What is the sampling procedure for a micronutrient survey within the S3M survey?

This is a big and somewhat undefined question.

GAIN has done a lot of micronutrient testing in S3M surveys and in their FACT surveys. This is testing of micronutrient levels in supplementation vehicles (e.g. iron / follate in flours). UNICEF and the Sudan FMOH looked at salt iodisation in the 2013 National S3M Survey. I have the GAIN work to hand and you can contact me off-forum if you need the reports and the methods or to be put in touch with the primary investigators at GAIN.

If you mean assessment of biological samples collected from human subjects then you should contact specialists and organisations who have done this (e.g. micronutrient surveys) before. Processes can be expensive and difficult. Micronutrient surveys are often expensive even with small sample sizes. S3M samples are usually very large and you many want to do the micronutrient survey on a small sub-sample (e.g. a few children for each PSU or all sampled children from 5% of PSUs). The sample would need to be carefully designed to allow mapping at state level.

The easiest to do is haemoglobin. UNHCR have been doing this for some years in their SENS surveys (like SMART but more indicators) at point of sampling using HemoCue devices. Non-invasive devices are becoming available.

I hope this is of some help.

Durria,

I recognised the name but I was not sure it was you.

You need to take advice from WHO (and others) about taking, labelling, storing, and transporting the specimens. I would have a nurse or phlebotomist in each team to draw blood for anything more substantial than a pinprick. Sharps and the risk of infection will need strict policing. I would use a reference lab. If you use a local lab then validate a subset of results with a reference lab making sure the local lab knows that this will happen.

With S3M we tend to estimate or classify for small areas. This require use to have reasonable large samples from each PSU and, for national surveys, many PSUs. In the planned S3M I think there will be about 3000 PSUs with 32 HHs from each PSU. If we assume that there will be 1.5 children per HH then the overall sample size will be about 144,000 children. To include them all in the micronutrient sample will be very expensive. What you can do is to calculate a state level sample size and try to get that as a subset of the S3M sample from each state. If (e.g) you had m = 1OO PSUs in a state and a required an overall sample size of n = 400 then you would need to sample n = 4 children from each PSU. Since you have many small clusters the design effect will likely be very small. If you have staff limits then you may want to take n = 15 children from m = 30 PSUs. I would advise not to go much below m = 25 PSUs from each state. As the number of cluster drops and the within-cluster sample size rises you may start to see substantial design effects.

You need to make sure that you can assign each specimen to the S3M PSU number. That will allow analysis by state and by region and overall.

I hope this helps.