Severe Wasting and Mortality Assessment

In the first stages of an emergency – for example when displaced people or refugees first arrive in a camp – a quick assessment is important of the severity of the situation. Levels or mortality risk and severe malnutrition, especially among children, are two of the conditions of interest. Mortality rates are not so easy to estimate as even with crisis levels (e.g 1/10,000/day) the situation may not be immediately obvious. For malnutrition also, the extent of mild-moderate malnutrition, which carries a considerable risk, needs a practiced eye to assess. However, the presence of severe malnutrition in children is obvious. It is relatively easy to show observers key clinical signs of severe wasting – of which the ‘baggy pants’ wasting of the buttocks is perhaps the most useful.

In a normal situation virtually no severe wasting is seen, even when there are the commonly encountered prevalences of mild-moderate underweight (around 30-50%) or wasting (around 5-10%). This makes observation of severe wasting (<-3SDs wt/ht) a candidate for an ‘existence’ test. This is specially useful where the population numbers are only very roughly known, because prevalences are not calculated – if severe wasting is observed, the hypothesis is, then there is a severe situation. Possibly this should be treated like the 1/10,000/day cut-off in mortality. If it could also be said that the absence of any severe wasting predicts a non-critical situation, this would be useful for planning; and, if a change is seen, for evaluation.

This can be examined from existing data, to get a sense of how good a predictor severe wasting is. Some results were given in Mason (2001, figure 6), which is provided as part of the PANDA emergency module. Some additional results are given below. However, this approach now needs to be field tested before it can be taken much further. A description of how nto field test and devlop the methods is then given later.

Using data from the ACC/SCN Refugee Nutrition Reporting System (RNIS) obtained from surveys in emergencies in Africa during the period 1993-1999, we looked at some basic indicators used by agencies to assess health and nutrition needs in the early stages of emergencies. These include: crude mortality rates, under five mortality rates and levels of (severe) wasting. Through cross tabulation of mortality rates and wasting levels at different, commonly used cut-offs, we tried to establish which cut-off would give the highest sensitivity and specificity for accurately identifying an emergency situation based on these indicators (see table attached at the end).

Results like these were used to draw up the flow chart shown in figure 6 of Mason (2001). Breaking down the severe wasting prevalences into three categories gives us these results.


Severe wasting prevalence	Mortality
>1.0%	Raised mortality certain: >0.3/10,000/day at 98% Crisis mortality levels likely: >1.0/10,000/day at 61%
0.5 – 1.0%	Raised mortality likely: 0.3/10,000/day at 70% Crisis mortality levels uncertain: >1.0/10,000/day at 30%
< 0.5%	Raised mortality uncertain: 0.3/10,000/day at 30% Crisis mortality levels unlikely: >1.0/10,000/day at 20%

This summarizes the probability of mortality rates being at crisis level (>1/10,000/day) or raised above normally expected levels (>0.3/10,000/day), when severe wasting is seen at low levels. The trigger prevalences are deliberately set low to be near a ‘yes/no’ observation – in a population of 1000 people, 1% would only be around 2 cases in under five children.

Based on these findings we describe a methodology for the use of these indicators in rapidly obtaining information that will inform managers of aid agencies on the current situation of the emergency as well as donors on the impact of the interventions. This needs to be tested as it is developed.

The first goal is to assess the validity of the use of the proposed indicators. Specific issues to be addressed are:

- Can an approach using these indicators effectively be used in the rapid assessment and monitoring of health and nutrition needs in the early stages of emergencies?

- Are the indicators identified as possibly appropriate for this approach indeed valid to accurately assess the current situation in an emergency?

- Is the methodology practically applicable and feasible in emergency situations, providing the health and nutrition information needed by NGO’s and required by donors?

Crude mortality rates are considered the most pertinent indicators of an emergency (CDC,1992). In current emergency practices the following indicators and cut-offs as an indication for an emergency are used:

Mild-moderate wasting (<-2SDs wt/ht) in children under five: >10% = alert; >20% = serious.

The aim is to investigate if the existence of even very few severely wasted children can substitute for these, adequately predicting raised mortality, until better information systems are established.

- If we find the above levels of (severe) wasting in children under five, is this indeed reflective of what actually is happening with regard to the severity of nutrition and health problems in the refugee/displaced population?

- Can we determine whether a situation is under control by looking at levels of severe wasting?

- Which other indicator could be used as an additional screen to improve sensitivity and specificity of the methodology (decreasing false negatives and positives)?

-assuming that a two stage cluster sampling needs to be used, a sample of 30 clusters of 30 families is needed to measure mortality (by recall) and nutritional status;

-calculation of crude mortality rates (#/ 10,000/day and, under five mortality rates (#/10,000/day), converted form the number of (age-specific) deaths reported in the defined time period

-calculation of levels of malnutrition in children under five using NCHS/WHO International reference tables (sex specific)

To validate information obtained form the survey, other available sources of information should be sought, on mortality and malnutrition, to compare with survey findings, such as:

-Use of community workers (health / social) responsible for certain part of the site to collect vital information on the population

-Collect age specific mortality and morbidity figures from health and compare observed trends with survey results

3. 3. Comparing of crude mortality rates and levels of malnutrition in children under five

-Establish what levels of malnutrition in children under five are seen with the observed crude mortality rate and age group specific mortality rates

-Establish what levels of age group specific mortality (esp. mortality >5) are coinciding with what levels of malnutrition and crude mortality rates

4. 4. Replication: take steps 1-3 at different sites and at different points in time and compare results

-Establish what levels of malnutrition in children under five are seen with the observed crude mortality rate and age group specific mortality rates at each site/point in time

-Establish whether changing CMR corresponds with changing levels of (severe) wasting and age specific mortality

-Establish which indicators (cut-offs) are most indicative of an emergency brought under control.

Measurement: Take anthropometric measurements, (weight and height) and observe for the presence of edema

Indicator: Calculate after every 1000 new incoming children the % of children < -2SD and < -3SD of the mean in the reference population

Conditions: This method is only appropriate in situations with continued movement of people. It does not provide information about changes in nutritional status within the refugee population that has been established at a site for some time.

Target: Children under five (as an indication of nutritional situation in total population)

Systematic sampling can be used in situations where the location of the population is well organized (well established rows of dwellings, accurate registration), otherwise a two-stage cluster sampling method should be used (see references).

Measurement: Take anthropometric measurements, especially weight and height and observe for presence of edema

Conditions: Although prevalence of malnutrition is a key indicator of the nutritional status of the affected population, the results should be interpreted carefully and within the context of current, specifically recent mortality rates in children under 5. Increased mortality rates may in fact mask a deteriorating nutritional situation. Therefore, information on nutritional status should be crosschecked with mortality data. (changes in mortality data from health services/ mortality data form grave counts / retrospective surveys)

There are a few methods that can be used to estimate mortality rates (CMR/U5MR), which can also be used together to verify data obtained from each method.

All burial site locations are monitored day and night by grave watchers measurement. Each death is recorded by age group through the questioning of the family/friends of the deceased.

Indicator: Number of deaths/10,000/day (converted from number/day/total population estimate)

Conditions: This method can be used in situations where people bury (or incinerate) the bodies of the deceased on a common and known site.

systematic sampling or a two-stage cluster sampling method should be used (see nutritional surveys). A minimum sample size of 3000 people is required to accurately measure crude mortality as well as under five mortality rates. To monitor mortality and measure change over time, larger sample sizes are needed. See reference for sample size calculations for measurement of mortality.

Measurement: Using a short questionnaire, record all deaths -disaggregated by age group- reported by the selected families/households over a well defined period of time (i.e. last two weeks/month)

Indicator: CMR: number of deaths/10,000/day converted from total number of deaths over the period/estimated mid-period population

U5MR: number of under 5 deaths/10,000/day converted from total number of child deaths over the period/estimated mid-period <5 population

Conditions: Standard probability sampling should be used to reduce sampling error. A sufficiently large sample size is needed especially when using surveys for mortality surveillance.

Be aware of tendency to underestimate mortality because of underreporting of deaths and overestimation of population and inaccurate age specific population estimates

Especially in a more established situation, mortality can be monitored through routine population registration where family of deceased report the death of a relative/household member.

Measurement: Recording of each death after reporting by family, disaggregated by age group

Compliance to reporting should be assured and therefore measures to encourage reporting may need to be put in place. Reporting of deaths can also be done by informant responsible for a section of the population (health volunteers / section chiefs / social workers)

Although it does not allow for calculation of mortality rates since deaths are not recorded within the health system in case people die at home, it provides information of changes and trends in mortality in the affected population. Therefore it is useful to monitor health service records in health services and check these against other sources of mortality information to monitor changes in mortality. In addition, the information will help to interpret survey results.

disease specific morbidity (including malnutrition) by age group at each health facility.

The following sources can be used to find more details on assessment methods described above.

Medicins Sans Frontieres, Nutrition Guidelines, Paris (1995) Annex 11 (pg. 143-145)

Reference table for Anthropometric measurements: National Center for Health Statistic(NCHS)/WHO reference table

Medecins Sans Frontieres, Nutritional Guidelines, Paris (1995) “Sampling methods” pg. 49-59

Diarrhoeal Disease Control Programme, WHO/Geneva 1989, Household Survey Manual: Diarrhoea Case Management, Morbidity and Mortality

Mason, JB. (2002). Lessons on the Nutrition of Displaced People. J Nutr. 132: 2096S-2103S. (Also in the Emergency module of the PANDA)

Cathrien Alons and John Mason (2000) Rapid Assessment and Monitoring in the Early Stages of Emergencies. Report to CARE.

[1] Supported by CARE contract # 0010000670-1

[2] This paper is also in the Emergency module of the CD ‘PANDA’. The material links to the work in ‘Lessons on the Nutrition of Refugees’ (Mason, 2002), and ‘Rapid Assessment in the Early Stages of Emergencies’, by Alons and Mason, 2000.

Severe Wasting and Mortality Assessment:

An Approach to Rapid Assessment of Mortality Risk in Emergencies[1] [2]