year. Causal analysis answers a simple question: in this context, which factors best explain acute malnutrition, and which of them can be modified?

Available methods

Steps of a rapid causal analysis

1. Formulate the question: population, area, period. For example, "why has GAM among children aged 6 to 59 months in the district remained above 12% for four years, with a peak in August?"
2. Secondary data review: SMART surveys from the last five years, monthly programme admission data (which show seasonality), food security surveys (food consumption score, coping strategies index), health data (malaria, diarrhoea, measles), water and sanitation data, demographic and health surveys, anthropological studies on care practices.
3. Build hypotheses: a list of possible risk factors, classified by level (immediate, underlying, basic) and by sector.
4. Targeted complementary data collection: separate focus group discussions (mothers, fathers, grandmothers, traditional birth attendants, leaders), interviews with health staff, observation of feeding and hygiene practices. No new heavy survey if the data already exist.
5. Weighting workshop: with the technical services, the cluster and community representatives, each hypothesis is scored according to the strength of evidence and perceived importance. The result is a hierarchy: major, important, minor or unconfirmed factors.
6. Feedback and recommendations: by sector, with concrete levers, and validation by the health authorities.

Grid of factors to examine

A useful causal analysis ends with a defensible prioritisation statement, for example: "In the northern communes, peaks of acute malnutrition follow peaks of malaria and diarrhoea by four to six weeks, and coincide with the period when mothers go to work in the fields; food insecurity plays an aggravating but secondary role." It is this statement that should guide the allocation of the budget between treatment, health, WASH and food security.

Seasonality, an under-used analytical tool

Plotting on the same graph, over three to five years, monthly OTP admissions, malaria and diarrhoea cases, cereal prices and the agricultural calendar almost always reveals a pattern. In Sahelian areas, the admissions peak typically falls between July and October, during the lean season and the rainy season. Surveys carried out after the harvest therefore underestimate the annual caseload: this is one of the reasons why needs calculations use an incidence factor and not point prevalence alone.

7Full example: reading and interpreting survey results in Wadi-Sora district

Fictional context: Wadi-Sora district (about 186,000 inhabitants) has, over the past eight months, received nearly 40,000 people displaced by armed conflict, spread across two sites and host families. A SMART survey was conducted in September by a national NGO, "Santé pour Tous Wadi", with technical support from the cluster. You are the nutritionist responsible for drafting the interpretation for the technical working group meeting.

Step 1: check quality before reading the results

Conclusion of this step: the data are usable. The prevalence figures can now be read.

Step 2: anthropometric results

Interpretation. GAM of 14.3% falls within the "serious" category of the commonly used WHO benchmarks (10 to 14.9%), and within the "high" category of the prevalence classification for wasting published by WHO and UNICEF in 2018 (less than 2.5% very low, 2.5 to less than 5% low, 5 to less than 10% medium, 10 to less than 15% high, 15% or more very high), just below the threshold of the "very high" category. But the confidence interval goes up to 17.5%: a critical situation cannot be ruled out. An honest interpretation states this explicitly. The difference between MUAC-based and weight-for-height-based prevalence is common: the two indicators do not identify the same children (MUAC selects more younger children and girls, weight-for-height more children with a slender build). They are always presented separately, without adding them together or substituting one for the other.

Step 3: mortality

The retrospective mortality module covers a recall period of 92 days. Calculation of the crude death rate (CDR): deaths / (mean population over the period × recall days) × 10,000.

Interpretation. The Sphere Handbook recommends comparing mortality with the baseline for the country or region: a doubling of the baseline rate signals an emergency. When the baseline rate is unknown, thresholds of 1 death per 10,000 people per day (CDR) and 2 per 10,000 children under 5 per day are used. Here, the CDR is close to the threshold and under-5 mortality reaches it. The survey also indicates that families attribute 9 of the 15 child deaths to "fever" and "diarrhoea": the health track is the priority.

Step 4: aggravating factors

• Measles vaccination coverage (9-59 months, card or recall): 61%, far from the 95% targeted by Sphere for children aged 6 months to 15 years in crisis settings.
• Diarrhoea in the previous two weeks: 31% of children.
• Water: 9 litres per person per day at the main site, compared with the Sphere benchmark of 15 litres.
• "Poor" food consumption score: 27% of displaced households, 11% of host households.
• Exclusive breastfeeding of infants under 6 months: 41% (small sample, to be interpreted with caution).

Step 5: the conclusion drafted for the technical working group

The nutrition situation in Wadi-Sora is serious, with a real risk of tipping into a critical situation (GAM 14.3%, CI 11.6 - 17.5%), and is accompanied by under-5 mortality at the emergency threshold. The dominant aggravating factors are health-related (diarrhoea, fever, low measles coverage) and water-related, against a background of food insecurity among displaced households. Recommendations: immediate scale-up of screening and treatment of SAM and MAM, measles vaccination campaign with vitamin A, urgent increase in water supply, targeted rations or transfers for displaced households, monthly monitoring of admissions and a new survey in six months, before the seasonal peak.

Step 6: what should not have been written

• "GAM is 14.3%, so the situation is not critical": this ignores the confidence interval and mortality.
• "Combined GAM (MUAC + weight-for-height) is 17%": without a definition validated by the national protocol and the technical working group, a combined prevalence cannot be compared with the standard thresholds.
• "Chronic malnutrition of 38.6% proves the emergency": stunting reflects long-term deprivation; it is serious, but it does not measure acute emergency and is not treated with RUTF.

8Beyond 6-59 months: infants, women, older people and micronutrient deficiencies

Surveys and programmes have historically focused on children aged 6 to 59 months, because thresholds, tools and products were designed for them. But a complete nutrition analysis in an emergency must also look at the groups that remain invisible in standard figures.

Infants under 6 months

MUAC is not routinely validated below 6 months and RUTF is not suitable for infants. Yet this group has high mortality in crises, particularly when breastfeeding stops. The 2023 WHO guideline on the prevention and management of wasting and nutritional oedema (acute malnutrition) in infants and children under 5 years broadens the notion of an "at-risk" infant: low weight-for-age, weight loss, breastfeeding difficulties, illness, sick or absent mother. The approach known as MAMI (Management of small and nutritionally At-risk Mothers and Infants), promoted in particular by ENN and LSHTM, offers a care pathway centred on the mother-infant pair.

• To assess: breastfeeding capacity (attachment, suckling, frequency), weight and weight trend, clinical danger signs, physical and mental health of the mother.
• Referral: inpatient care in the presence of a danger sign or significant weight loss; otherwise, outpatient breastfeeding support with close weight monitoring.
• What not to do: apply the 6-59 month criteria, give RUTF, distribute infant formula without individual assessment.

Pregnant and lactating women

There is no single international MUAC threshold for defining acute malnutrition in pregnant or lactating women. National protocols most often use a MUAC below 230 mm, sometimes 210 or 220 mm depending on the country and population. MUAC is preferred to weight during pregnancy because it varies little with pregnancy. Maternal malnutrition has direct consequences for low birth weight, and therefore for neonatal mortality and infant malnutrition. In an emergency, the analysis includes at a minimum the proportion of pregnant and lactating women with a low MUAC, antenatal care coverage and coverage of iron and folic acid supplementation.

Adolescents, adults and older people

• Adults: body mass index (BMI), with commonly used thresholds of less than 16 for severe malnutrition and 16 to less than 17 for moderate malnutrition; MUAC is often used for screening, with thresholds varying between protocols (for example less than 185 mm for a severe form in non-pregnant adults).
• Adolescents: BMI-for-age z-score according to the WHO references.
• Older people: few validated thresholds; assessment is clinical and functional (recent weight loss, mobility, ability to prepare and eat rations, teeth). Dry rations distributed without adaptation are often unusable for them.
• People living with HIV or tuberculosis: increased energy requirements, specific national protocols, close linkage with treatment programmes.

Micronutrient deficiencies in crisis settings

In populations that depend on a limited food ration over long periods (refugee camps in particular), deficiency diseases thought to have disappeared re-emerge. The documented outbreaks of scurvy, pellagra and beriberi in camps in East Africa and Asia bear witness to this.

For anaemia, WHO classifies public health significance according to prevalence: less than 5% no public health problem, 5 to 19.9% mild, 20 to 39.9% moderate, 40% or more severe. Sphere notes that a high prevalence of anaemia among children or women is a warning signal that should trigger a multisectoral response.

An analysis report that only discusses GAM among children aged 6-59 months is incomplete. Technical working group reviewers and donors expect at a minimum a mention of infants under 6 months, pregnant and lactating women, and the risks of deficiencies linked to the composition of the ration.

9Common mistakes, contextual variations and what donors look at

The most common mistakes in the field

Variations by context

• Sudden-onset crisis (mass displacement, flood): there is no time for a SMART survey in the first weeks. Exhaustive or rapid MUAC screening on arrival or at registration, health facility data and initial rapid multisectoral assessments are used. The situation is described as "probable" and a survey is planned as soon as access and stability allow, generally within two to three months.
• Protracted crisis: annual or biannual surveys, nutrition surveillance systems (sentinel sites), in-depth causal analyses. The main risk is the normalisation of high prevalence: GAM of 12% for ten years remains a serious situation.
• Camp: known and accessible population, simpler surveys (systematic sampling possible if a map of shelters exists), realistic exhaustive screening. For refugee camps, UNHCR uses its SENS surveys (Standardised Expanded Nutrition Surveys), which integrate anaemia, IYCF, coverage and food security.
• Urban area: dispersed, mobile populations mixed with residents; household lists rarely available. Urban displaced people are often under-represented in surveys. Spatial sampling methods or neighbourhood-targeted surveys are used.
• Limited access and remote management: surveys conducted by local teams trained remotely, quality control reinforced by tablets with GPS coordinates, time-stamping and daily data upload; daily review of the plausibility report by the remote supervisor.
• Insecurity: inaccessible clusters must be documented and replaced according to a rule defined in advance, and the report must specify which population is not represented. A survey that silently excludes the worst-affected areas underestimates severity.
• Local partner with low capacity: accompany rather than substitute. Training, standardisation and analysis carried out together, with technical support from the cluster or an international partner, and joint signature of the report.

What donors, evaluators and technical working groups look at

• The source and date of the prevalence data cited in a proposal, and their validation by the technical working group or the cluster. ECHO and BHA explicitly require the needs analysis to be based on recent and recognised data.
• Consistency between the needs analysis, the country's humanitarian needs overview and the IPC classification.
• The existence of a causal analysis and how it justifies the proposed combination of activities.
• Consideration of vulnerable groups beyond children aged 6-59 months and disaggregation of data by sex and age, required by most donors.
• Compliance with the national protocol and with the thresholds adopted by the country.
• Data protection: no named lists of children in annexes, no identifiable photos.

10Interview questions and key references

Interviews for nutritionist, nutrition programme manager or cluster information management officer positions almost always include a technical section on measurement and analysis. Here are frequent questions and the elements of a strong answer.

1. "A survey gives GAM of 9% by MUAC and 15% by weight-for-height in the same population. How do you explain this, and which one do you use?"

A good answer explains that the two indicators identify partly different groups of children (age, sex, body shape), that neither is "wrong", that the classification of the situation uses weight-for-height prevalence according to the reference thresholds, while programmes admit on both criteria or on MUAC alone if the national protocol provides for it. It mentions the importance of first checking data quality (standard deviation, excluded values) before interpreting the gap.

2. "How do you ensure the quality of an anthropometric survey?"

The candidate mentions the sample size calculation, several days of practical training, the standardisation test, the calendar of events, daily equipment checks, supervision with re-measurement, daily review of the plausibility report and final validation by the technical working group. They give precise thresholds (standard deviation, percentage of excluded values).

3. "Calculate the under-5 mortality rate: 1,200 children, 18 deaths, 90-day recall period. What do you conclude?"

Expected answer: 18 / (1,200 × 90) × 10,000 = 1.67 deaths per 10,000 per day. This rate is below the threshold of 2 used when the baseline is unknown, but it must be compared with the country's baseline rate: if it is double that rate, the situation is an emergency. A good candidate also points out the width of the confidence interval with a small number of deaths.

4. "Why carry out a causal analysis before designing a programme? Give an example."

The candidate shows that they have seen an analysis change a strategy: for example, the finding that malnutrition peaks followed malaria peaks led to the integration of bed net distribution and community case management of malaria into the nutrition programme. They cite a method (rapid workshop, Link NCA) and the notion of modifiable factors.

5. "You suspect a scurvy outbreak in a camp. What do you do?"

Clinical confirmation by a doctor, case definition, active case finding, notification to the ministry and to the Health and Nutrition clusters, analysis of the ration composition with WFP and UNHCR, treatment of cases, addition of foods rich in vitamin C or fortified foods, monitoring of the trend.

6. "Your director wants to publish a GAM of 16% from a mass screening to raise funds. What do you say?"

The answer explains that a screening is not a representative survey, that it cannot be compared with classification thresholds, that publication without validation by the technical working group would damage the organisation's credibility with the cluster and donors, and proposes using the data as a warning signal to request a survey or an IPC analysis.

Key references

• The Sphere Handbook (2018), Food Security and Nutrition chapter, in particular the standards on food security and nutrition assessment: Sphere Association.
• WHO Child Growth Standards (2006) and the prevalence thresholds for malnutrition published by WHO and UNICEF (2018).
• WHO Guideline on the prevention and management of wasting and nutritional oedema (acute malnutrition) in infants and children under 5 years (2023).
• SMART Methodology Manual and ENA for SMART software: SMART Initiative, Action Against Hunger Canada.
• IPC Technical Manual (current version), acute malnutrition component: IPC Global Partnership.
• Conceptual Framework on the Determinants of Maternal and Child Nutrition (2020): UNICEF.
• Link NCA method guide: Action Against Hunger.
• SENS Guidelines: UNHCR.
• MAMI tools and resources: Emergency Nutrition Network (ENN) and London School of Hygiene and Tropical Medicine.
• WHO reference documents on the assessment of anaemia and micronutrient deficiencies, and documents on scurvy, pellagra and beriberi in emergencies (WHO, UNHCR).