3. Goals of a Survey
Evaluate the scope and severity of a
humanitarian crisis.
Determine the needs of a new
program.
Evaluate already existing programs.
Advocacy.
4. When to do a Survey?
4
Surveys should take place when the
data collected will answer questions
that will influence activities and
determine actions.
5. 5
Snapshot Question
What are the differences between
nutritional surveys, nutrition
surveillance and rapid assessment?
6. Methods for Data Collection
6
Rapid
assessment
Nutrition
Survey
Nutrition
Surveillance
Rapid appraisal
Medium-term
appraisal
Continuous
appraisal
Qualitative/
Quantitative/
Cross sectional
snapshot
Cross sectional
snapshot
Quantitative/
Longitudinal
trends
Observational /
Sample with
survey
instruments
Periodic,
standardized
data collection
Objective
Data Type
Method
Secondary source
7. Limitations of Nutrition Surveys
7
Cross-sectional in nature.
Provides a snapshot at a given moment of
time (not trends).
Unable to establish causality.
Provides insufficient information for causeeffect analysis.
Cross sectional survey data should be
used in conjunction with other contextual
information
12. Need for:
12
Standardized methodology - timely and reliable
data.
Simplified, clear guidance for non-epidemiologists/
statisticians to prioritize humanitarian assistance.
Addressing key issues regularly faced by field
workers.
Emphasis on representative sampling AND
information on the quality of the measurements.
13. Problem: No Standardisation
13
Data quality in disaster settings poor.
Ethiopia 1999-2000:
Source: Spiegel P et al, JAMA 2004
Among 67 nutrition and mortality
surveys, only 6 (9%) met all 5 eligibility
criteria (being valid and precise).
14. History
14
The SMART methodology was developed in
2006 by a panel of experts in epidemiology,
nutrition, food security, early warning systems
and demography.
SMART was originally developed to assess
acute malnutrition and mortality in
emergencies.
It is now used in all settings, including
development and displaced populations.
15. Who uses SMART?
15
Today, SMART is recognised as the standard
methodology by national Ministries of Health,
donors, implementing partners such as
international NGOs and UN agencies that wish to
undertake nutrition and mortality surveys in
ALL settings (emergency, development,
displaced populations).
SMART is also incorporated into many national
nutrition protocols.
16. Surveys using SMART
16
Produce representative, accurate and precise estimates of:
Global Acute Malnutrition (GAM).
Chronic malnutrition (Stunting).
Underweight.
Retrospective mortality.
These four indicators gathered through the SMART
methodology provide the best available validated data that
can be used for effective decision making and resource
allocation.
19. Why is mortality data needed?
19
Benchmark the severity of a crisis.
Establish a baseline for comparison (programmatic
purposes).
Provide an empirical basis for advocacy purposes.
Complement a surveillance system.
Provides information on the demography of the
target population.
20. What indicators to include
20
Main indicators:
1.Crude Death Rate.
2.Under-5 Death Rate.
Other possible analyses:
Age-specific death rate.
Sex-specific death rate.
(Cause-specific death rate).
22. Limitations of Mortality Data
22
Data is retrospective: data on deaths will be late;
deaths may have occurred months ago.
Mortality is multi-causal: it is difficult to identify
what specific programs have reduced mortality
rate.
22
23. SMART can also evaluate:
23
Additional indicators, including:
Anemia; water, sanitation and hygiene (WASH);
coverage of blanket programs, immunization, and
vitamin A distribution; food security contextual
information.
The number of additional indicators is
recommended to be kept to a minimum to
ensure high quality data.
24. Key SMART Innovations
1.
2.
3.
4.
5.
6.
User-friendly software -ENA Software- that follows every
step of a SMART survey.
Flexibility in sample size calculation.
Regularly updated, clear sampling guidance based on field
experiences, research and best practices.
Rigorous standardization test and analysis.
Plausibility check to follow data quality and identify
where the problems are.
Improved census procedure for the Mortality
component.
25. Increased data credibility
25
Rigorous standardisation of field procedures.
Data quality checks.
Standardised automated data analysis.
Consistent and reliable survey data is
collected and analysed.
26. ACF-CA: SMART Project Convenor
26
ACF-CA, a core member of the GNC, is the SMART Project
Convenor, and in collaboration with the SMART Technical
Advisory Group and Centers for Disease Control and Prevention
(CDC Atlanta) establishes and maintains:
Standardisation and translation of tools for survey managers.
The SMART www.smartmethodology.org website.
On-line technical forum.
Partnerships with other agencies in trainings & survey support.
The SMART experts at ACF-Canada
in collaboration with CDC
currently provide training and remote technical
support in SMART.
Notes de l'éditeur
Notes for trainers:
There are various definitions of a survey. In the Emergency Nutrition Assessment (2004), Save the Children defines a survey as “a method of gathering information about a large number of people by talking to, or measuring only, a sample of them; a way to collect information on people’s needs, behaviour, attitudes, environment and opinions, as well as on personal characteristics such as age, nutrition status, income and occupation.”
Notes for trainers:
If you are using a survey to evaluate a program impact, you should pay close attention to the confounding factors that might have played a role in the improvement or deterioration of the situation where the intervention took place.
A nutritional survey, for example, can partially measure the resulting change in the nutritional status of the population. We can only evaluate partially the efficiency and effectiveness of these interventions, since we must also take into consideration any other factor that could have influenced the nutritional status.
The Sphere Minimum Standard requires that an assessment of nutritional status is conducted when a targeted nutrition program is implemented (Sphere Project, 2004), but this does not mean that in an emergency situation, the program is delayed until a survey is completed.
Finally, a survey, for example, also serves as an advocacy tool (addressed to governments or other funding agencies, to be able to lobby for support, raise awareness of a situation.)
Notes for trainers:
Are health and nutrition surveys relevant to “rapid assessments” ?
Answer: Yes – while there is usually no time to implement an actual survey (primary data collection), rapid assessments will often require reviewing secondary data or survey reports.
Furthermore, a rapid assessment will assist in deciding whether or not a more in-depth survey is required as well as the type, objectives and geographical coverage of the survey.
Looking at the table:
A survey is usually implemented for decision-making in the short to medium term. Given the fact that a survey requires significant planning and resources, it usually requires several weeks for planning and implementation.
Quantitative data is usually collected which provides a cross sectional picture of the nutritional status of the population at a single point in time.
Data is collected at the household level and individuals within that household.
A representative sample is taken using a survey instrument or tool.
A survey is more valid than a rapid assessment, especially when the findings are compared with a previous survey.
Many useful manuals and guidelines, but Many humanitarian agencies suffered from a critical lack of specialized manpower in these areas.
Key issues: How to handle children with disabilities; what to do if a household as no children; what if the child was at school or in the clinic.
Problems in Ethiopia included:
1. Non-probabilistic sampling, 2. Sample Size < 500 children aged 6-59 months, 3. < 25 clusters, 4. Weight for Height not measured/poorly measured as anthropometric index.
The listed variable are the absolute minimum that are required to determine the prevalence of protein-energy malnutrition. However, these alone will not provide all the potentially useful information.
Your rapid assessment, secondary information source, and the objectives of the survey are important to consider when deciding what other information you wish to collect in the survey.
This is an example for anthropometry. The same kind of work needs to be done whatever the indicators assessed (e.g., death rates).
Advocacy: Death rates are the most compelling for advocacy purposes. It is powerful to use for decision makers and policy makers.
Benchmark: It is difficult to attribute change in mortality to any one intervention but as a general baseline it is very useful information.
Surveillance: Mortality is best measured through surveillance (real time) to track weekly trends. In stable countries we refer to mortality surveillance as “vital statistics” (i.e., counting deaths from death certificates). Surveys are less valuable because they are retrospective. But in some contexts we do not have mortality surveillance (vital statistics). Where we don’t have surveillance, surveys may be the only option for gathering this information.
----
An elevated death rate can indicate that there is a health problem in a population BUT cannot determine its cause.
Usually, public health workers at the start of an emergency have to rely on cross-sectional surveys to determine current nutrition status and death rates in the recent past. Ideally, surveys complement a functioning surveillance system to estimate acute malnutrition, verify surveillance data, and answer specific questions that the surveillance system is not providing or about areas that the surveillance system is not covering.
The crude death rate and the under-five mortality rate are the most frequently used measures used in emergency contexts.
Note: It we often do not get enough sample size to have strong confidence in the U5DR.
Age specific: U5 or any other age group.
Multi-causal: could be anything from outbreak of disease to violent death. Cannot well attribute mortality to any one cause. Can’t be used well to
measure affect of mortality trends to any specific program.
The members of a survey team must be able to answer the community leader’s and the family’s questions, and provide them with necessary explanations. They must understand the reasons for the mortality survey.
Identify the most vulnerable groups.
A death rate can point out a health problem but not establish a cause-and-effect relationship. Moreover, investigating causes/ location of death poses numerous methodological challenges very difficult to deal with in emergency and transition situations. Thus, SMART does not recommend to collect these variables when it is not objectively necessary.
Notes to trainers:
The basis of SMART methodology relies on many manuals and guidelines, as that of Save the Children Fund: Emergency nutrition Assessment: Guidelines for field workers (2004).
Some elements come from organizations such as MSF, FANTA, ACF, WHO, but also epidemiological and statistical norms.
With SMART, there is a manual that comes with the software.
Simplifies data collection and completion of reports from specific data sets.
SMART is added to established and approved methods.
Other software exists such as EpiInfo, EpiNut, Anthro, Excel, SPSS, EPI/ENA, etc.