Dear Anonymouse 118, This is for clarification. what is your admission criteria? Thanks

I think you mean CSAS or SQUEAC. There is also a cheaper method called SLEAC which use LQAS classifiers. There is also a new method called S3M that is a development of CSAS. This is still under development. CSAS and SLEAC are simple survey based approaches. The principal differences are (1) CSAS estimates coverage and SLEAC classifies coverage (e.g. SLEAC can tell you if coverage is below 50%), and (2) CSAS can produce detailed maps of coverage. If you are dealing with IDPs both the CSAS and SLEAC methods will need you to do some work on developing and testing a case-finding method (you should get in touch with VALID International Ltd as they have already done some work on this with Darfur IDPs). I usually do the development and testing of the case-finding method with the survey team supervisors. This usually takes 3 to 4 working days (so if you have 3 survey teams that will be 9 - 12 person days). CSAS usually divides the survey area into small areas of less than about 150 sq. km (much smaller for densely populated areas). There are usually about 30 areas. The rule is that one survey team samples one area in one day (this also effects the choice of small area size) sampling between 3 and 6 villages per day. With the example of three teams of 3 people the main survey will take 10 days (30 person days). Data analysis and reporting are very simple. I think you will need to employ a consultant for about three weeks. SLEAC is quicker. A typical SLEAC survey can be done with three days fieldwork (you still have to get the case-finding right). You need to factor in the travel time when employing a consultant. SQUEAC is a very different method that can estimate and map coverage but also provides in-depth discovery and analysis of barriers to coverage (note that CSAS and SLEAC do provide some data on barriers). As a rule the minimum time needed for a first SQUEAC with untrained staff is about 18 working days working with 6 - 9 people (more people won't make it much quicker). Subsequent SQUEACs are much quicker. If you use a W/H based case definition then workload will probably increase above what I have above due to the poor ability of people to be able to recognise low W/H and the need to measure and weigh suspected cases. I always recommend using a few small teams as this allows for closer supervision and gives better quality data. It does mean, however, that the consultancy cost increases. VALID International Ltd. are the leading agency for these sorts of surveys. There are a few freelance consultants for CSAS and SQUEAC. Other issues : Indirect methods of assessing coverage are usually inaccurate. Observing a difference in prevalence between WHZ and WHM is not particularly surprising since WHM < 80% is slightly more restrictive than WHZ < -2. For example, using the NCHS 1977 reference for a non-stunted 12 month old female (height = 74.3 cm) we need a weight of < 7.75 kg to have a WHZ < -2 but a weight of < 7.55 kg to have a WHM < 80%. The difference you report is rather large. You may want to check the original data. I am not sure why you do not quote prevalence by MUAC since the preferred admission criteria is based on MUAC. The use of W/H is associated with low coverage. In this population, W/H will probably overestimate prevalence due to low sitting to standing height ratios. I hope this helps.

@Kiross: we use MUAC for community-based screening and/or weight for height at the clinic level screening - a child who does not meet both criteria would still be admitted. @Mark: thanks so much for the explanation and time estimates; that is very helpful. I don't know why the survey was done the way it was - not using MUAC. That NGO may not have been aware of the newer developments, given that they still used the NCHS reference.

Hi, Thanks for clarification. it seems to me that you are using mixed criteria (MUAC or WFH). if this is so, i think it will be difficult to calculate the coverage using indirect method as the method for prevalence estimation and admition are different. the other point is, you stated as "a child who does not meet both criteria would still be admitted" this means the program is blanket. then what is the important of daily screening? or is it for surveilance purpose? Thanks Kiross Tefera

Hi Kiross, Children who have <-3 WFH but not MUAC <115mm will be admitted and vice versa, and those with edema, which we have not seen in our area. That means only children with SAM are admitted. I think you could still use an indirect method even using WFH and/or MUAC in the facilty-based screening. Several countries I have been to have this same procedure in their national protocol and I have seen NGOs use this in other places. Coverage surveys still take place using LQAS and SQUEAC in those places. Thanks for your advice!

Hi Mark, Can I get a little clarification on your estimates of resources for the different types of surveys? You said it will take 3 teams of 3 people about 10 days to do a CSAS survey, and 6-9 people about 18 days to do a SQUEAC survey the first time it's done. There is a fair amount of set-up work, especially using SQUEAC. How many of these days are field days? Also, if the staff are fairly quick to pick up on the methods, how long should the SQUEAC take second or third time around? Thanks for the help, Merry

It is easy to work out how long a CSAS survey will take because CSAS is just a survey method. One survey takes about as long as another. Think of is as SMART for coverage. SLEAC is a quicker version of CSAS (about 4 - 5 days per district) but does not provide the level of detail of mapping that CSAS provides. We recently completed a national SLEAC of a West African country in six weeks. That's an entire country of 13 regions all mapped out in 30 working days! SQUEAC is a very different thing from CSAS or SLEAC. It is a set of tools. It usually takes about 18 working days to go through the use of all the tools in the toolbox. This is usually overkill for a single coverage assessment but it makes for a thorough training. A full SQUEAC by a trained SQUEACer should about eight days (say two weeks to be on the safe side). I find that I learn very little after about five days but spend the next five days proving what I already know. After that you might spend a day or two a month on SQUEAC activities and do a more intensive bit of SQUEACing every six months or a year. To do all this, however, requires some familiarity with the SQUEAC toolbox ... hence the full 18 days training. I have contacted Saul Guerrero at ACF to add something about their recent experiences with rolling out SQUEAC with iNGO staff. My understanding is that they can do a pretty good job with remote support in about ten days. I may be wrong so I have asked him to comment here himself. I hope this helps.

Hi Mark That is useful. One more quick question. Once you have your areas selected for the small area surveys in a SQUEAC survey, how do you know what is the minimum number of cases you have to find in each area? Thanks again, Merry

The process is that you select the communities to sample and then sample them using either house-to-house and door-to-door (H2H) sampling or use the active and adaptive case-finding (AACF) procedure that was developed for CSAS surveys. Both of these methods are capable of finding all, or nearly all, cases in the community. The sample size is the number of cases that you find. An LQAS sampling and classification plan is then calculated and applied. This is a very different way of working than most people are used to. We do not know how many cases we will find in a small area survey unless we have a good idea of local incidence, durations of untreated episodes, and duration of SAM before admission. This make it difficult to plan a survey to get a sample size although it makes sense not to sample more than one or two small villages (you have to give yourself the chance of finding some cases!). SAM is a rare condition and SQUEAC is a rapid method. This means that small-area surveys often have small sample sizes (sometimes we surprise ourselves find a lot of cases). This brings with it the spectre of error. It is important to realise that the sample will have a very large sampling proportion (i.e. the sample will include a very large proportion - in tests using capture-recapture methods always > 80%, of cases in the sampled communities). This reduces sampling error. Also, we are not surveying "cold" in the sense that have a prior hypothesis regarding coverage. We decide our survey areas based on a lot of data (e.g. an old SMART survey found cases there but we have not admitted anyone from there, we have not had good outreach there, the area is much more than a half-distance between markets from the nearest site offering CMAM services, and so-on). This is the equivalent of a two stage screen. The first stage boost the positive and negative predictive values of the second test even when the second test would perform badly if used on its own. Important issues are ... (1) You need to be sure that your case-finding method works. We have found that AACF is easy in most rural settings but fails in large towns and in IDP / refugee camps. We have used social network analysis with multiple informants to make AACF work in IDP settings (Darfur camps) and in urban settings (Lusaka, Zambia). We have found that H2H is a good fallback in most camp and urban settings. (2) The community sample is not taken PPS. This type of sample will put you in bigger villages whereas you will want to include sub-villages and hamlets. SAM is often a disease of economic and geographical marginalisation. We often replicate this marginalisation in our programs and in our sampling. We tend to use spatial methods such as CSAS or transect sampling since these tend to avoid problems such as sampling along roads (PSS and convenience sampled will tend to do that). Remember that the sample is "purposive" and we can allow bias. This is known as "optimal bias" ... we intentionally bias our sample to where we believe cases to be and where we believe those cases to be uncovered. (3) Timing is important. Don't sample on CMAM clinic days, market days, distribution days, national or religious holidays, &c. We also find that we occasionally need to sample at particular times of the day. You won't (e.g.) find mothers and kids at home at certain times ... we have used labour timetables (e.g. to avoid times when mother are away in the fields doing agricultural labour) and meal timetables (e.g. they'll be at home will preparing / eating meals). This is why (e.g.) you get bothered by telephone salespersons when you sit down to your evening meal ... they know that someone will be home at that time. I hope this helps.

Hi Mark/Merry To follow up on the issue of the number of days needed to "complete" a SQUEAC. I dont think there is a standard answer, as it will depend on a series of issues which include (but is not limited to) the size of the programme area, population density, number of villages to sample, etc. The first SQUEAC also tends to take slightly longer than subsequent ones. Based on our recent experiences, we now calculate for around 14 days, of which around 8 are used for data collection (for all three stages) and the rest for analysis and write-up. This would apply mostly to SQUEACs done with the support of an experienced investigator, but I feel that it is important to try to continue to aim for a similar timeframe for all exercises if these are to be truly mainstreamed by nutrition organisations. If you would like any further information, don't hesitate to let me know. Best Saul