Menu ENN Search
Language: English Fran├žais

Use of single coverage estimator with expanded OTP protocol

This question was posted the Coverage assessment forum area and has 7 replies.

» Post a reply

Hugh Lort-Phillips

Global Coverage Advisor

Normal user

10 Nov 2019, 06:00

In the Cox's Bazar refugee settlements, nutrition sector partners are using an expanded protocol for OTP. Children are admitted to OTP based on regular international standards (MUAC <115mm and WFH <-3 Z score) (oedematous cases are referred straight to inpatient care), however since July 2019, they are discharged only when they become well nourished (MUAC ≥125mm and WFH ≥-2 Z score). 

I am supporting the sector with a coverage survey using the SLEAC methodology. If they were using a regular discharge criteria (as they were before July), I would classify coverage using the single coverage estimator by estimating Recovering cases out of the programme and incorporating this figure into the denominator. 

However is it appropriate to do this when an expanded protocol is in use?

My concern is that, because of the expanded protocol, there may be a large number of OTP Recovering cases IN the programme which would result in a disproportionately large number of Recovering OUT cases, leading to an inaccurate estimate and classification. 

Many thanks in advance for any advice or opinions on this. 

Mark Myatt

Epidemiologist at Brixton Health

Frequent user

11 Nov 2019, 10:05

Hugh,

I am not sure what you mean by an expanded protocol for OTP. Please explain. I think you mean that cases are kept in the program until they are not severely or acutely malnourished (i.e. MUAC ≥ 125mm and WFH ≥ -2 z-scores). This is not so odd.

I can see the source of your concern ... we calculate the number of recovering cases OUT of the program using an estimator that includes the recovering cases IN the program. I am not sure that we have to worry much about this as long the assumption that incidence and coverage do not change rapidly over time still holds.

I would adopte a "suck it and see" approach and calculate point, period, and single coverage estimates. The single coverage estimate should be quite close to the average (mid-point) of the point and period coverage estimates. If you have concerns about the validity of your single coverage estimate then you can present the point and period coverage estimates with interpretations:

  • Point coverage reflects the ability of a program to find and recruit cases. The point coverage estimator does not account for recovering cases and so does not directly reflect the program's ability to retain cases from admission to cure. 
  • Period coverage reflects the ability of a program to find, recruit, and retain cases. The period coverage estimator does directly reflect the program's ability to retain cases from admission to cure but tends to overestimate program performance because the denominator does not include recovering cases that are not in the program.

I hope this is of some use.

Hugh Lort-Phillips

Global Coverage Advisor

Normal user

11 Nov 2019, 10:59

Mark, as ever thanks for your speedy response. 

I should have been clearer about the expanded protocol - it does indeed mean that children in the OTP are discharged only when they fulfill the MAM treatment discharge criteria (>= 125mm and WFH Z-score >= -2). 

Thanks for your suggestion - we'll have a look at the results and take a decision based on how the different estimators look. 

Thanks again,

Hugh

Paul

Frequent user

11 Nov 2019, 11:52

Hi Hugh,

I think I see what you might be concerned about, do let me know if this response captures it. There are different examples / adaptations of expanded criteria for different countries. The effect of the expanded protocol will depend where you are. The potential is for the SAM child to be recovered with lower doses (e.g. 2 packets per day) of RUTF either while still SAM or (1 packet per day) when they reach 'MAM' status (by anthropometry) during recovery. 

This lower dose during treatment potentially affects the mean length of a treatment episode and the subsequent calculation of K. The effect of this is likely to be more pronounced if the age profile of admissions shows a greater number of children aged > 2 years (they need bigger doses, e.g. 3-5 packets per day, during initial recovery than is sometimes available with expanded criteria). You might see this if there are a high proportion of WFH admissions vs MUAC admissions since MUAC will tend to recruit younger cases, for whom the lower doses of RUTF would be the 'normal' ration.

The differences in mean LOS are however likely to be fairly small and potentially you could ignore them. However you could check the average length of stay for cured cases and calculate K based on k = mean LOS in months / 7.5 (mean length of an untreated episode in months). Your interpretation should include considerations of ration size, mean LOS and case finding methodology (using MUAC / WFH).

Check also if there is an updated figure for the mean length of an untreated episode. I heard on the grapevine that UNICEF / / Global Nutrition Cluster is involved in the development of different figures for different contexts. The Rohingya context is likely to be unusual and (possibly) not equivalent the stable context used for the estimate of 7.5 months. Ultimately the concerns may be academic and relevant only to a topline figure for the coverage. As Mark says, point and period coverage may be more utilitarian and diagnostic to focus on the programme changes needed to improve recrutiment and treatment protocols respectively.

I hope this helps,

Paul

Mark Myatt

Epidemiologist at Brixton Health

Frequent user

11 Nov 2019, 13:34

Thanks for the clarifications. It looks like Hugh has a SAM treatment program discharging case then they are not severely or acutely malnourished following, I think, current WHO guidelines.

I am used to referring to the type of program that Paul describes as "combined" (since they treat both MAM and SAM in the same program) or "simplified". I think he is right when he suggests calculating "k" using observed lengths of stay in the program being assessed. The effect may be larger than Paul anticipates. A change of mean LoS from 10 weeks to 12 weeks might increase the estimate of R_out by about 20%.

Sheila Isanaka's group has been working with UNICEF on the duration of untreated SAM with the focus on burden estimation where the incidence from prevalence correction factor is often based on the they are not severely or acutely malnourished. Their work my be of use at finding "k" in the coverage context.

I think there are problems with case-finding in coverage surveys for MAM treatment programs as active/adaptive case-finding methods tend not to perform well for MAM cases. I would probably use door-to-door screening to find MAM cases.

I think the single coverage estimate has an intuitive meaning and is relative unbiased. It also addresses the issue of which of point of period coverage to focus on. I also think that point and period coverage do have their uses provided it is clear what they measure and how they are biased. I think a concentration on the "topline" coverage figure can be unhelpful as the barriers and boosters data in SLEAC and SQUEAC are likely to be more useful in identify problems and improving coverage.

Hugh, Please let us know how you get on.

Paul

Frequent user

11 Nov 2019, 14:38

Hi Mark,

I'm not entirely sure - but I thought the "expanded protocol for OTP" may have referred to the guidance in the MAM decision tool (2017) which refers to "exceptional CMAM programming" and "expanded admission criteria" which gives various options for admission criteria / RUTF / RUSF dosages in emergencies when supplies may not be adequate.

Thanks for the clarifications.

Paul

Hugh Lort-Phillips

Global Coverage Advisor

Normal user

13 Nov 2019, 06:07

Thanks Paul and Mark for your detailed comments and suggestions.

It was my fault for the confusion - I assumed that "expanded protocol" was a universal term which was used to describe the current admission and discharge criteria for the OTP. As it is TSFP also operates in the camps distributing RUSF with the regular admission criteria (<-2 WFH z-score and <= 125mm). BSFP also operates across all camps. All children aged 6-59 months are eligible for this BSFP: WSB++ is the product distributed. 

The mean LoS for the OTP was 10 wks (2.5 months) in September 2019. And 11 weeks (2.75 months) for the TSFP. So I shall use these in the calculation of the K factor when calculating Rout. As for the length of stay for an untreated episode of SAM or MAM, I'm not sure what the correct figure is to use in this context. I will make enquiries while I am here. 

Mark - case finding is being done for SAM and MAM cases at the same time. So teams are conducting door to door screening. 

Good point about not concerntrating on the topline coverage figure! I always emphasize the importance of the investigation of barriers and boosters during surveys I support. In this case however, there is a very real interest from partners to know camp level classifications and combined camp coverage estimates following the low coverage classifications identified during the assessment in 2018. Parallel programme data investigations and qualitative investigations are however ongoing to try to understand the root causes of high and low coverage. 

If partners are happy for me to share the results, I will happily share them with you both in due course.  

Mark Myatt

Epidemiologist at Brixton Health

Frequent user

13 Nov 2019, 10:14

Hugh,

No worries over the confusion. Our field is full of jargon and acronyms. We love words like "expanded" and tend to apply them to pretty standard programming practice.

There is no "length of stay for an untreated episode" as there is no treatment to "stay" in. This is the duration of the disease episode (i.e. from the start of an episode to recovery or death) without treatment. It is difficult to know but there has been some progress on estimating it. It looks to vary considerably from place to place.

As for sharing results ... My interest is in seeing how the different coverage estimators compare and whether the single coverage estimator returned a nonsense estimate.

Good luck
 

If you have any problem posting a response, please contact the moderator at post@en-net.org.

Back to top

» Post a reply