For Recommended Survey Equipment, please see Resources.
1.1 What are the different elements of the SMART methodology?
The SMART methodology includes three main elements:
3) Food Security.
Guidelines and recommendations to conduct a survey using the SMART methodology are provided for each of these three elements.
1.2 Why measure nutritional status of children 6-59 months?
Acute malnutrition in children 6-59 months is closely linked with risk of death and is used to draw conclusions about the situation of the health status of the whole population, not just young children. Children aged 6-59 months are more vulnerable than other age groups to external factors (such as food shortage or illness) and their nutrition status is more sensitive to change than that of adults in many (although not all) populations.
1.3 Why measure mortality?
Mortality is the most critical indicator of a population’s improving or deteriorating health status and is the type of information to which donors and relief agencies most readily respond.
1.4 Why measure food security?
Food security information helps in understanding and interpreting nutrition and mortality survey data. SMART uses the Household Economy Approach, which measures various social, economic and medical factors and their impact on livelihoods, in order to evaluate food security
SMART’s input in terms of food security context analysis is to facilitate and standardize the gathering and analysis of information needed to predict how and at what point a shock may or may not impact a population, especially its nutritional status. The SMART context analysis is “a cut-down version of the Household Economy Approach (HEA) designed for local use”. The HEA itself is derived from Amartya Sen’s Entitlement Theory of risk.
The main purpose is to simulate different scenarios after a shock based on a comparison between known data from key-informants for a defined reference year and the data for the period of interest.
The advantage to this methodology is that the level of uncertainty related to data quality is assessed at each step of the process, thus allowing to corroborate –or not-- the previous steps (internal consistency) and then, more broadly, to test the user’s initial hypothesis. This gives to the model a high power of prediction while mortality measurements are retrospective and anthropometry telling you about the current situation.
1.5 How should I use the different elements of the manual and how are they related to the ENA software?
To conduct a survey using SMART, you should use the different modules (or elements) of the SMART manual to guide you through the process, from planning, training survey teams, analysis and cleaning of the dataset to report writing. The SMART manual is to be used along with the ENA software. The ENA software contains different screens which follow the survey process in a step-wise manner:
3) Data Entry Anthropometry & Results Anthropometry.
4) Data Entry Mortality.
5) Results Mortality and Options.
Please refer to the Capacity Building Toolbox to download the manual for ENA software.
1.6 Why should I conduct a food security assessment with SMART?
1.7 Where can I find answers to any technical questions about the SMART survey methodology (for planning, sampling, data collection, final data cleaning and analysis and/or reporting)?
2.1 In order to conduct a survey using the SMART methodology, do I always have to use all three elements together? In other words, do I always have to estimate malnutrition prevalence and crude mortality rate, as well as conduct a food security assessment in the survey area?
No, it is not always recommended to use the three elements together when conducting a survey using SMART. Each survey should be adapted to its context, therefore the survey planners should carefully select which modules of the SMART methodology are needed.
For example, in some contexts, it may be necessary to combine the anthropometry and mortality modules to conduct a survey, whereas in other contexts, it may only be necessary to conduct an anthropometric survey without the mortality component. Similarly, survey planners may decide to use additional indicators as part of the nutritional survey (e.g. hemoglobin concentration, diarrheal disease data, vitamin A supplementation coverage, measles vaccination coverage). These decisions need to be made carefully during the planning stage, keeping in mind that each additional dataset collected during the survey should be reported and lead to action. If the additional data will not be reported or will not be used, it would be a waste of resources and a waste of time to collect such an indicator. Moreover, additional collected data adds to the surveyors’ burden of work and may affect the overall quality of the dataset.
2.2 What is the difference between an anthropometric survey and a nutritional survey?
In the context of an emergency, an anthropometric survey only includes the basic measurements (including age, weight, length/height, gender, oedema) needed to calculate nutritional indices which are then used to derive estimates of malnutrition prevalence. MUAC is also often measured during anthropometric surveys. The target population is most often children 6-59 months of age.
In contrast, a nutritional survey includes additional data used to assess the nutritional and health status of the target population. Examples of such data may include hemoglobin concentration, vitamin A supplementation coverage, measles vaccination coverage, diarrheal diseases, and infant and young child feeding practices, among others.
2.3 If I conduct a nutritional survey, how do I decide which data to collect in addition to the basic anthropometric measures?
In order to make an informed decision, it is necessary to understand the context of the survey through the collection of secondary data (context data). The checklist below will help you determine the additional data that is needed in your survey:
• Decision to undertake a survey agreed upon by all parties involved.
• Objectives of survey defined.
• Geographic area and target population to be surveyed carefully defined.
• Timing of survey finalized.
• Expected malnutrition prevalence and crude death rate.
• Already existing relevant information gathered.
Please refer to the SMART manual for additional information.
Sample Size Calculations
3.1 If I conduct a nutritional survey, how do I decide on the final sample size to use in the survey?
The SMART tools (SMART manual and ENA software) allow you to calculate the sample size for a combined anthropometric and mortality survey adapted to the context, instead of using a 30 by 30 cluster survey (with 900 children) everywhere when it is not needed.
The more outcomes you measure in a survey, the more complex the sample size calculation becomes and the more demanding the survey is in terms of time and resources. In emergency contexts, it is best to keep the survey simple so that results can be gathered quickly and simply, thereby ensuring that resources are used efficiently, essential data is collected and that results are reliable.
If you are conducting a nutritional survey with multiple outcomes, you may refer to the CDC/WFP manual (2005) entitled ‘A manual: Measuring and Interpreting Malnutrition and Mortality’ found at the following link: http://www.unhcr.org/45f6abc92.html
3.2 When doing calculation with other software, I am getting different sample sizes compared to the calculation made with ENA 2011. For example, with an expected prevalence of 8%, a desired precision of 2.5% and 1.5 as design effect, I obtain a sample size of 739 children with ENA 2011, 678 with EPI INFO STATCALC, and 679 by using the SCHWARTZ equation in an Excel sheet. In general, the sample sizes are higher with ENA as compared to other methods. What equation is ENA using and what do such differences mean? It seems as though the sample size is already raised by 10% in ENA. Please provide clarifications on this matter.
First of all, please ensure that you have the latest version of ENA (ENA 2011). Please refer to the Capacity Building Toolbox to download the manual for ENA software.
The difference in the sample size calculation formula is such that for cluster surveys we use t=2.04, whereas for simple random sample we use 1.96. The other software mentioned does not take into account the fact that the number of degrees of freedom in the cluster surveys formula equals the number of clusters minus 1. Thus, the calculation using ENA is more appropriate.
If you try entering 1 as design effect and switch between the Cluster and Random checkbox on top of the Planning page, you will see the sample size change, although prevalence, precision and design effect remain the same. This is because of change in t from 1.96 to 2.04.
3.3 SMART recommends a specific approach to the planning process instead of using the standard 30x30 cluster design (30 clusters x 30 households/children). What are the benefits of this approach to the overall survey results?
SMART methodology promotes a bottom-up approach, whereby the surveyors form the basis for planning, and are not forced to survey an unrealistic number of households (or children) per cluster. With the 30x30 cluster approach, surveyors are required to survey 30 households (or children) per cluster (usually to be completed within one day), without taking into consideration their skills, the context, or the feasibility of it. The latter approach (top-down) could negatively affect the overall quality of the data, as surveyors may need to rush and take shortcuts to finish data collection by the end of the day. When using the SMART approach, good data quality is the most important aspect to consider; it ensures that reliable and comparable results are found. It should always be emphasized that proper planning and standardization of the process will help achieve this goal.
3.4 How is the number of clusters and cluster size derived when using SMART?
After calculating the final survey sample size, the survey planner needs to derive the cluster size by estimating the number of households (or children) that a team can realistically survey during one day's work. The overall calculated sample size is divided by this cluster size to arrive at the number of clusters to include in the survey. Of course, the final number of clusters will always depend on what is feasible in terms of available resources and timing. As a general rule, it is always better to have more clusters and a smaller cluster size for the results to be more statistically efficient. This usually results in lower design effects and smaller sample sizes being needed to achieve the required level of precision.
4.1 If I conduct a nutritional survey, can I enter the additional nutritional data in ENA?
Yes, you can enter the additional nutritional data in ENA by adding variables to the Data Entry Anthropometry screens (Data View). Please refer to the Capacity Building Toolbox to download the manual for ENA software. If you add a variable to your database, make sure to add the acceptable ranges in the Variable View screen; this will ensure that extreme values (such as data entry errors or outliers) will be detected by turning red. Also make sure to add the label and values of the new variable in the Variable View screen; this will ensure that your data dictionary is up-to-date.
If yes, will ENA also automatically analyze these additional data (similarly to the automatic analysis of the nutritional indices in the Results Anthropometry screen)?
No, ENA 2011 will not automatically analyze these additional data. However, it is now possible to use the statistical calculator to obtain some descriptive statistics for variables broken down by cut offs or ranges. It is also possible to export the data to Excel and then use Excel or more sophisticated software (such as Epi Info or SPSS) for analysis.
You may also use the Epi Info/ENA hybrid software for analysis of the additional variables. For more information, please click on the following link:
4.2 How do you adjust the flags in ENA (for example, -4 instead of -3)?
You can adjust the ranges of the SMART and WHO flags on the bottom right of the Options Page in ENA, although it is not recommended to do so for the plausibility check. You should keep the set ranges (-3, +3 for SMART) in order to be able to compare the plausibility check with other surveys in terms of representativeness and data quality. Only in the case of excellent supervision and full confidence in your survey teams would you adjust the ranges for the flags for your survey report.
You can also decide to add some “data entry flags”. The software will alert you if your values are outside the set range by colouring outliers in purple. To do so, go to the Anthropometry screen and click on the Variable View tab at the bottom left to adjust these ranges.
Note: Make sure you understand the difference between data entry flags and plausibility flags.
Please refer to the STP Modules 5 and 7 (Anthropometry and Plausibility Check) in the Capacity Building Toolbox.
4.3 In previous versions of ENA, values entered in additional columns (e.g. Measles) were obscured by (Measure, Strata, Clothes, Weight Factor) when data was exported to Excel. Has this bug been rectified?
Yes. To have the columns appear on your data entry screen for Measure, Strata, Clothes, Weight Factor, go to the Options page on the bottom left-hand corner and check the corresponding box.
4.4 I am having trouble generating Word and/or Excel files with ENA 2011 using previous versions of Office. Do you know how to solve this problem?
One explanation for this problem could be that there is a space in the path for the installation folder. If this is the case, the connection to MS-Office doesn’t work under Windows XP. To install ENA, click on the following file http://www.nutrisurvey.de/ena2011/ena2011.exe (Version 2011, Nov. 8th 2011). After installing you will find an icon for the program on your desktop. The default installation directory is the user directory, but if you use Windows XP please install the program in c:\ena2011 or d:\ena2011 to avoid problems with the connection to MS-Office.
4.5 ‘Survey Date’ values are often ‘flagged’. Is there a correct format to enter dates?
ENA can usually detect various date formats; however the most common date format is day/month/year. If your date values are flagged, try closing and re-opening the document.
Also check the regional language settings in your computer’s Control Panel. It is possible that your laptop has the US American format where dates are entered in ‘mm/dd/yy’. Under the START menu, click on Control Panel. Under Regional and Language Options look for Change date, time or number format. Select the UK format of ‘DD/MM/YY’. Double check that both the short and long date formats are ‘day, month, year’.
4.6 How do I use the new WHO Child Growth Standards to calculate and report prevalence of malnutrition with the ENA software?
The ENA software allows you to calculate nutritional indices by either using the NCHS reference 1977 or the WHO Growth Standards 2006 in the Data Entry Anthropometry screen. Survey results should be reported using both the NCHS reference and the WHO Growth Standards until the WHO Growth Standards have been fully adopted in the country where you conducted the survey. For example, depending on what is appropriate and acceptable, you may decide to include the NCHS reference prevalence estimates in the main body of the survey report while reporting on the WHO Standard prevalence estimates in the appendix, or vice versa.
4.7 What is the difference between the WHO Anthro 2005 software and the ENA for SMART software?
WHO Anthro 2005 software consists of three modules: 1) Anthropometric calculator, 2) Individual assessment and 3) Nutritional survey (cross-sectional). The ENA for SMART software has been designed specifically to be used for combined anthropometric and mortality surveys (cross-sectional) along with the SMART manual. Anthro software, therefore, has a larger scope of use. Both software contain similar features for entering and analyzing anthropometric data from children and will generate identical results.
As compared to the nutritional survey module of Anthro software, the ENA software contains more screens and different options to help survey planners and managers plan a survey, standardize survey teams, perform data quality checks, and write final survey reports. In addition to the Data Entry Anthropometry and Results Anthropometry screens found in the two types of software, ENA for SMART software contains: 1) a planning screen to calculate sample sizes, randomly select clusters and generate random number tables, 2) a training screen to analyze data from the standardization test on weight and height/length measurements, 3) a plausibility report to check survey data quality, 4) two screens for mortality data entry and analysis, and 5) an option to automatically generate a report with the main survey results incorporated. Additionally, ENA can generate standard questionnaires for the collection of both anthropometric and mortality data, among other useful options (e.g. double entry, merge surveys).
Click on the following link to read more about WHO Anthro 2005 software, its benefits and recommended uses: http://www.who.int/childgrowth/software/en/
Anthropometry/ Standardization Test
5.1 What is the problem with using flexible/portable tapes for height in SMART - especially in areas where portability is an issue (e.g. teams working for hours or wading through swamps)?
Even though the flexible/portable tapes for height are more practical to carry around, SMART does not recommend using these tapes, as they introduce measurement bias for the following reasons:
The recommended height board weight is between 5 and 7 kg (the Shorr model for 2011 weighs 4.6 kg). Newer models come with a carrying bag for the field.
For Recommended Survey Equipment, please see Resources.
5.2 When team members only have basic literacy and/or numeracy, can the MUAC tape be folded in half to find the mid-point, instead of dividing its length by 2?
We strongly advise not to fold the MUAC tape. Folding the MUAC tape introduces measurement bias, since the fold introduces some millimetres of error. Some survey managers recommend using a string and folding it in half to find the midpoint.
5.3 Is 10 children the minimum that can be used for a Standardization Test?
Yes, SMART recommends 10 children as the minimum for reasons of precision and accuracy in the enumerators’ measurements.
Plausibility Check/Final Report
6.1 After running the Plausibility Check, I obtained the following results for the Digit Preference Test for MUAC (see Table below). How could these results be explained?
Digit preference MUAC:
Digit Preference Score: 41 (0-5 good, 6-10 acceptable, 11-20 poor and > 20 unacceptable)
Your MUAC data was most probably entered in cm. You need to convert to mm for ENA to produce correct results. Remember, ENA always assumes that MUAC is in mm.
7.1 What is the difference between the SMART methodology to measure mortality rate as compared to the other methods used to measure mortality rate (namely past household census method (PHH census) or current household census method (CHH census)?
The method recommended by SMART to measure mortality addresses the issue of migration or population movement. In an emergency, it is likely that more people will both leave and join households. With equal in- and out- migration, the SMART, the CHH, and the PHH methods should all give the same results. With excess in-migration, then CHH underestimates and PHH overestimates mortality rates. SMART takes additional time during a census (to determine people leaving and entering the household) but is more accurate and gives data for in- and out-migration rates.
SMART Methodology - 2012