The Anthro Survey Analyser is an online tool developed by the Department of Nutrition for Health and Development at the World Health Organization (WHO), that allows the user to perform comprehensive analysis of anthropometric survey data. The tool incorporates standard methodology as in the WHO Anthro Software Nutrition Survey module and adds more accurate calculations of confidence intervals and standard errors around the estimates taking into account complex sample designs, whenever is the case. In addition to age, sex, type of residence and sub-regions/districts, the Anthro Survey Analyser provides also stratified analysis results by wealth quintiles, mother education and allows users to define one more stratification factor relevant to the survey/research.
The tool is interactive and allows users to visualize outcomes on input data, z-score distributions, data quality assessment summaries. In addition, it provides a semi-filled summary survey report template with main results, including graphs and tables. The current version of the tool provides results for four of the anthropometric indexes: height-for-age, weight-for-age, weight-for-height and body-mass-index-for-age. Users are able to output two files in Excel format:
The summary survey report template is exported in Word. It is a summary report template laying out guidance on minimum required details to follow good practice in reporting, including some output for data quality assessment. The report also includes main findings (graphics and tables) regarding prevalence estimates by different disaggregation factors for the five main indicators, namely stunting, wasting, severe wasting, overweight and underweight, as well as data quality assessment statistics.
Ensure that your data is prepared in the correct format for the Anthro Survey Analyser. Refer to the section on Data Preparation in the Quick Guide–specifically, the table on the accepted values for each variable–to find the best way to prepare your data to ensure that your variables will be mapped in the tool.
The Anthro Survey Analyser does include validation checks for each mapped input variable used for the analysis, and provides user-friendly messages to guide the user in detecting potential mismatches. However, the user is strongly encouraged to perform preparation of the data prior to importing the file into the Anthro Survey Analyser. Table 1 provides guidance on accepted values for each of the variables to be mapped as input to the analysis and information on whether the variable is compulsory. In order to have the best accuracy for the output estimates, the user should know what the compulsory variables are as well as the recommended format for each of the mapped variables.
The data file to be imported should be in a comma delimited format (.csv). The file can be created in any spreadsheet software used for the organization, analysis, and storage of data in tabular form such as Microsoft Excel. Once the data is properly organised it can be saved as or transferred to a .csv format.
Attention: This application is based on R code. Therefore any variable label can only contain characters, numbers, _, and -. It should not include spaces or symbols. This also applies to the file name to be imported. For example, names such as country survey.csv or survey2013&2014 are not accepted.
| Variable | Compulsory or optional? | Accepted values and other details |
|---|---|---|
| Age-related variables: Date of birth & date of visit (recommended) or Age (in days, or in months) |
Compulsory | Date of birth AND date of visit: DD/MM/YYYY. The use of the two date variables for calculating the exact age of the child is the recommended, best practice approach. Note: if DAY is missing for the date of birth, it should be replaced by 15. If month or year is missing, the date value should be set to missing/blank. If Date of birth and Date of visit are provided, the mapping of the variable Age will not be available to the user. Age: numeric. If in days, it is calculated as date of visit minus date of birth (integer value). If in months: it is calculated as age in days divided by 30.4375 (float value). In this case, decimals should always be provided for more accurate calculations of z-scores. Note: For all cases where age is missing, only results for weight-for-height will be computed and children will be accounted for the total sample size (0 to 5 years), but not classified in the age groups. |
| Sex | Compulsory | Numeric or text. For males '1', 'M', or 'm'; for females '2', 'F', or 'f'. |
| Weight | Optional | Numeric, float value (in kilograms with decimals) |
| Length or height |
Optional | Numeric, float value (in centimeters with decimals). |
| LH measure (Standing or recumbent position for height or length measurement) |
Optional |
Character. Recumbent length ('L' or 'l') or standing height ('H' or 'h'). Note: it is recommended that recumbent length is used for children aged less than 731 days and standing height for those aged 731 or more days. As such, if this information is missing, the tool applies the values of 'L' or 'H' according to this recommendation. |
| Oedema | Optional | Character. For no oedema 'N', 'n', or '2'; for oedema cases 'Y', 'y', or '1'. Note: For all cases where this information is not provided, the tool's code considers the cases as having no oedema. Z-scores for all weight-related indexes will be set to zero when oedema is present. For prevalence calculation purposes, children with oedema are classified as having severe malnutrition (weight-for-height < -3SD, weight-for-age < -3SD, and BMI-for-age < -3SD) |
| Sampling weight | Optional | Numeric float. If not provided, all children will be assumed to have identical weights (i.e. equal to 1); that is, unweighted analysis will be carried out. If provided, all children with missing sampling weights will be excluded from the analysis. |
| Cluster | Optional | Numeric integer. If not provided, all children will be assumed to have identical cluster (i.e. a one-cluster design) with the default value of the cluster set to be equal to one. If provided, all children with missing cluster values will be excluded from the sample. Note: Clusters must be nested within strata. This means one cluster cannot belong to more than one strata. In this case, the application will not calculate prevalence estimates. |
| Strata | Optional | Numeric integer. If not provided, all children will be assumed to have identical strata, i.e. a one-strata design where the default strata value is equal to one. If provided, all children with missing strata will be excluded from the sample. |
| Residence type | Optional | Numeric integer or character. Any values are accepted. The recommended values of 'Rural' or 'Urban', however, are preferable for the purposes of output interpretation. |
| Geographical region | Optional | Numeric integer or character. |
| Wealth quintiles | Optional | Numeric integer or character. From 1 to 5 with 1 being the poorest and 5 the richest. |
| Mother's education | Optional | Numeric integer or character. Recommended values are 'None', 'Primary', and 'Secondary'. Any number of categories or values are accepted for the analysis, provided sample sizes are sufficient in all categories. However the common, standard recommended categories are no education, primary school, and secondary school or higher (corresponding to 'None', 'Primary', and 'Secondary' respectively). Note: Mother?s education refers to the highest level of schooling attained by the mother. |
| Other grouping variable | Optional | Numeric or character. Any variable that is of interest for obtaining results from stratified analysis. |
| Filter variable(s) | Optional | Numeric or character. Note: Binary variables (with values of '0'/'1' or 'Yes'/'No') are preferable to facilitate the selection of included records by the applied filter. |
| Recoding missing data | Blank or empty cell. In case of missing value codes such as 9999, 9998, etc., the cells with missing values should be replaced with a blank/empty cell before uploading the file in the application. |
Once Upload data is activated, click on Browse to locate the dataset to be analysed on your computer. Keep in mind that the uploaded files are required to be in a csv format.
Variable mapping requires the user to manually select the variables from the dataset that corresponds to the variables used for analysis. As a part of the data validation, only the formats specified in Table 1 is possible for each variable selection. The tool is able to recognise the correct format for each variable. In the instance that no variable-specific format is found in any of the available variables in the dataset, a pop-up warning message will be seen.
Please refer to the Quick Guide, a PDF document you can find at the bottom of this page, which provides comprehensive guidance on using the Anthro Survey Analyser.
Alternatively, please contact us on anthro2005@who.int.
This is version 1.0.1 of the Anthro Survey Analyser (25-09-2019).
The Anthro Survey Analyser is intended to be a useful tool for individuals in National Statistics Offices, data collection specialized agencies or programs, research centres, and any other institutions responsible for the analysis of anthropometric child indicators. It can be especially useful for users who do not have access to standard statistical software to analyse surveys.
In terms of output, there are some additions in the Anthro Survey Analyser:
expanded format with the following measures included:
The application was developed using R: A Language and Environment for Statistical Computing and shiny: Web Application Framework for R . Additional R packages used to support the application include:
The license and terms of use are compiled in a PDF file that you can download here. It governs the online and offline version of the tool.
If you have any further questions please contact anthro2005@who.int
We would appreciate any feedback or ideas on the Anthro Survey Analyser. Please send an email to anthro2005@who.int.
When reporting a problem, please include the following information:
If you discovered a security issue, we kindly ask you to contact us privately by e-mail so we have enough time to investigate and fix the issue.
Thank you for your interest in the Anthro Survey Analyser!