Routinely MUAC screening for severe acute malnutrition should consider the gender and age group bias in the Ethiopian non-emergency context.

Early identification of children <5 years with severe acute malnutrition (SAM) is a high priority to reduce child mortality and improved health outcomes. Current WHO guidelines for community screening for SAM recommend a Mid-Upper-Arm Circumference (MUAC) of less than 115 mm to identify children with SAM, but this cut-off does not identify a significant number of children with a weight-for-height Z-score <-3. To establish new specific MUAC cut-offs, pooled data was obtained for 25,755 children from 49 SMART recent surveys in Ethiopia (2016-2019). Sensitivity, proportion of false positive, and areas under receiver-operator characteristic curves (AUC) were calculated. MUAC below 115mm alone identified 55% of children with SAM identified with both methodologies. MUAC was worse in identifying older children (21%), those from a pastoral region (42%) and boys (41%). Using current WHO cut-offs, the sensitivity (Se) of MUAC below 115mm to identify the children severly malnourished screened through Weight-for-height below-3 was 16%. Analysing the ROC curve and Youden Index, Se and Specificity (Sp) were maximal at a MUAC < 133 mm cut-off to identify SAM (respectively Se 61.1%, Sp 81.4%). However, given the high proportion of false-positive children, according to gender, region and age groups, a cut-off around 125 mm to screen SAM could be the optimal one. In Ethiopia, implementation of a MUAC-only screening program for the identification of severe acute malnutrition with the actual cut-off of 115 mm would be unethical as it will lead to many children remaining undiagnosed and untreated. In addition, future study on implementation challenge on screening children with a higher cut-off or gender/age sensitive ones should be assessed with the collection of mortality and morbidity data to ensure that the most in need are being taking care of.

Introduction

Ethiopia has experienced a rapid improvement on several nutrition indicators over the past fifteen years. However, the expected decrease in the number of stunted children from 5.85 to 5.11 million in 2025, corresponding to an average annual reduction rate (AARR) of 2.26%, is far from the projected AARR of 4.75%, needed to reach the 2025 World Health Assemble target [1]. In addition, over the last fifteen year, the prevalence of acute malnutrition has almost not changed (from 12% in 2000 to 10% in 2016) [2]. According to the 2019 Ethiopian Humanitarian Response Plan (HRP), almost two million children will be affected by acute malnutrition and will require assistance.

In Ethiopia, severe acute malnutrition (SAM), i.e. Mid Upper Arm Circumference (MUAC) <115 mm or weight-for-heigh z-score (WHZ) < -3 or nutritional oedema, affects approximately 1 million children under the age of five. Acute malnutrition places a child at a great risk for death and increases the risk of morbidity and stunting [3], therefore it is essential to prevent and treat acute malnutrition at an early stage. However, the 2025 World Health Assembly targets for Ethiopia will be challenging to achieve if current trends continue.

In the latest 2013 guideline for the management of severe acute malnutrition, the World Health Organization (WHO) recommends using Mid Upper Arm Circumference (MUAC) at community level to screen for SAM [4]. A recent meta-analysis published in 2018 [5] supports this recommendation that communities through different platforms could perform routine screenings of young children with MUAC and improve thereby detection of SAM cases. However, over the last years, several papers have highlighted that the current practice to identify children with acute malnutrition, at least in a non-emergency context, can be improved considerably by increasing the current WHO recommended cut-offs for MUAC [6-9]. In Cambodia [6] and India [9], the authors suggest that a cut-off towards 12.5cm to 12.8cm would increase the efficiency to screen children with severe acute malnutrition (SAM). Such change will allow to find with MUAC alone 50% of all the children with a limited proportion of false positive (<10%).

The present study aimed to answer several research questions posed in the revised WHO guidelines for the screening of severely malnourished children, especially focusing on the discrepancy between MUAC and WHZ, defining optimal MUAC cut-offs for routine system in non-emergency settings adapted to the Ethiopian context, and the influence of age and gender for detecting SAM in children under 5 years of age.

Methods

We used pooled data from several SMART surveys carried out by the Government of Ethiopia and UNICEF, covering 4 years (2016–2019). Data on weight, height, MUAC, gender and age were available for a total of 26,806 Ethiopian children, from SMART surveys conducted in 49 different woredas/districts from 6 provinces of Ethiopia between 2016 and 2019. A two-stage cluster sampling approach based on a population sampling frame probability proportion to size (PPS) of all smallest rural villages was used as per SMART survey methodology. The sample size for anthropometric and retrospective mortality survey was determined by entering expected prevalence of malnutrition and mortality rate, desired precision, design effect, percentage of children under five, average household size and per cent of non-responsive households. Data quality was maintained by provision of standardization test and field practice during training, field level supervision by survey coordinators during data collection, daily checking and exchanging feedback on completed questionnaires and measurement error. The plausibility check in the ENA ensured that the quality of these data was within the expected quality standard.

All parents of participants gave their written consent. The target population was children aged 6–59 months. Children’s height, measured to the nearest 1 mm (UNICEF measurement boards) were collected from each child. Weight was measured to the nearest 0.1 kg (Seca, Hamburg, Germany), with children wearing only light clothes. The nutritional status of children was defined using MUAC, length or height-for-age (L/HAZ), WHZ, and weight-for-age (WAZ) z-scores, calculated according to the Child Growth Standard of the WHO [10] using the WHO Anthro software. To ensure the accuracy of the data, extreme values were excluded from the analysis: WAZ < −6 or > 5; L/HAZ < −6 or >5; WHZ < −6 or > 5. Excluded values represented less than 4% of the total values (see Fig 1). The quality of the survey was also assessed through the z-score of WHZ (-0.615±1.048) as recommended by Grellety and al [11].

Fig 1

Sample used for the analysis.

We used the following cut-offs to define acute malnutrition or wasting: WHZ < −2 Z-scores (according to the WHO growth charts) or MUAC < 125 mm, and severe acute malnutrition (SAM) was defined as: WHZ < −3 Z-scores or MUAC < 115 mm. Stunting was defined as a height-for-age < −2 Z-scores. The child’s age in months was calculated by subtracting the date of the visit from the date of birth of the child; results were used as a continuous variable. Gender was considered as a binary variable.

We also investigated the potential bias of MUAC and WHZ as indicators for severe acute malnutrition. For SAM, a new variable with three categories was created: those with a MUAC<115 mm but a WHZ>-3; children with a WHZ<-3, but a MUAC>115mm; children with a WHZ<-3, and a MUAC<115mm with non-wasted children (WHZ>-3 and MUAC>115mm) used as reference category. For global acute malnutrition, similar categories were created using WHZ<-2 and MUAC<125mm. This variable was then used as a response variable to calculate sensitivity of each indicator to identify children with acute malnutrition or SAM.

To assess the performance of different MUAC cut-offs compared to the current cut-off recommended by WHO to detect severe and moderate acute malnutrition, receiver operating characteristic curves (ROC curves) were constructed. The sensitivity and proportion of false-positive (1-specificity) of MUAC were determined using wasting (WHZ<-2 z-score in children under 5 y) as acute malnutrition is the combined case load of low MUAC and/or low WHZ. The ROC curve is the plot of sensitivity versus proportion of false-positive of MUAC cut-offs. The area under curve (AUC) is the area between the curve and the segment (0,0) and (1,1), which corresponds to a random classifier. A larger AUC indicates a more accurate diagnosis of acute malnutrition defined by WHZ cut-offs [12]. Data analysis was performed using SPSS version 20.0 (SPSS, Inc., Chicago, IL).

To evaluate the performance of our analysis, the corresponding Youden index, which is the sum of sensitivity and specificity minus one, and reflects the overall capacity of an early warning model to detect a disease, was calculated: 1 indicating a perfect test, and 0 an imperfect test [13]. These analyses were conducted overall, and by gender, regions (agrarian versus pastoral) and age groups. Pastoralists are communities whose primary livelihood activity is livestock production/sales; whereas agrarian are those whose primary livelihood activity is crop production/sales, complemented with livestock production/sales. We used three criteria to which any new cut-off should adhere: i) Higher Youden index than the present cut-off; ii) Accuracy between 0.8 and 1; and iii) A proportion of false positive below 1/5 of non-malnourished children. The latter criteria is to prevent overburdening of health centres with too many false-positive cases of SAM. Calculations were done for all children overall, as well as for specific age groups, for girls, boys and region.

Results

From the total of 25,755 children for whom data was available, 51% were male and 70.4% living in Agrarian zone. Mean (±SD) age was 31.7(±15.2) months (44% of children were less than 2 years old, 66% between 2 and 5 years old). Mean (±SD) MUAC was 141(±12) mm and ranged between 90 mm to 234 mm. Overall, 9.4% and 8.3% of the children were identified as GAM by WHZ<-2SD or MUAC<125mm respectively, while the prevalence of SAM (MUAC<115 mm or WHZ<-3) was identical for both indicators at 1.3%. The proportion of children indentified by both indicators was 19.2% for GAM and only 8.9% for SAM (Fig 2). Using MUAC-only identified 55.7% of the acute malnourished children and 55.2% of the severely malnourished children (Fig 2).

Fig 2

Pie charts showing the proportion of children with GAM and SAM diagnosed by both MUAC < 125 mm and WHZ < -2SD (green) or by MUAC alone (yellow) or by WHZ alone (blue).

Subgroup analyses revealed large differences in the degree by which children were diagnosed by MUAC-only, ranging from 29.4% to 81.4% for GAM and from 21.4% to 75.1% for SAM, depending on the gender, region or age groups (Table 1). Less than half of malnourished boys and less than half of children living in the pastoral region were identified by MUAC. And MUAC identified less than ¼ of older children with SAM.

Table 1

Identification of global and severe acute malnutrition by weight-for height Z-score, mid-upper arm circumference, or by both criteria according to gender, age-group and age.

	GAM subject	WHZ<-2 only	MUAC<12.5 only	Both criteria	MUAC diagnosis	SAM subject	WHZ<-3 only	MUAC<11.5 only	Both criteria	MUAC diagnosis
Gender
Boy	1,891	52.9	26.5	20.6	47.1	316	59.5	30.4	10.1	40.5
Girl	1,945	35.8	46.3	17.9	64.2	304	32.6	59.9	7.6	67.5
Age-group
6-23mo	1,944	18.5	54.8	26.6	81.4	373	24.9	61.4	13.7	75.1
24-59mo	1,892	70.7	17.8	11.6	29.4	247	78.5	19.8	1.6	21.4
region
Pastoral	1,392	57.7	24.8	17.5	42.3	246	58.5	31.3	10.2	41.5
Agrarian	2,444	36.6	43.2	20.2	63.4	374	38.2	53.7	8	61.7

ROC curves (Fig 3) showed optimal MUAC cut-offs of 138 mm and 131 mm to detect also acute malnutrition and SAM identified by a low WHZ respectively. When MUAC <131 mm is used to identify SAM, the sensitivity and specificity for our population surveyed reached respectively 61.1% and 81.7% with an accuracy of 0.814.

Fig 3

ROC curve of the MUAC score against WHZ<-2SD (I) and WHZ<-3SD (II).

In Table 2, other MUAC cut-offs for detecting SAM are presented with consideration for other parameters than just the AOC (false positive, accuracy and Youden index). Increasing the MUAC to screen children for SAM from 115mm to 125mm could improve the sensitivity of the test by 2 to 10 times, depending on the sub-group assessed without overburdening the health post with too many false positive cases (less than 18%). Promoting a higher cut-off such as 131mm would continue to increase the sensitivity but the specificity would reduce significantly. For example, almost a fifth of the children screened will be false positive with a MUAC cut-off of 131 mm, while it is less than 8% for a cut-off of 125 mm. With the suggested cut-off of 125 mm, almost 30% more children would have been diagnosed as severely acute malnourished. For our complete data set, the number of children diagnosed with SAM would have increased 333 to 430 children.

Table 2

Evaluation of screening test of nutritional status by different cut-offs of MUAC and WHZ (to detect severe acute malnutrition) in children aged 6–59 months.

	Sensitivity (%)	False Positive (%)	Accurarcy	Youden Index	Difference with highest Youden index (%)
all sample
MUAC<11.5cm for WHZ<-3 (current practices)	16.1	1	0.978	0.15	-0.28
MUAC<12.5cm for WHZ<-3 (most practical)	44.4	7.8	0.915	0.37	-0.06
MUAC<12.7cm for WHZ<-3 (Considering other parameters)	52.6	11	0.885	0.4	-0.03
MUAC<13.1cm for WHZ<-3 (optimal AUC)	61.1	18.3	0.814	0.43	0
Gender
boys
MUAC<11.5cm for WHZ<-3 (current practices)	14.5	0.7	0.978	0.14	-0.36
MUAC<12.5cm for WHZ<-3 (most practical)	45.9	6.2	0.93	0.4	-0.1
MUAC<13.1cm for WHZ<-3 (optimal AUC for all)	65.9	16.2	0.835	0.5	0
girls
MUAC<11.5cm for WHZ<-3 (current practices)	8.6	1.5	0.97	0.07	-0.25
MUAC<12.5cm for WHZ<-3 (most practical)	28.2	9.5	0.894	0.19	-0.13
MUAC<13.1cm for WHZ<-3 (optimal AUC for all)	48.2	20.3	0.791	0.28	-0.04
Age group
6–23 months
MUAC<11.5cm for WHZ<-3 (current practices)	35.4	2.7	0.967	0.33	-0.28
MUAC<12.5cm for WHZ<-3 (most practical)	77.8	17.2	0.827	0.61	0
MUAC<13.1cm for WHZ<-3 (optimal AUC for all)	84	25.9	0.644	0.58	-0.03
24–59 months
MUAC<11.5cm for WHZ<-3 (current practices)	2	0.3	0.986	0.02	-0.4
MUAC<12.5cm for WHZ<-3 (most practical)	20.2	3.1	0.961	0.17	-0.25
MUAC<13.1cm for WHZ<-3 (optimal AUC for all)	44.4	9.5	0.9	0.35	-0.07
Region
Pastoral
MUAC<11.5cm for WHZ<-3 (current practices)	14.8	1	0.971	0.14	-0.3
MUAC<12.5cm for WHZ<-3 (most practical)	41.4	7	0.919	0.34	-0.1
MUAC<13.1cm for WHZ<-3 (optimal AUC for all)	61.5	17.2	0.824	0.44	0
Agrarian
MUAC<11.5cm for WHZ<-3 (current practices)	17.3	1.1	0.988	0.16	-0.27
MUAC<12.5cm for WHZ<-3 (most practical)	47.4	8.2	0.914	0.39	-0.04
MUAC<13.1cm for WHZ<-3 (optimal AUC for all)	60.7	20.8	0.81	0.4	-0.03

Discussion

It is accepted and highlighted in the WHO guidelines [14] that MUAC and WHZ indicators identify different categories of malnutrition and therefore different groups of acutely malnourished children. The present study confirms an earlier study from Southern Ethiopia [15] and studies in several other countries [9] that MUAC and WHZ identify different sub-sets of children. Also, our large sample from Ethiopia is highlighting similar findings from Tadess et al. [14] in that MUAC characterised a larger proportion of girls and young children as severely malnourished as compared to WHZ. In our study, about 60% of the boys and 79% of the older children failed to be identified as SAM through a routine community screening using current cut-offs. Those biases have been already described for Cambodia [7,16]. It is not surprising that MUAC is associated with age, as the current MUAC cut-off is age independent, while it is known that MUAC increases with age [16]. Hence, the older the child is, the less likely s/he will be detected as having SAM by a low MUAC. This bias of MUAC to detect younger children but also shorter ones was also already highlighted by other recent publication [17-18]. In most case, those children were considered at a higher risk of death [19-23] and responding well to treatment [9, 24–25]. However, according to a recent meta-analysis, there is increased risk of death with low WHZ as well [26].

Interestingly, a strong difference was also observed in our study between Pastoral and Agrarian regions, as MUAC failed to identify more children from Pastoral region (59%) than from Agrarian ones (38%). This might be due to body shape difference between regions measured by sitting-to-standing height ratio (SSR) as described by different authors [9,27] and even between different weather exposure [28-29]. In south Sudan, the SSR of patoralist population than settled one like agrarian population [9]. Difference in SSR might influence the diagnosis by WHZ independently of MUAC [9,27]. Therefore MUAC and WHZ are associated with different aspects of body composition, and therefore identify different groups of children with malnutrition [9].

Age-dependent or perhaps even gender/region specific MUAC cut-offs would be more appropriate, however, it would need the development of such MUAC and additional training of the field. MUAC is clearly an easier and quicker screening tool at community level than weight-for-height. Mothers and health volunteers have shown their capacity of detecting SAM [5] in several countries of Africa with good sensitivity and accuracy [30,31]. Acute malnutrition screening should not be reduced to mortality risk screening but also to prevent morbidity, impaired physical, cognitive development, and associated micronutrient deficiencies [32]. Therefore, the cut-off of 115 mm is incorrect as single criteria to screen for severe acute malnutrition, as most children with a WHZ<-3 are missed. Increasing the MUAC cut-off for screening at 131mm as estimated by the ROC curve would increase significantly the sensitivity (from 16% to 61%), but at the same time, almost 26% of young children would be tested false positive for severe acute malnutrition. Acknowledging the potential and valid criticism of placing equal importance on Sensitivity and Specificity, a cut-off at 125mm seems the optimal one, as less than 8% of the screened children would be false positive and the sensitivity would move towards 44%, with an accuracy 91%. However, the suggested cut-off point for screening should further be studied in real practice or implementation, including measuring the additional burden to health system and cost-effectiveness in Ethiopia. It is important to start adcocating for targeted interventions to prevent wasting. Those false positive population could be one of them as they are towards the thine line of being considered as acute malnourished.

To avoid overburdening the Ethiopian health system with false-positive cases of severe acute malnutrition, we suggest including two MUAC cut-offs for different purposes. The first at MUAC<125mm could be used at community level to ensure inclusion of as many children with SAM as possible. This could identify, according to our data set, over 69% of the children with a WHZ<-3. Then, as a second step, all children with a MUAC below the screening cut-off (e.g., 125 mm) should be assessed at a health facility for weight, height and MUAC measurements following the WHO cut-offs which initiate treatment for severe acute malnutrition if SAM is confirmed and if not could initiate appropriate counselling or other programs to prevent future acute malnutrition problem. This approach may improve the cost-effectiveness of the screening programs and the treatment as the sensitivity will be significantly improved. Similar systems in other countries were proposed in other research [6,33].

Any gender analysis of acute malnutrition using the Ethiopian Demographic Health Survey (EDHS) should not undervalue the bias highlighted in this analysis. The EDHS only uses the weight-for-height methodology and therefore this would not identify the 46% of the children with a low MUAC-only, if the same trends are observed as in our study.

Limitations of the study

A limit of the current study is that data used in this analysis was from none nationally represented surveys and therefore that the population screened by the different methods and the new MUAC cut-offs presented here are only adapted to the children assessed through those 3 years SMART surveys. Hence, the aim of the present study is not to suggest using the presented optimal cut-offs as international reference but as possibility for community screening in Ethiopia. However, the study highlights that in Ethiopia by using the actual cut-off of 115 mm results in almost 50% of children with SAM not receiving the vital treatment and/or preventing measures needed. Therefore, we recommend additional analysis in Ethiopia to develop a context wise community screening process to ensure that at least 80% of the children in need are correctly identified and treated.

Conclusion

To conclude, the present study showed the ability of MUAC and WHZ to identify children with severe acute malnutrition. Both indicators showed gender, region and age bias. To ensure that no child with severe acute malnutrition is left without proper treatment and follow-up, a step-wise approach should be defined using MUAC<125mm used for community screening purpose and even the development of gender and age sensitive MUAC. However, further implementation studies need to better understand the health burden, the impact on morbidity/mortality and cost of additional screening.

(XLSM)

Click here for additional data file.

29 Nov 2019

PONE-D-19-28795

Routinely MUAC screening for severe acute malnutrition should consider the gender and age group bias in the Ethiopian non-emergency context

PLOS ONE

Dear Dr Laillou,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

As the two authors clearly reported MUAC and WHZ identify different sub-sets of children with SAM and the two measurement should be regarded as independent from each other. So how it would be possible to assess the validity of MUAC based on WHZ cutoff values? Can WHZ be considered as a gold standard measurement for validating MUAC cutoff values? As provided below, the same serious concern has also been raised by both of the reviewers.

It is not clear how the authors managed the data of children with nutritional oedema. Did you exclude or retained them in the analysis? I recommend to exclude them because, from practical perspectives, such children are automatically considered as SAM cases irrespective of their MUAC or WHZ measurements. Further, retaining them in the analysis may bias (overestimate) the WFH measurement of the children.

The authors pooled data from several SMART surveys but said noting about the quality of arthrometric data of the original surveys. Did you exclude any SMART survey from the analysis due to quality concern? What were the quality assurance approaches employed by the individual surveys?

The recommendation of the second author for validating MUAC against Age- and Sex-specify MUAC cutoffs should also be considered.

We would appreciate receiving your revised manuscript by Jan 13 2020 11:59PM. When you are ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter.

To enhance the reproducibility of your results, we recommend that if applicable you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). This letter should be uploaded as separate file and labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. This file should be uploaded as separate file and labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. This file should be uploaded as separate file and labeled 'Manuscript'.

Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.

We look forward to receiving your revised manuscript.

Kind regards,

Samson Gebremedhin, PhD

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at http://www.journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and http://www.journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. PLOS requires an ORCID iD for the corresponding author in Editorial Manager on papers submitted after December 6th, 2016. Please ensure that you have an ORCID iD and that it is validated in Editorial Manager. To do this, go to ‘Update my Information’ (in the upper left-hand corner of the main menu), and click on the Fetch/Validate link next to the ORCID field. This will take you to the ORCID site and allow you to create a new iD or authenticate a pre-existing iD in Editorial Manager. Please see the following video for instructions on linking an ORCID iD to your Editorial Manager account: https://www.youtube.com/watch?v=_xcclfuvtxQ

3. Please include additional information regarding the parental consent obtained for this study. For example, were the parents well informed prior to consenting?

4. We note that you have indicated that data from this study are available upon request. PLOS only allows data to be available upon request if there are legal or ethical restrictions on sharing data publicly. For information on unacceptable data access restrictions, please see http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions.

In your revised cover letter, please address the following prompts:

a) If there are ethical or legal restrictions on sharing a de-identified data set, please explain them in detail (e.g., data contain potentially identifying or sensitive patient information) and who has imposed them (e.g., an ethics committee). Please also provide contact information for a data access committee, ethics committee, or other institutional body to which data requests may be sent.

b) If there are no restrictions, please upload the minimal anonymized data set necessary to replicate your study findings as either Supporting Information files or to a stable, public repository and provide us with the relevant URLs, DOIs, or accession numbers. Please see http://www.bmj.com/content/340/bmj.c181.long for guidelines on how to de-identify and prepare clinical data for publication. For a list of acceptable repositories, please see http://journals.plos.org/plosone/s/data-availability#loc-recommended-repositories.

We will update your Data Availability statement on your behalf to reflect the information you provide.

5. Please remove your figures from within your manuscript file, leaving only the individual TIFF/EPS image files, uploaded separately.  These will be automatically included in the reviewers’ PDF.

Please format all the references according to the standard of the journal.

Additional (section-by-section) Editor Comments:

Background section

Line 63-66: can you please provide more details on the findings of the studies that recommended for increasing the current WHO recommended cut-offs for MUAC?

Methods section

Line 74-77: Please provide a brief highlight of the methodological approach of the SMART surveys included in the analysis.

Can you please add a concise paragraph that provides an overview of Ethiopia including clear description of the pastoral and non-pastoral regions of the country?

Results

Please provide the proportion of children selected from the pastoral and agrarian regions.

Discussion

The authors concluded MUAC performs worse in identifying SAM children from a pastoral region of Ethiopia and justified that the finding can be due to “body shape difference” between Pastoral and Agrarian regions. But no explanation had been provided in what parameters the “body shape” is different between the two regions. Can you provide a brief highlight of the finding of the study by Myatt et al. [ ref # 18]?

Line 171: (reference)????

Conclusion

Line 210 “Both indicators showed gender, region and age bias.” Did you identify any bias associated with the use of WHZ?

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: No

Reviewer #2: No

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: No

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: No

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: This manuscript deals with an important topic.

The authors have made fundamental scientific mistakes by assuming weight for height as a gold standard and by trying to determine optimal cut offs by comparing the number of children detected by different anthropometric measures. The only way this can be achieved is by a study examining outcomes. Several studies have shown that weight for height and MUAC identify different groups of children and that MUAC has better predictive performance for mortality. Thus assuming weight for height as a gold standard is problematic because it has worse performance. The statement that switching to a MUAC based program would be unethical simply cannot be determined from the analysis that was done. It might be true, but it might also be true that a weight for height based program or a program based on both measures is unethical if it expends resources on a large number of children at low risk or it it fails to target those at greatest risk of death or neurodevelopment consequences. The other issues of ethical importance are coverage and cost benefit since unlimited resources will not be available and would need to be addressed in policy making decisions.

Overall the findings of how the two measures identify different populations in Ethiopia is of some interest in determining caseloads of potential different policies. The current WHO guidelines suggest using either measure but analysis of outcomes has led to a proposal that using a higher MUAC cutoff value would have better performance. Although the authors also suggest this, it should also only be based on a study that determines outcomes.

Reviewer #2: Routinely MUAC screening for severe acute malnutrition should consider the gender and age group bias in Ethiopian non-emergency context.

General comments.

The issue, of whether to use Weight-for-height or MUAC is still hotly debated, so far so that one has “believers” and “non-believers”.

This is mainly because there is never going to be a way to solve this issue. Firstly, what one is doing, is evaluating a screening tool to decide which one is best. This qualitative appraisal is done using sensitivity and specificity analyses and likelihood ratios to evaluate the text from a clinical perspective. The primary condition for this evaluation is that we “know” who is ill. Or to put it in another way ; that we know who is malnourished. Problem here is that we decide on this using an indicators that we are trying to evaluate. Here MUAC is evaluated against weight-for-height, or vice-versa. The flaw is that we are evaluating one indicator against another one without having a definition of who is really malnourished. We miss a “Golden Standard”.

The comparisons done in this present paper are therefore not very useful. Moreover, they say what many other papers have said already. Both indices find other populations and the overlap is small.

The second issue, is that arguments in favour of one or the other indicator are based on analyses of risk of mortality. Some studies find that MUAC is more predictive for mortality and hence should be used. The flaw here is that malnutrition is taken as an exposure variable, where it is rather an effect of a multi-causal problem. There a many different reasons why a child could be malnourished, and it is these reasons that are responsible for the increased mortality risk, and not the weight loss per sé.

Using malnutrition, based on an indicator, to predict mortality , is therefore not entirely correct from a strict epidemiological point of view.

So what can the authors do? They should have analysed more directly the effect of using a MUAC with single cut-off irrespective of gender. They should have done this using anthro and calculating the age and gender specific z-scores. Then they could have compared the misclassification comparing MUAC single cut-off with MUAC gender and age specific and distribution cut-off. This would have been more informative. The question of the study should have been whether to use MUAC with a single cut-off or use an age and gender specific one.

More specific comments

I have the impression that the authors have treated this “lightly” and not invested a lot in the existing literature. Moreover, on the HWO site is a useful systematic review on the matter: https://www.who.int/nutrition/publications/guidelines/updates_management_SAM_infantandchildren_review1.pdf

The paper would have benefitted from a more exhaustive literature evaluation.

Line 108. The Youden index is not correctly defined. It is not de difference between Se and SP; but rather Se+Sp minus 1.

How is accuracy measured and defined??

The rate in line 109 is not a rate, because the is no incidence and time line.

Table 2 is not useful , see my first main comment. The same goes for the AUC analyses.

Line 124: the figures are not corresponding to the figures.

Line 138. The false positive rates and the health post activities are not clearly explained. Is it because in the field MUAC is used , to be confirmed at the health post by Weight-for-height? In that case , indeed a lot of false positive will be identified. This is something to be avoided because it decrease the confidence the population will have in the health care providers. They go the health centre because they are told something is wrong with their child to be told late that there is no problem. The next time this happens, they will not go to the health centre again , because there previous experience thought them that nothing will happen. Late on, the authors use the higher sensitivity of MUAC to justify its use; forgetting the false positives at the health centre will have very important negative effects on health system appreciation. Line 187- 191

The discussion

This should be more focussed on the main question: MUAC one for all or a age and gender specific cut-off.

The argument of mortality should be re-evaluated against the second main comment I made.

Line 166: Why are logistic problem a challenge for MUAC???

Line 167: why is it a more “robust” screening tool???

Line 172: reference ?

Please also evaluated the quality of the surveys by giving an idea of the standard deviations as in the paper by Grellety. PLoS One. 2016 Dec 28;11(12):e0168585. doi: 10.1371/journal.pone.0168585. eCollection 2016. The Effect of Random Error on Diagnostic Accuracy Illustrated with the Anthropometric Diagnosis of Malnutrition. Grellety E1, Golden MH2.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: James A Berkley

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files to be viewed.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at figures@plos.org. Please note that Supporting Information files do not need this step.

27 Feb 2020

Editor Comments:

Background section

• Line 63-66: can you please provide more details on the findings of the studies that recommended for increasing the current WHO recommended cut-offs for MUAC?

We have summarized the main findings as requested.

Methods section

• Line 74-77: Please provide a brief highlight of the methodological approach of the SMART surveys included in the analysis.

Added as requested by the editor.

• Can you please add a concise paragraph that provides an overview of Ethiopia including clear description of the pastoral and non-pastoral regions of the country?

Added as requested for the agrarian and pastoral difference in the methodology section. We have also added an overview of the burden of acute malnutrition in Ethiopia in the introduction.

Results

• Please provide the proportion of children selected from the pastoral and agrarian regions.

As requested, we have added the # of children per area

Discussion

• The authors concluded MUAC performs worse in identifying SAM children from a pastoral region of Ethiopia and justified that the finding can be due to “body shape difference” between Pastoral and Agrarian regions. But no explanation had been provided in what parameters the “body shape” is different between the two regions. Can you provide a brief highlight of the finding of the study by Myatt et al. [ ref # 18]?

We have added a more substantial explanation on those factors which could influence body structure and therefore the use of MUAC or WHZ as a stand-alone.

• Line 171: (reference)????

Mistake, it has been taken out

Conclusion

• Line 210 “Both indicators showed gender, region and age bias.” Did you identify any bias associated with the use of WHZ?

We have included how WHZ is also associated to bias in the conclusion and that both methodologies are necessary

Reviewer 1:

The authors have made fundamental scientific mistakes by assuming weight for height as a gold standard and by trying to determine optimal cut offs by comparing the number of children detected by different anthropometric measures. The only way this can be achieved is by a study examining outcomes. Several studies have shown that weight for height and MUAC identify different groups of children and that MUAC has better predictive performance for mortality. Thus assuming weight for height as a gold standard is problematic because it has worse performance. The statement that switching to a MUAC based program would be unethical simply cannot be determined from the analysis that was done. It might be true, but it might also be true that a weight for height-based program or a program based on both measures is unethical if it expends resources on a large number of children at low risk or it it fails to target those at greatest risk of death or neurodevelopment consequences. The other issues of ethical importance are coverage and cost benefit since unlimited resources will not be available and would need to be addressed in policy making decisions.

We thank the reviewer for his comments, but do not agree with his statements. As we have tried to make clear in the manuscript is that cases of acute malnutrition consist of children with either a low weight-for-height, a low MUAC or both. There is no gold standard, and we have also not claimed that WHZ is superior of MUAC or vice versa. The purpose of our manuscript is to describe a way to identify as many children as possible with acute malnutrition. As MUAC is being used far more often in communes as screening tool than WHZ, is seems logic to try to optimize MUAC cut-off points to enable finding most children with also a low WHZ, without overburdening the health system. The statement from the reviewer “Several studies have shown that … MUAC has better predictive performance for mortality. Thus assuming weight for height as a gold standard is problematic because it has worse performance.” is wrong. The paper by Grettely and Golden [ref] shows clearly that WHZ is a better predictor of death than MUAC. Early papers were biased as children with both a low WHZ and MUAC were lumped together, creating a statistical artefact. Regardless of whether WHZ or MUAC predicts death better, children with a low WHZ or with a low MUAC are at an increased risk for death, something that has been shown by many studies. Treating only children with a low MUAC, and ignoring children with a low WHZ, is therefore clearly unethical. We have addressed the point on the coverage and the number of children falsely diagnosed in the discussion.

Overall the findings of how the two measures identify different populations in Ethiopia is of some interest in determining caseloads of potential different policies. The current WHO guidelines suggest using either measure but analysis of outcomes has led to a proposal that using a higher MUAC cutoff value would have better performance. Although the authors also suggest this, it should also only be based on a study that determines outcomes.

In our opinion, the paper by Schwinger, Grellety, Golden, analyzing death as outcome in >15,000 children is conclusive and warrants treatment of children with severe acute malnutrition, regardless on the indicator used to identify the severe acute malnutrition. It serves nothing to endlessly repeat relatively small, under-powered studies on outcomes of severe acute malnutrition.

Reviewer #2:

The issue, of whether to use Weight-for-height or MUAC is still hotly debated, so far so that one has “believers” and “non-believers”.

We do not categorize ourselves in Believers or not. We have just facts from Ethiopia. What do we do with the children not screened through MUAC alone. To date in Ethiopia, MUAC is used as the sole screening tool.

This is mainly because there is never going to be a way to solve this issue. Firstly, what one is doing, is evaluating a screening tool to decide which one is best. This qualitative appraisal is done using sensitivity and specificity analyses and likelihood ratios to evaluate the text from a clinical perspective. The primary condition for this evaluation is that we “know” who is ill. Or to put it in another way ; that we know who is malnourished. Problem here is that we decide on this using an indicator that we are trying to evaluate. Here MUAC is evaluated against weight-for-height, or vice-versa. The flaw is that we are evaluating one indicator against another one without having a definition of who is really malnourished. We miss a “Golden Standard”.

As we have tried to make clear in the manuscript is that cases of acute malnutrition consist of children with either a low weight-for-height, a low MUAC or both. There is no gold standard, and we have also not claimed that WHZ is superior of MUAC or vice versa. The purpose of our manuscript is to describe a way to identify as many children as possible with acute malnutrition. As MUAC is being used far more often in communes as screening tool than WHZ, is seems logic to try to optimize MUAC cut-off points to enable finding most children with also a low WHZ, without overburdening the health system. As the condition ‘ill’, or in this case, acute malnutrition, is defined by a low WHZ and/or MUAC, theoretically, we would have captured 100% of cases with both indicators (which is different from for example an infectious disease, where you can have more negative indicators)

The comparisons done in this present paper are therefore not very useful. Moreover, they say what many other papers have said already. Both indices find other populations and the overlap is small.

See above. The main purpose of the study was not to show the lack of overlap (which has indeed been shown before), but how many more cases of severe acute malnutrition could be identified, and how this would affect case-load and false-positive rates.

The second issue is that arguments in favour of one or the other indicator are based on analyses of risk of mortality. Some studies find that MUAC is more predictive for mortality and hence should be used. The flaw here is that malnutrition is taken as an exposure variable, where it is rather an effect of a multi-causal problem. There a many different reasons why a child could be malnourished, and it is these reasons that are responsible for the increased mortality risk, and not the weight loss per sé.

Using malnutrition, based on an indicator, to predict mortality, is therefore not entirely correct from a strict epidemiological point of view.

We fully agree that there are many underlying causes of malnutrition, but the increased risk for mortality is one of the major reasons to treat malnutrition. The same can be said about malaria. There are many underlying reasons why a child becomes infected by malaria (lack of bednets, poor WASH with pools of water, etc), but the fact that the child is at an increased risk for death is reason for treatment.

So what can the authors do? They should have analysed more directly the effect of using a MUAC with single cut-off irrespective of gender. They should have done this using anthro and calculating the age and gender specific z-scores. Then they could have compared the misclassification comparing MUAC single cut-off with MUAC gender and age specific and distribution cut-off. This would have been more informative. The question of the study should have been whether to use MUAC with a single cut-off or use an age and gender specific one.

This works has been done before, among others by us. See for example Fiorentino et al. that we have done in Cambodia. We do not see however how this would have addressed our prime objective of the study : based on current WHO guidelines for identifying (severe) acute malnutrition, how can assure that most children with (severe) acute malnutrition as diagnosed (and treated).

The comparison of MUAC vs MUAC-for-age or MUAC-for-gender, leaves out the complete part of children diagnosed with WHZ alone, in our case, ~50% of children. We will concentrate another paper on this topics.

More specific comments

I have the impression that the authors have treated this “lightly” and not invested a lot in the existing literature. Moreover, on the HWO site is a useful systematic review on the matter: https://www.who.int/nutrition/publications/guidelines/updates_management_SAM_infantandchildren_review1.pdf

We do not agree that we have take this lightly, but we have included more literature as requested.

The paper would have benefitted from a more exhaustive literature evaluation.

We have added some more literatures into the paper.

Line 108. The Youden index is not correctly defined. It is not de difference between Se and SP; but rather Se+Sp minus 1.

Thanks for the reviewer, it was a mistake from our in the explanation. We have changed it and the table accordingly.

How is accuracy measured and defined??

The percentage of biological implausible measures were very low. 3.8% of the values had a flag and were excluded from the analysis. As written, in the document, we have followed the recommended definition from WHO. To ensure the accuracy of the data, extreme values were excluded from the analysis: WAZ < −6 or > 5; L/HAZ < −6 or >5; WHZ < −6 or > 5. MUAC values were from 90cm to 200cm.

The rate in line 109 is not a rate, because the is no incidence and time line.

We have change it as suggested by the reviewer.

Table 2 is not useful, see my first main comment. The same goes for the AUC analyses.

See our comments previously as well, as we do think it is relevant.

Line 124: the figures are not corresponding to the figures.

What do you mean? If you ass for GAM 36.5+19.2=55.7% and for SAM 46.3+8.9=55.2…so yes they are corresponding to the figures.

Line 138. The false positive rates and the health post activities are not clearly explained. Is it because in the field MUAC is used , to be confirmed at the health post by Weight-for-height? In that case , indeed a lot of false positive will be identified. This is something to be avoided because it decrease the confidence the population will have in the health care providers. They go the health centre because they are told something is wrong with their child to be told late that there is no problem. The next time this happens, they will not go to the health centre again , because there previous experience thought them that nothing will happen. Late on, the authors use the higher sensitivity of MUAC to justify its use; forgetting the false positives at the health centre will have very important negative effects on health system appreciation. Line 187- 191

We do understand the concern, but if we are starting to work on the prevention of wasting, the false positive could be one of our focus as well. As they could cross the thin line which will categorize them as acute malnourished. We will add this notion in the text.

The discussion

This should be more focussed on the main question: MUAC one for all or a age and gender specific cut-off.

See our commented earlier. This works has been done before, among others by us. See for example Fiorentino et al. We do not see however how this would have addressed our prime objective of the study : based on current WHO guidelines for identifying (severe) acute malnutrition, how can assure that most children with (severe) acute malnutrition as diagnosed (and treated). The comparison of MUAC vs MUAC-for-age or MUAC-for-gender, leaves out the complete part of children diagnosed with WHZ alone, in our case, ~50% of children.

The argument of mortality should be re-evaluated against the second main comment I made.

Line 166: Why are logistic problem a challenge for MUAC???

We have erased this sentence as the only challenge is additional training and the development of the cut-off for MUAC gender and age based

Line 167: why is it a more “robust” screening tool???

Thanks to the reviewer, we have taken out the term of robust as it was out of line.

Line 172: reference ?

Thanks to the reviewer for highlighting this. We have resolved the issue as it was a mistake.

Please also evaluated the quality of the surveys by giving an idea of the standard deviations as in the paper by Grellety. PLoS One. 2016 Dec 28;11(12):e0168585. doi: 10.1371/journal.pone.0168585. eCollection 2016. The Effect of Random Error on Diagnostic Accuracy Illustrated with the Anthropometric Diagnosis of Malnutrition. Grellety E1, Golden MH2.

Included in the methodology as recommended. Our standard deviation for WHZ is pretty good as only 1.048SD.

Submitted filename: reviewers.docx

Click here for additional data file.

3 Mar 2020

Routinely MUAC screening for severe acute malnutrition should consider the gender and age group bias in the Ethiopian non-emergency context

PONE-D-19-28795R1

Dear Dr. Laillou,

We are pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it complies with all outstanding technical requirements.

Within one week, you will receive an e-mail containing information on the amendments required prior to publication. When all required modifications have been addressed, you will receive a formal acceptance letter and your manuscript will proceed to our production department and be scheduled for publication.

Shortly after the formal acceptance letter is sent, an invoice for payment will follow. To ensure an efficient production and billing process, please log into Editorial Manager at https://www.editorialmanager.com/pone/, click the "Update My Information" link at the top of the page, and update your user information. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to enable them to help maximize its impact. If they will be preparing press materials for this manuscript, you must inform our press team as soon as possible and no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

With kind regards,

Samson Gebremedhin, PhD

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

20 Mar 2020

PONE-D-19-28795R1

Routinely MUAC screening for severe acute malnutrition should consider the gender and age group bias in the Ethiopian non-emergency context

Dear Dr. Laillou:

I am pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please notify them about your upcoming paper at this point, to enable them to help maximize its impact. If they will be preparing press materials for this manuscript, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

For any other questions or concerns, please email plosone@plos.org.

Thank you for submitting your work to PLOS ONE.

With kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Samson Gebremedhin

Academic Editor

PLOS ONE