Only MUAC for admission and discharge?

Just reviewing this thread ... someone asked about:

    Roberfroid D et al. Background review. Utilization of mid-upper
    arm circumference versus Weight-for-height in nutritional
    rehabilitation programmes. 2013

The big problem, IMO, with this review is that it suggests that the type of exemplified by the Grellety study (discussed above) would provide appropriate and useful evidence (this is the in the concluding sentence of the review). I think, given the problems with this type of evidence (discussed above), that the review has not been very useful. I fear that it has inspired a lot or not very relevant work ans sown confusion.

Dear Kiros,

You asked me to comment the systematic review:

Utilisation of mid-upper arm circumference versus weight-for-height in nutrition rehabilitation programmes: a systematic review of evidence.

http://www.who.int/nutrition/publications/guidelines/updates_management_SAM_infantandchildren_review1.pdf

The current discussion is whether MUAC alone or MUAC and WFH should be used to select children in need of treatment. I am not sure this is really what is discussed in this WHO working document.

To choose between WFH and MUAC as admission criteria, you have to give preference to the indicator which selects from the community children with the highest risk of death and which are most likely to respond to treatment (currently, RUTF + antibiotics).

This issue was investigated in 6 different community based studies examining ROC curves (sensitivity plotted against specificity for a whole range of cut-off) of MUAC and WFH to predict deaths (Bairagi, 1981) (Alam et al., 1989) (Vella et al., 1994) (Van den Broeck et al., 1996) (Briend et al., 2012) (Mwangome et al., 2012). All of them showed that MUAC had a ROC curve above that of WFH. In several of these studies, a prediction as good as that of WFH could have been obtained by just tossing a coin. A review in 1994 on the relationship between anthropometry and mortality based on studies available at that time concluded that “the most consistent observation is that WFH is the least effective predictor of mortality” (Pelletier, 1994).

Before going to MUAC, one has to check that selected children respond to treatment. The response to treatment of children selected by MUAC compared to those selected by WFH has little been investigated. The WHO working document, however, quotes 4 studies examining this response. None of them suggest that children selected by MUAC respond differently.

So in summary, there is consistent evidence that MUAC better selects high risk children who also respond to treatment. The conclusion is clear: we should go for MUAC.

None of the community studies on anthropometry and mortality is mentioned in the WHO working document, and the association between MUAC and risk of death not highlighted, although this association is behind the current move of programmes towards MUAC, independently of its easiness of use in the community. The only reference to mortality in this WHO working document is based on treated children with an unknown representativity of the examined sample compared to the overall population. Also, there is no mention of the specificity of WFH compared to MUAC, which considerably weakens the conclusion of the analysis. A randomly allocated variable not associated with mortality can predict all deaths in a sample if a very low specificity is accepted.

The WHO working document, and also two recent papers (Isanaka et al., 2015) (Grellety et al., 2015) raise the question of children of the risk of death of children with WFH <-3 and MUAC > 115mm, currently not selected by standard MUAC only programmes. And indirectly advise for possibly using both indicators. I don’t think, however, this is the correct approach to improve the identification process (Briend et al., 2012). ROC curves show that increasing the MUAC cut off is will be more effective and we also know this will be programmatically much easier. So I suggest that instead of continuing this sterile discussion about including children with WFH <-3, we should consider increasing the MUAC cut off. Children with a MUAC of 116 mm are hardly better off than those currently selected and regrettably there is no discussion about the opportunity to treat them as well, although MSF in different programmes used this approach successfully (see: http://files.ennonline.net/attachments/2401/FEX-50-Web-Final.pdf p 108).
This option of increasing the MUAC cut off instead of keeping WFH < -3 is not discussed in the WHO working document.

A frequent objection to increasing the MUAC cut-off is that this will increase dramatically the number of patients admitted for treatment. This is true, there will be an increase of patient numbers, and the decision has to be thought carefully, taking into account treatment capacity. The effect on treatment coverage has to be assessed. However, adding children with WFH <- 3 in a MUAC<115 mm only programme would also increase the number of patients by about 60%. See again the table derived from 522 surveys collected by Mark (with about 500 000 children in total) I mentioned in a previous post on the breakdown of different categories of SAM children:

MUAC < 115 mm and WHZ < -3 23%
MUAC < 115 and WHZ >-3 39%
MUAC > 115 and WHZ < -3 38%

One could easily increase the MUAC to make it more inclusive, increasing the number of high risk children detected, i.e. increasing the sensitivity of screening, without going beyond the patient numbers that you would get by also including all children with WFH < -3.

Ideally, the MUAC cut off for SAM treatment should be based on risk assessment and response to specific SAM treatment. We know that the risk of death no more decreases beyond 125 mm (Myatt et al., 2006). We have no information on a cut off based on response to treatment, however. Even if children with a MUAC 120-125 mm are at a high risk, this does not mean they need a specific treatment with RUTF, they may well respond to a less expensive food supplement or an adequate diet based on local foods, with mineral and vitamin supplements, if needed. More research is needed on this.

The WHO working document also ends up suggesting that analysing current programmes will help to decide if it is relevant to use WFH<-3 as additional admission criteria or if MUAC <115 mm alone is sufficient. This statement has to be qualified, as this is correct only if SAM children treated are representative of SAM children seen in the community, something very difficult to achieve. For this, you have to make sure that different categories of children are referred to treatment in the same way and have the same treatment seeking behaviour. A real challenge. The Grellety et al. study should be an eye opener in this respect as it is based on a sample from which nearly all high risk low MUAC children were excluded and no conclusion relevant to admission criteria can be obtained from it (see my previous post). In addition to that, data from treated children are difficult to interpret. After a few days, there are few deaths and many of them are unrelated to malnutrition if treatment is correctly implemented.

An additional limitation of the data analysis of current programmes is that most of them are not adapted to analysing the alternative and more attractive option of increasing the MUAC cut off. Current programme do not screen these children and all children who have a MUAC > 115mm and WFH <-3 are censored.

I let you appreciate the relevance of this WHO document in light of these comments. Of note, I fully support Mark’s statement that well known WFH uncritical supporters are overrepresented among the authors of this WHO working document. The lead author himself has even just published an article highlighting that MUAC < 115 mm does not identify correctly children with WFH <-3 (déjà vu hundreds of times !!!; just wondering how this can still be published) and rejects the use of MUAC on this basis (Roberfroid et al., 2015). Of course, MUAC and WFH do not select the same children, else these two indices would be just as good to assess the risk of dying, whereas MUAC is better. Interestingly, the only reference to mortality in this paper is based on the Grellety study, used to claim that WFH <-3 is much better to detect children who are going to die, but failing to mention that the selection process eliminated virtually all high risk children with a MUAC < 115 mm. No mention either of community based studies.

A personal note to end up. My first papers on MUAC and mortality were published in 1986 (Briend and Zimicki, 1986) (Briend et al., 1986), nearly 30 years ago now. Since then, it has been confirmed many times that MUAC is very good to identify high risk children, much better for that than WFH and easy to implement at the community level, even by mothers (Blackwell et al., 2015). I have enough of the ongoing discussion, I think we should stop saying we have not enough evidence for moving to MUAC. I am rather depressed to see this repeated at the end of this WHO working document which fails to mention the evidence in favour MUAC being a better predictor of the risk of death. I am also upset to see this repeated uncritically in this forum, including by the moderator, neglecting the evidence. Those who want more evidence should read the relevant literature, and avoid relying on poor documents censoring the evidence. Else, they should describe the type of additional study they need to be convinced, if ever, it is just too easy to say there is not enough evidence beyond limits. As shown by the data of the Grellety study, a dogmatic attitude in favour of low WFH children eliminates from treatment many high risk children and results in many unnecessary deaths (by the way, the authors of the Grellety study examined data from an ongoing programme, they did not design it as far as I know and should not be held responsible for the selection process which denied treatment to those most at risk and for the excess of deaths which resulted from this). I am afraid the ongoing debate on whether or not to include children with MUAC > 115 mm and WFH <-3 is counterproductive and diverts our attention from the important reconsideration of current MUAC cut-off in relation to response to treatment. And also diverts on attention from the broader and even far more important issue of access to treatment and current low programme coverage. All this leads to more preventable deaths still occurring among high risk children who remain untreated. No wonder quite a few of us have difficulty in keeping calm in this debate which has been going on for too long.

I hope this helps,

André

References

Alam, N., Wojtyniak, B., Rahaman, M.M., 1989. Anthropometric indicators and risk of death. Am. J. Clin. Nutr. 49, 884–888.

Bairagi, R., 1981. On validity of some anthropometric indicators as predictors of mortality. Am. J. Clin. Nutr. 34, 2592–2594.

Blackwell, N., Myatt, M., Allafort-Duverger, T., Balogoun, A., Ibrahim, A., Briend, A., 2015. Mothers Understand And Can do it (MUAC): a comparison of mothers and community health workers determining mid-upper arm circumference in 103 children aged from 6 months to 5 years. Arch. Public Health Arch. Belg. Santé Publique 73, 26. doi:10.1186/s13690-015-0074-z

Briend, A., Dykewicz, C., Graven, K., Mazumder, R.N., Wojtyniak, B., Bennish, M., 1986. Usefulness of nutritional indices and classifications in predicting death of malnourished children. Br. Med. J. Clin. Res. Ed 293, 373–375.

Briend, A., Maire, B., Fontaine, O., Garenne, M., 2012. Mid-upper arm circumference and weight-for-height to identify high-risk malnourished under-five children. Matern. Child. Nutr. 8, 130–133. doi:10.1111/j.1740-8709.2011.00340.x

Briend, A., Zimicki, S., 1986. Validation of arm circumference as an indicator of risk of death in one to four year old children. Nutr. Res. 6, 249–261. doi:10.1016/S0271-5317(86)80129-4

Grellety, E., Krause, L.K., Shams Eldin, M., Porten, K., Isanaka, S., 2015. Comparison of weight-for-height and mid-upper arm circumference (MUAC) in a therapeutic feeding programme in South Sudan: is MUAC alone a sufficient criterion for admission of children at high risk of mortality? Public Health Nutr. 18, 2575–2581. doi:10.1017/S1368980015000737

Isanaka, S., Guesdon, B., Labar, A.S., Hanson, K., Langendorf, C., Grais, R.F., 2015. Comparison of Clinical Characteristics and Treatment Outcomes of Children Selected for Treatment of Severe Acute Malnutrition Using Mid Upper Arm Circumference and/or Weight-for-Height Z-Score. PloS One 10, e0137606. doi:10.1371/journal.pone.0137606

Mwangome, M.K., Fegan, G., Fulford, T., Prentice, A.M., Berkley, J.A., 2012. Mid-upper arm circumference at age of routine infant vaccination to identify infants at elevated risk of death: a retrospective cohort study in the Gambia. Bull. World Health Organ. 90, 887–894. doi:10.2471/BLT.12.109009

Myatt, M., Khara, T., Collins, S., 2006. A review of methods to detect cases of severely malnourished children in the community for their admission into community-based therapeutic care programs. Food Nutr. Bull. 27, S7–23.

Pelletier, D.L., 1994. The relationship between child anthropometry and mortality in developing countries: implications for policy, programs and future research. J. Nutr. 124, 2047S–2081S.

Roberfroid, D., Huybregts, L., Lachat, C., Vrijens, F., Kolsteren, P., Guesdon, B., 2015. Inconsistent diagnosis of acute malnutrition by weight-for-height and mid-upper arm circumference: contributors in 16 cross-sectional surveys from South Sudan, the Philippines, Chad, and Bangladesh. Nutr. J. 14, 86. doi:10.1186/s12937-015-0074-4

Van den Broeck, J., Eeckels, R., Massa, G., 1996. Validity of single-weight measurements to predict current malnutrition and mortality in children. J. Nutr. 126, 113–120.

Vella, V., Tomkins, A., Ndiku, J., Marshal, T., Cortinovis, I., 1994. Anthropometry as a predictor for mortality among Ugandan children, allowing for socio-economic variables. Eur. J. Clin. Nutr. 48, 189–197.

Benjamin,

I find your mail rather confusing, with a presentation of my ideas and Mark’s of your own which I don’t fully recognise as mine. I invite those who want to know my position to read my posts and my own papers and not to rely on second hand comments.

Apparently, you missed an episode. See my previous posts. The Grellety et al paper (1) that you quote in favour of using WFH is based on a sample from which nearly all children with a MUAC < 115 mm and WFH >-3 were excluded. This results in a sample with nearly all children having a WFH <-3. This explains the artificially high proportion of deaths in children with WFH <-3 compared to MUAC < 115 mm. This study based on a highly truncated sample not at all representative of what you see in the community cannot be used to make any statement in favour of adding WFH when selecting children for treatment.

Also, please note that the Grellety et al. paper (and none of the others you quote) does not contribute in any way to discuss the relevance of slightly increasing MUAC to increase its sensitivity, the most appropriate answer to the concern that MUAC < 115 mm could be too restrictive.

Apparently, you missed also that this low proportion of MUAC<115 mm only children in this sample resulted from an attempt to have more “truly malnourished” children, ie with a low WFH by eliminating all young small children. These excluded children have a high risk of death and were denied treatment in the described programme. This presumably resulted in an excess of preventable deaths in the target population. I will now on use the data presented in this paper to argue that excessive confidence in WFH when running programmes can lead to excess deaths.

1 - Grellety E, Krause LK, Shams Eldin M, Porten K, Isanaka S.Comparison of weight-for-height and mid-upper arm circumference (MUAC) in a therapeutic feeding programme in South Sudan: is MUAC alone a sufficient criterion for admission of children at high risk of mortality? Public Health Nutr. 2015 Oct;18(14):2575-81.

We keep coming back to tangentially relevant or misleading evidence:

Givjalva-Eternod et al. (2015) - Interesting but not, I think, directly relevant as the "outcome" measure is of body composition (not mortality). The finding of this paper are at odds with other papers that address this issue. Further work is probably needed although I am not sure if it will be relevant to CMAM programming.

Roberfroid et al. (2015) - Showing that MUAC and WHZ select different children is exactly the point. WHZ perform badly at what we need it to do. MUAC performs better that WHZ at what we need it to do. If MUAC and WHZ selected the same children then MUAC and WHZ would perform equally well as what we need then to do. I avocate for the use of MUAC exactly because because it is different from WHZ (which does very badly at what we need it to do). Please point me to the post where I misrepresent the Roberfroid et al. (2015) paper. I will check and, if needed, write a correcting note.

Grellety et al. (2015) - This has been discussed at length above. This is not appropriate or useful evidence.

Isanaka et al. (2015) - Here we have evidence that children with low MUAC respond to treatment. This is supports using MUAC for admission. We already knew this. The article does not address the issue of mortality risk in untreated cohorts which is what we need to decide whether the discordant MUAC-/WHZ+ children need the full CMAM protocol or could be treated in (e.g.) TSFP type programs. Clinical studies miss the point somewhat (see above re Grelley et  al. (2015)).

It does seem that we need more evidence (although I fear that it will be ignored by those with a "dogmatic attitude in favour of low WFH children"). What you have presented above is not the evidence we need. At best these articles present tangentially relevant evidence. Grellety el al. (2015) is so inappropriate and biased as to be misleading. Your "Further investigations ... " suggest more of the same pseudo-evidence.

I am not sure where the evidence for some of the statements about MUAC and WHZ comes from. We know from the CTC research program that (e.g.) low MUAC kids in SFP die at five or six times the the rate as the same kids in OTP. This was a key piece of evidence for allowing MUAC to be used in OTP programs. I cannot understand the logic behind classifying the identification and admission of children who need and can benefit from the program as over-diagnosis. This is the sort of "logic" that gives rise to the sort of programming described in Grellety et al. (2015).

Please ... rather than keep producing and flaunting tangentially relevant or misleading evidence why not start collecting so useful and appropriate evidence?

Sorry ... see here for a discussion of the Grellety et al. (2015) study.

Dear MA Chowdhury,

You are concerned that MUAC selects preferentially too many young children. At this stage, please note that the main argument in favour of MUAC is its ability to select high risk children. So the question you ask should be translated into:

Should we correct MUAC for age (and sex) to select children with a high risk of death?

This issue was examined for the first time in a short paper from R Bairagi published ion 1982 who showed that the ROC curves of MUAC corrected and uncorrected for age were nearly identical. I re-examined this issue in my first paper on MUAC in 1986 using age as a continuous variable and also found that adding age (and height) in logistic models did not improve the assessment of the risk of death.

These results were then forgotten, and a similar study was published in 2012 by a Danish group who were not aware of our results. And found the same thing. Correcting MUAC for age or sex (with the then just published WHO growth standards for MUAC) did not improve the assessment of the risk of death. We can describe this as an example of scientific amnesia whereby the same discovery is made several times independently by different groups which ignore each other.

The bias towards younger children makes sense as well from the nutritional point of view (see Briend et al, 2015). When energy and nutrient intake is inadequate, children leave from their fat and muscles. Muscle seems critical for survival in case of infection. And in relation to body weight, young children have a low muscle mass and are presumably more vulnerable.

This increased vulnerability of young children may also explain the poor performance of WFH who is an index based on deviation from a standard, ie a variable obtained from a statistical analysis without any connection to function or pathophysiology.

Measuring muscle mass is nearly impossible in young children, and I am not aware of a paper showing how it grows in normal children. Avaialbe data come from a few post mortem studies. The paper from Ethiopia quoted in Benjamin’s post measured lean body mass, which is completely different and does not have the same nutritional meaning. Its interpretation is difficult and the relevance of this paper for this discussion very low.

Bairagi, R., 1981. On validity of some anthropometric indicators as predictors of mortality. Am. J. Clin. Nutr. 34, 2592–2594.

Briend, A., Zimicki, S., 1986. Validation of arm circumference as an indicator of risk of death in one to four year old children. Nutr. Res. 6, 249–261. doi:10.1016/S0271-5317(86)80129-4

Briend A, Khara T, Dolan C. Wasting and stunting--similarities and differences: policy and programmatic implications. Food Nutr Bull. 2015 Mar;36(1 Suppl):S15-23.

Grijalva-Eternod CS, Wells JC, Girma T, Kæstel P, Admassu B, Friis H, Andersen GS.
Midupper arm circumference and weight-for-length z scores have different associations with body composition: evidence from a cohort of Ethiopian infants.
Am J Clin Nutr. 2015 Sep;102(3):593-9. doi: 10.3945/ajcn.114.106419. Epub 2015 Jul 29.

J Rasmussen, A Andersen, A B Fisker, H Ravn, M Sodemann, A Rodrigues, C S Benn and P Aaby. Mid-upper-arm-circumference and mid-upper-arm circumference z-score: the best predictor of mortality? European Journal of Clinical Nutrition 66, 998-1003 (September 2012)

Dear MAChoudhry,

Thank your for your thoughtful post.

The issue of age-independence and MUAC does cause confusion. We use MUAC because if is a good predictor of mortality that can be averted with appropriate nutritional and clinical support. The age-independence of MUAC is assessed with regard to perfromance at identifying these children. If you see MUAC selecting a lot of young male children it is because these children are at high mortality risk. You'l also see MUAC pick up the young stunted children. This is also because these children are at high mortality risk. We know that these children can be treated successfully with the CTC/CMAM protocol. Follow-up studies show good post-discharge survival (often as good or better than never-SAM children).

A MUAC-only program does not "exclude" older children. It will admit older children with low MUAC (who will be at high risk of death and can be treats successfully). It does exclude older children who are at lesser mortality risk. See above for a critique of the program reported in Grelletty et al. (2015). This program decided to exclude young children who were both stunted and wasted and at high mortality risk.

WRT increasing MUAC thesholds ... Yes. This is a better approach than including low WHZ but high MUAC children. Note that we have already done this once (from 110 mm to 115 mm) around 2009. I think this is a far better option than introducing or retaining WHZ. I think a key questions is at what MUAC threshold can we safely replace high intensity care (i.e. OTP) with lower intensity care (w.g. TSFP). At present we have this (almost everywhere at 115 mm). This may be the right threshold. This is (IMO) a useful avenue for further research now that we have good TSFP products.

The proportions were from 552 SMART surveys from 48 countries. These are not perfectly representative but very few surveys are perfectly representative. I think the differences between the figures from the 552 surveys and the Grellety et al. (2015) study are so large as to rule that study out as useful and appropriate evidence. Clinic base studies are of little help with these questions.

We expect MUAC-only programs to better reflect proportions in the population because they tend to have higher coverage due, in large part, to the simplicity and low cost of achieving high spatial and temporal coverage of case-finding activities. I think it is possible for a WHZ program to achieve high coverage but it will be a very expensive program with poor targeting of resources. I have done many coverage assessments and a consistent finding is that WHZ (for a number of reasons)( damages program coverage.

I am not supporting MUAC just because it better targets resources at those in need. It also enables better coverage so that program treat more children at high risk. I think appropriate targeting, high coverage, and good programmatic efficacy are what make for effective programming. I fear that this is missed in much of the MUAC vs. WHZ discussions.

WRT "which children are actually wasted?" ... I don't think I really care much about this. Some people (mostly in then emergency and development nutrition field) will only define wasting as low WHZ. In a clinical context wasting is often defined as loss of peripheral muscle and fat mass (which is low MUAC). My main focus is, however, on child survival so I want an indicator that is a good predictor of mortality that can be averted with appropriate nutritional and clinical support. I suppose that if I were interested in treating children with high body surface area to mass ratios compared to a specially selected reference population then I would push for us to be using WHZ.

WRT additional benefits over survival. That may be the case but any model of hierarchy of needs will have "being alive" at the base of the pyramid. To be blunt ... dead children never grow up and never grow up to be great composers. I see CMAM as primarily a child-survival program and this guised my choices.

I hope this helps.

Benjamin

I am not sure that the evidence you present supports any of these hypotheses. The clinical data is (IMO) not very relevant and the Robertfroid et al. (2015) review was highly selective in the evidence presented to the extent that borders (IMO) on fraud. This "evidence" challenges little because it does not present useful and appropriate evidence (see the discussion in this thread). I think we may, in fact, be talking about two different things. Your questions may be very different from mine. Your evidence does not address my questions.

Young and stunted children without acute nutritional needs would not be picked up by MUAC as you suggest. The children that would be picked up would by young, stunted, and wasted. These children have multiple nutritional deficits and are at heightened mortality risk. When we recruit these children into OTP programs they exhibit rapid gains in both weight and MUAC usually following a typical growth curve (i.e. very rapid gains in the first few weeks slowing and approaching an asymptote). When we don't recruit them into OTP and put them into SFP they do not do well (i.e. they die at a much greater rate in SFP than in OTP). This is presented in Myatt et al. (2006). That paper used the case-definitions of the time. It is not a properly designed study (as is usual with natural experiments) but the finding that small, young, low MUAC kids benefit from therapeutic feeding is clear. The practice, as in the program reported bt Grellety et a. (2015), of excluding these children is dangerous and likely to lead to avoidable deaths.

BTW : I cannot look at the natural experiment data and apply the new MUAC case-definition as the study population was children with MUAC < 110 mm. I can confirm that all had a WHZ > 70% (NCHS). The mortality rates (cumulative incidence) were:

    In OTP : 0% (no deaths observed in 40 cases)
    In SFP : 14.8% (8 deaths observed in 54 cases)

The 95% CI on the risk difference was 3.15% to 26.47% (i.e. putting these kids in SFP was clearly a bad idea).

Note how times change. Ten years ago the discordancy was MUAC+/WHZ-. A small study ensured that these kids got what they needed (although the height bar, as applied in Grellety et al. (2015), has undermined this by renewed exclusion of these children without supporting evidence).

We could do a similar (properly designed) trial ... put the MUAC-/WHZ+ kids into SFP and see what happens to them. We'd want to have close monitoring. The Bangladesh trial suggest that they will do well in SFP. I think this is the best way to be sure that our programs deliver the appropriate level of care to these children. I'd be happy to work on such a trial in any capacity.

I have access to a great deal of data. Some is my own data but much has been given / loaned to me for specific analyses. I am sure that I can obtain permissions to use or share much of this data for different analyses. You will understand that I cannot do this without permission from the data owners. I think we should start by working out what our research questions on this topic should be (you seem to have a very different set of questions to me but I amy have several blind spots). We can then see what can be answered by data in hand (and try to do that) and what will need new data. I am sceptical that clinical case-series will be useful. I think that a cohort study looking at what happens to the MUAC-/WHZ+ kids with low intensity treatment might be useful. I do not have data on that.

I have started a thread here.

I hope that we can come together there to work out what we need to do to move forward with an informed debate on issues around MUAC-only programming (a wider topic than has been debated here) and the appropriate treatment of children with MUAC > 115 mm and WHZ < -3.

See you there!

We seem to have moved away from legitimate concerns about the discordant (i.e. MUAC - / WHZ+) cases towards a hand-waving attempt at justifying for denial of treatment to young and small children with low MUAC at very high risk of death if left untreated.

Returning to the issue of discordant (i.e. MUAC - / WHZ+) cases ...

I looked again at Figure 1 and Table 2 in Grellety et al. (2015). I worry that the analysis here (IMO) neglects the denominator.

The table in Figure 1 shows:

    Cumulative incidence of death =  27 / 719 = 3.76% (95% CI = 2.49%; 5.42%)	in MUAC +
    Cumulative incidence of death = 13 / 1486 = 0.87% (95% CI = 0.47%; 1.49%)	in MUAC - / WHZ +

The RR is 4.29 (95% CI = 2.23; 8.27).

It is clear that the mortality risk in the MUAC + group is considerable higher than in the MUAC - / WHZ + group.

We could take a person-days at risk / incidence density approach (data from Figure 1 and Table 2):

    MUAC

        27 deaths per 719 persons per 51.57 days
	(27 / 51.57) deaths per 719 persons per day
	(27 / 51.57) * (10000 / 719) per 10000 per day
	7.28 / 10000 / day (95% CI = 4.81 to 10.29 / 10000 / day)

    MUAC - / WHZ +

        13 deaths / 1486 persons / 50.82 days
        (13 / 50.82) / 1486 persons / day
        (13 / 50.82) * (10000 / 1486) / 10000 / day
        1.72 / 10000 / day (95% CI = 0.92 to 2.78 / 10000 / day)

The IRR is 4.33 (95% CI = 2.18; 8.20). This is essentially the same analysis as above as there was little difference in terms of length of stay between the two groups.

Someone should check my calculations.

We cannot generalise much from this sample or similar samples for many reasons (see the discussions above) but in this program the MUAC + group were at very much higher mortality risk than the MUAC - / WHZ +. Since the program excluded many young and stunted and low MUAC kids using an (IMO) unwarranted height bar we can expect the observed c. 4 fold increase in risk to be an underestimate.

I did the analysis above because I was interested to see mortality rates in the two groups.

Those who promote the continued use of WHZ in CMAM programs are concerned about the 1.72 / 10,000 / day mortality rate in the MUAC - / WHZ + group. That is an elevated (i.e. above acceptable) mortality rate if observed in the general population. We are not, however, looking at a general population ... 0.87% mortality in an OTP patient group is exceptionally low. The 95% CI suggests that what we see in this this group in this program is not statistically significantly different from 1 / 10,0000 / day.

I think we should be careful with this evidence and not be quick to dismiss concerns about here being real need in the MUAC - / WHZ + children. I think we can conclude that there is need but that need may be different. The MUAC - / WHZ + may not require the high intensity treatment provided by OTP and do very well with a lower intensity therapy. I think that we need to concentrate on this question.

Also looking at Figure 1 ... in MUAC in the MUAC - / WHZ + tends to be low (most are below 125 mm). This suggest (to me) that MUAC only programming with a MUAC < 125 mm admission with two clinical pathways (i.e. MUAC < 115 mm get OTP, MUAC between 115 and 125 mm get something less intensive) would (1) address the concerns of WHZ proponents, (2) retain the advantages of MUAC only programming, and (3) contol costs and crowding (if designed properly).

Benjamin,

When you assess growth of low MUAC children receiving RUTF, the comparison group should be children not receiving treatment. Comparison with low WFH children also receiving RUTF is not really relevant. Also, note that measuring weight gain in relation to initial WFH will show an artificially high effect, as WFH is strongly correlated with initial weight which is used to calculate weight gain. You can play with computer generated random numbers to see this bias at work.

Risk of deaths and sensitivity of different indicators to assess the risk of deaths should be compared at similar specificity level. Adding children with a WFH < -3 to a MUAC <115 mm only programme will result in a 60% increase of patient numbers. Mark just did a quick calculation on the 552 surveys and found that this increase in patient number is roughly equivalent to what you would get by increasing MUAC cut off around 120 mm. Number of deaths observed in children with WFH <-3 should be compared with number of deaths of children less than 120 mm MUAC if we want to compare screening schemes selecting about the same number of children.

I think you will have a hard time to convince any one with a few notions of epidemiology that community based studies represent lower quality evidence compared to hospital based studies when discussing screening at the population level. And this is what you do in the WHO document and again repeatedly in this forum. And you and Roberforid in his last 2015 paper put forward the Grellety et al study, which is an embarrassment to the humanitarian community as its table shows clearly that it is based on a programme which left the most high risk children untreated. You should stop doing that.

And please kindly stop quoting the 2015 Roberfroid paper to reject MUAC as it does not agree with MUAC. This has been said again and again and is not an accepted argument. Or quote this paper, as I will do, to argue that MUAC selects children who tend to be younger, and more stunted, and hence more at risk compared to low WFH children.

I wonder how useful it is to continue this discussion.

See the discussion above. I have engaged with your "evidence" and your "facts" and find them lacking relevance and intellectual rigour.

I reviewed some of the evidence in one of your "key exhibits" again just this morning and posted a message looking at the mortality claims being made.

Come back with some relevant facts and relevant evidence and I will take you seriously.

As for my fear of evidence ... you seem to have missed my post here. I want evidence of direct rather than tangential relevance. Why not join in? I am certain I do not have all the answers. I am also equally certain that you do not have all the answers.

I think it is time for us to stop bickering and move forward on these important issues.