New approach to simplifying and optimising acute malnutrition treatment in children aged 6–59 months: the OptiMA single-arm proof-of-concept trial in Burkina Faso

The Optimising treatment for acute MAlnutrition (OptiMA) strategy trains mothers to use mid upper arm circumference (MUAC) bracelets for screening and targets treatment to children with MUAC < 125 mm or oedema with one therapeutic food at a gradually reduced dose. This study seeks to determine whether OptiMA conforms to SPHERE standards (recovery rate > 75 %). A single-arm proof-of-concept trial was conducted in 2017 in Yako district, Burkina Faso including children aged 6–59 months in outpatient health centres with MUAC < 125 mm or oedema. Outcomes were stratified by MUAC category at admission. Multivariate survival analysis was carried out to identify variables predictive of recovery. Among 4958 children included, 824 (16·6 %) were admitted with MUAC < 115 mm or oedema, 1070 (21·6 %) with MUAC 115–119 mm and 3064 (61·8 %) with MUAC 120–124 mm. The new dosage was correctly implemented at all visits for 75·9 % of children. Global recovery was 86·3 (95 % CI 85·4, 87·2) % and 70·5 (95 % CI 67·5, 73·5) % for children admitted with MUAC < 115 mm or oedema. Average therapeutic food consumption was 60·8 sachets per child treated. Recovery was positively associated with mothers trained to use MUAC prior to child’s admission (adjusted hazard ratio 1·09; 95 % CI 1·01, 1·19). OptiMA was successfully implemented at the scale of an entire district under ‘real-life’ conditions. Programme outcomes exceeded SPHERE standards, but further study is needed to determine if increasing therapeutic food dosages for the most severely malnourished will improve recovery.

Acute malnutrition is a major public health problem, affecting 49·5 million children worldwide each year and contributing to nearly half of all annual childhood deaths (1,2) . One quarter of all acutely malnourished children are in Africa with the largest number, 5·1 million, in West Africa (2) . In this region, the burden of acute malnutrition overlaps with the highest rates of child mortality (2) . Progress in reducing acute malnutrition needs to accelerate in the region if it is to meet WHO Global Nutrition Targets by 2025 (3) .
Although acute malnutrition is a continuum condition, it is arbitrarily divided into moderate (MAM) and severe (SAM) categories defined by mid upper arm circumference (MUAC) or weight-for-height Z-score (WHZ). MAM (MUAC 115-124 mm or WHZ between -2 and -3) and SAM (MUAC < 115 mm or WHZ <-3 or oedema) are managed separately, with programmes overseen by different UN agencies, and using different protocols and products (ready-to-use supplementary food or fortified-blended flours for MAM and ready-to-use therapeutic food (RUTF) for SAM). Such separation complicates delivery of care, contributes to low coverage and creates confusion among caregivers (4) .
Confusion is further exacerbated by a complicated case definition using MUAC and WHZ to determine programme eligibility, even though WHZ does not offer a clear advantage over MUAC for identifying children at near-term risk of death (5) . MUAC-only programming is expanding as evidence accumulates that weight gain and MUAC gain track each other and that the trajectory of weight and MUAC gain is maximal during the first 2-3 weeks of treatment (6)(7)(8) . MUAC is therefore becoming a stand-alone practical tool for all phases of nutrition programming: screening, admission, monitoring recovery and determining discharge.
In addition to the tangle of agencies and case definitions, SAM and MAM programmes are chronically underfunded with only 25 % of SAM cases treated globally in 2016 and 16 % of MAM cases reached by the World Food Program in 2017 (9,10) . An optimised allocation of resources is therefore needed. A programme integrating SAM and MAM treatment in a single MUAC-based protocol using only RUTF showed the cost of RUTF for integrated treatment to be less than half that for standard SAM management (11) . Other studies found that identification and treatment of children earlier in the wasting process led to fewer hospitalisations and that inpatient care was shown to be twice as costly as outpatient SAM management (12,13) . Furthermore, the treatment of MAM has been shown to be cost-effective in reducing mortality risk by more than 10 % (14) .
We piloted a new MUAC-based and oedema approach for treating acute malnutrition in Burkina Faso with a single-arm proof-of-concept trial called Optimising treatment for acute MAlnutrition (OptiMA). OptiMA made three strategic changes to the current Burkina Faso National Nutrition protocol (15) : screening for acute malnutrition was task-shifted to caretakers via community-based training in the use of MUAC bracelets; the case definition for acute malnutrition was modified to MUAC < 125 mm and/or oedema and only one product was used for treatment (RUTF) at a gradually reduced dose based on a child's weight and MUAC status. We hypothesised that the OptiMA protocol was at least as effective as the national nutrition protocol in terms of children's recovery rate and other standard programme indicators when compared against international SPHERE standards (16) .

Location, period and study population
This single-arm proof-of-concept trial was conducted in Burkina Faso's Yako Health District. In 2016, Yako had a general population of approximately 415 000, with the under-five population estimated at 82 000, and was served by fifty-four health centres plus a district hospital. Each health centre was staffed by a nurse who routinely provided acute malnutrition treatment. Prevalence of MAM and SAM by WHZ just prior to implementation of OptiMA was 8·2 % and 1·9 %, respectively (17) . Health centre staff were trained on the OptiMA protocol in November and December 2016. Inclusions for the study occurred from January through December 2017, and follow-up of study participants ended in March 2018. Children were considered eligible for enrollment if they were aged 6-59 months with a MUAC < 125 mm or bipedal oedema and presented to any of the fifty-four outpatient clinics in Yako District.

Study protocol
Screening for acute malnutrition. A mass community-based campaign was conducted from June to August 2016 to train mothers and caretakers throughout Yako District to use MUAC bracelets and check for oedema in order to screen their own children for malnutrition. Following the mass campaign, family MUAC trainings became a routine activity at all fifty-four health centres and continued to the end of the study.
Admission/discharge criteria, therapeutic feeding products and dosage. Table 1 summarises the differences between the current national protocol and OptiMA, which was implemented in all of the Yako's fifty-four health centres for the duration of the study. Children presenting with MUAC ≥ 125 mm and WHZ < −3 were treated in a 'non-MUAC eligible' arm according to the current national protocol.
In contrast to the weight-based RUTF ration in the national SAM programme, which is fixed at 150-200 kcal/kg per d (628-837 kJ/kg per d) for the course of treatment, the OptiMA RUTF ration was calibrated to the child's degree of wasting based on the combination of MUAC status and weight. Thus, more nutritional support was given to the most severely malnourished and then gradually reduced as the child's MUAC increased. Children with MUAC < 115 mm or oedema received 175 kcal/ kg per d (732 kJ/kg per d) of RUTF. Children with MUAC 115-119 mm, either at admission or during the course of treatment, received 125 kcal/kg per d (523 kJ/kg per d) of RUTF, and children with MUAC ≥ 120 mm received 75 kcal/kg per d (314 kJ/kg per d) of RUTF (with a minimum of one sachet/d) until discharge from the programme.
Systematic medical treatment. All children underwent malaria rapid testing upon admission and at any point during their participation if clinical signs of malaria were detected. All children with a positive malaria rapid diagnostic test were prescribed an artemisinin-combination treatment. Amoxicillin 90 mg/kg per d for 7 d was prescribed for all children with MUAC < 120 mm or oedema. Albendazole was given to children if they had no deworming in the previous 4 months.

Defaulter follow-up
Community health workers traced children who were classified as defaulters to their caretakers' homes and completed a standardised form recording vital status, MUAC measure and reasons for defaulting.

Primary and secondary outcomes
The primary outcome was recovery defined as the proportion of children achieving a MUAC ≥ 125 mm and absence of oedema for two consecutive weeks, in good clinical health, with a minimum programme stay of 4 weeks. Secondary outcomes were RUTF consumption and other standard programme indicators: death (occurring after programme admission and prior to recovery), default (being absent for three consecutive visits), non-response (failure to achieve recovery after 12 weeks in the programme) and hospitalisations (children referred for inpatient care at admission or at any point during the course of treatment).

Data collection procedures and monitoring
One of the five supervisors visited each health facility bimonthly to ensure protocol adherence and correct completion of patient records. There was no increase in clinical staff to manage the nutrition programme.
Data were collected using the national programme individual outpatient records (modified to include information on family MUAC-training). These forms were entered into an anonymised Access database. Data monitoring was conducted from completed archived patient records and included searching for each child's anonymous unique identifier in health centres in order to ensure that all children enrolled were in the database.
Socio-demographic information on the caretaker and child, mode of referral, distance to the health centre, caretaker MUAC-training prior to admission, Amoxicillin prescription and results of malaria rapid diagnostic test were collected at the first visit. The child's weight, MUAC, temperature, clinical symptoms and amount of RUTF ration were recorded at each weekly visit. Children's length was measured at admission and once a month thereafter. Weight was measured to the nearest 100 g with a Salter scale, and length was measured to the nearest 0·5 cm on a height board with the child in a supine position (or standing if taller than 85 cm). MUAC was measured to the nearest mm with a MUAC bracelet demarcated in 1 mm increments. At each visit, supervisors ensured that scales were correctly calibrated and MUAC bracelets and height boards were in good condition.

Data analysis
Data analysis was performed using RStudio software (RStudio, Inc.).
WHZ, weight-for-age Z-score and height-for-age Z-score (HAZ) were calculated using WHO 2006 growth standards.
In order to evaluate adherence to the new OptiMA dosage tables, the difference between weekly RUTF ration provided to the child and the theoretical OptiMA ration as calculated by the new tables were described.
Children who were alive but erroneously classified as recovered (after a single MUAC measurement ≥ 125 mm) or defaulter (prior to three consecutive weeks of absence) or non-respondent (before 12 weeks of participation without achieving discharge criteria), and who did not correspond to another category of exit, were reclassified as 'Alive, unconfirmed status'.
Programme outcomes were described overall and stratified by MUAC category at admission with their 95 % CI and considered at least as effective as the international SPHERE standards if the lower limit of the 95 % CI was greater than or equal to the reference value. Multivariable survival analysis to identify variables predictive of recovery was run using a shared frailty model (18) (extension of the Cox proportional hazard model), with a random effect on health centres, and by assuming a Weibull distribution for the baseline hazard function. A univariate analysis by fitting a separate model for each covariate was done, variables having a P-value ≤ 0·2 were entered into multivariate analysis. P-value < 0·05 was considered as statistically significant in the final model.
To investigate the contribution of factors that explain the recovery (i.e. the proportion of recovered children explained by each variable of interest), adjusted attributable fraction (AF) of risks was estimated based on the final multivariate model. The AF for each factor was obtained by the difference between the predicted number of recovered children in a simulated dataset where exposure to each variable was removed and the real number of recovered children in OptiMA database (19) . The adjusted AF were estimated in relation to the recovery at the median length of stay in the programme. We computed 95 % CI using bootstrapping (19) .

Ethics
The present study was approved by the Ethics Committee for Health Research (2016-6-067) and Technical Review Committee for Clinical Trial Authorizations (5003720165EC0000) of Burkina Faso's Ministry of Health. The trial was registered on clinicaltrials.gov (NCT03027505). Caregivers gave written consent (signature or fingerprint) prior to enrollment for all children included in the study. All data were anonymised when entered into the database, and unique identification numbers were coded. A data safety monitoring board conducted an independent interim analysis, reviewed consolidated results and endorsed completion of the study.

Results
Of the 5638 children included in the nutrition programme from January to December 2017, 4958 were included for analysis (Fig. 1). Children omitted from analysis were those who relapsed, were readmitted, had missing data, did not meet OptiMA selection criteria or were not MUAC eligible (MUAC ≥ 125 mm and WHZ < -3). Non-MUAC eligible children are described in online Supplementary file 1.
Home visits were conducted for 233 children classified as defaulters: 113 were alive and physically seen (average MUAC was 124, SD 7·1), thirty-eight were declared alive by the caretaker but not physically seen, four were deceased and seventy-eight could not be found. Among the seventy-seven defaulters with a MUAC < 115 mm at admission, the average number of visits before default was 4·7 and their last MUAC measure was 115·1 mm on average (SD 8·2).
Of the twenty-two deaths, 72·7 % were younger than 24 months, 59·1 % were girls and 54·5 % had a MUAC < 115 or oedema at admission, while this group accounted for only 16·6 % of all admissions. The average time from admission to death was 18·8 (SD 19·2) d, with six, eight and eight deaths occurring on the day of admission, 1-21 and 22-60 d after admission, respectively. Among these sixteen deaths that occurred after the day of admission, ten children had MUAC < 115 mm, nine were girls and the median age was 14 months (interquartile range 7·7, 26·2) Among all recovered children and those admitted with MUAC < 115 mm or oedema, length of stay was 5·8 and 8·1 weeks, MUAC gain was 9·8 and 17·5 mm, weight gain was 3·7 and 4·7 g/kg per d and the amount of RUTF provided was 51·0 and 87·4 sachets, respectively ( Table 5). The OptiMA dosage table provided a slightly lower than the anticipated average daily energetic intake to the most malnourished at 131 kcal/kg per d (548 kJ/kg per d).      UNICEF Burkina Faso plans an average of 140 sachets per SAM child treated. The comparable group in OptiMA is the 1778 children who met standard SAM admission criteria and whose average RUTF ration was reduced by nearly half at 72·2 sachets per child treated. The 3072 children treated under OptiMA who fit the current MAM case definition consumed an average of 54·3 RUTF sachets/child which is lower than the 60-90 ready-to-use supplementary food sachets/child typically planned for MAM programming. Table 6 shows factors associated with recovery after controlling for age and sex: alive mother status (adjusted hazard ratio 1·39; 95 % CI 1·02, 1·89), child of a caretaker who received MUAC training (adjusted hazard ratio 1·09; 95 % CI 1·01, 1·19) and who lived close to health centres (adjusted hazard ratio 1·25; 95 % CI 1·11, 1·42) were more likely to recover. Conversely, children who did not receive seasonal malaria chemoprevention during July and October were less likely to recover (0·87; 95 % CI 0·81, 0·94).

Discussion
This proof-of-concept trial evaluated a therapeutic nutrition protocol for children affected by acute malnutrition, defined as MUAC < 125 mm or oedema, and treated with one product (RUTF) at a gradually reduced dose based on a child's weight and MUAC status. The OptiMA protocol was implemented in a 'real-life' setting by Burkinabe Ministry of Health personnel in all fifty-four health centres of Yako District, with operational support from a national and international non-governmental consortium and an established UNICEF-supported RUTF supply chain. Nurses had minimal difficulty adapting to this new protocol as evidenced by excellent adherence to the new dosing table and application of the new admission criteria. Programme outcomes overall were satisfactory with substantial reduction in RUTF ration compared with standard SAM treatment. However, recovery and non-response rates were unsatisfactory in children admitted at MUAC < 115 mm or oedema.
Overall recovery and mortality rates reported here, 86·3 and 0·4 % respectively, compare favourably to the 83 % recovery and 4 % mortality reported by a similar MUAC-based study in Sierra Leone, which had a more rapid reduction in RUTF dosage than OptiMA (11) . A MUAC-based programme previously implemented in the same district of Burkina Faso and using MUAC < 120 mm for admissions reported 91·6 % recovery and 1·1 % mortality, but the present study did not use a reduced RUTF regimen and had a less stringent discharge criterion of a single measure of MUAC ≥ 124 mm (6) .
Recovery in the category of children admitted at MUAC < 115 mm or oedema (70·4 %) was poorer than anticipated; this was largely attributable to high non-response (10·6 %) and defaulter rates (9·3 %). Non-responders and defaulters did achieve an average MUAC of 119·5 and 115·1 mm at exit, respectively, while the average MUAC at admission in the MUAC < 115 mm group was 109·5 mm. Mortality remained low at 1·5 % and is similar to the mortality rate of 1·6 % reported for children with MUAC < 115 mm at admission in the prior MUAC-based programme in this same district (6) . Elsewhere among MUAC-based programmes without RUTF reduction, recovery rates for this category have ranged from 57·4 % in India, 63·4 % in Malawi and 82 % in Sudan (20)(21)(22) . In the MUAC-based programme with RUTF dose reduction in Sierra Leone recovery for this sub-group was 75·3 % (23) ; however, nutritional oedema was more prevalent in Sierra Leone (2·9 v. 0·7 % in OptiMA). This range of reported recovery rates with and without RUTF dose reduction suggests that recovery for this category is challenging. Other important factors like underlying disease (HIV, tuberculosis, sickle cell disease) might have contributed to a low recovery rate.
Similarly, recovery rate was the lowest (64·3 %) for children admitted with MUAC < 115 mm and WHZ < -3 (8·8 % of all admissions) and is again driven primarily by a high nonresponse rate at 12·5 %. Stunting was also highly prevalent in this group, reaching 57·9 %. Previous studies described a correlation between shorter length and treatment non-response rate, suggesting that severely wasted and stunted children may require other interventions over a longer time period to improve growth trajectory (7,24) .
The sub-optimal outcomes for the small proportion of severely wasted and stunted children should be interpreted in light of overall programme performance. Under OptiMA, only 16 % of children treated were admitted with MUAC < 115 mm or oedema, while 84 % were admitted with MUAC between 115 and 124 mm, whereas the proportion of children in each category in the Sierra Leone study was 30 and 70 % (11) . This might suggest that the OptiMA programme achieved good coverage and caught most children early in the wasting process. Thus, those children who exhibit severe wasting and stunting, in spite of apparently good access to treatment, may represent either a history of low birth weight or malnutrition secondary to chronic illness, both of which lower expectations for treatment outcomes.
Current SAM programmes typically plan RUTF consumption between 120 and 150 sachets per child treated, while MAM programmes plan between sixty and ninety sachets of ready-to-use supplementary food per child (25,26) . However, comparisons of average rations must consider variations in discharge criteria. Under OptiMA, discharge was stringent: two consecutives measures of MUAC ≥ 125 mm. In this light, the overall RUTF consumption of 60·8 sachets average per child per course of treatment is highly encouraging. However, further study will be needed to determine if increasing RUTF dosage for children admitted with MUAC < 115 mm will improve recovery and nonresponse even if other factors appear to influence recovery rate as shown in the multivariate analysis.
Female sex bias in MUAC-based nutrition programmes has been documented in previous studies (11,27) , similar to what the present study observed with 58·4 % of girls at admission. Isanaka et al. also clearly show that WHZ score preferentially selects males (27) . Thus combining anthropometric criteria, MUAC and WHZ serve to maintain sex balance in nutrition programmes. Using MUAC as the single anthropometric criterion for programme admission may require improving MUAC sensitivity for boys and specificity for girls. The multivariable survival model highlighted that recovery was positively associated with male children older than 24 m, living within 10 km of the health facility, whose mother was alive and had been trained to use MUAC. Recovery was negatively associated with children who had not received seasonal malaria chemoprevention between July and October. Most of these results have already been found in other studies. It is well known that female and younger children present lower MUAC (28) . As for distance to the health facility, a study from Ethiopia showed that SAM children living within 25 min of the facility were 1·53 times more likely to recover (29) . It has also been shown that mothers screening their children by MUAC can lead to earlier initiation of care (12) . The low proportion of self-referral by mothers (12·9 %) is likely due to an underestimation because many mothers sought confirmation from a Community Health Worker (CHW) after using MUAC at home and then reported being referred by a CHW at admission. To the best of our knowledge, it is the first time that a positive association is demonstrated between training mother to use MUAC bracelet and recovery, likely due to a better care-seeking behaviour resulting from such trainings. The negative association between children who did not receive CPS should be viewed with caution because the reason why these children did not receive the seasonal malaria chemoprevention was unknown and we cannot exclude a confusion bias or even reverse causality. Of these six variables associated with recovery, AF calculation highlighted, unsurprisingly, that recovery was primarily driven by maternal vital status and living within 10 km of the health facility.
The main limitation in the present study is the absence of a comparator group. In order to determine how MUAC-based integrated SAM and MAM protocols perform compared with current protocols that rely on MUAC and WHZ criteria, additional randomised trials, in multiple contexts, are necessary. Another limitation is that we did not collect data on possible RUTF sharing. But, it was previously shown in a study on MAM treatment in the same location that nearly two-thirds of children consumed all of the prescribed LNS ration themselves (30) .
Lastly, this analysis relied on routinely collected programme data. Extensive data monitoring was carried out throughout the study to ensure that all children found in health centre registers were retrieved in the OptiMA database. To our knowledge, such monitoring is rarely performed in programmatic studies. The strength of the present study is that it shows it is possible to conduct robust data monitoring in a 'real-life' programme to generate a high-quality individualised database. However, it was not without challenge. The reclassification of children erroneously categorised at exit-influenced OptiMA's recovery rate downwards. Adherence to routine amoxicillin prescription for children with MUAC < 120 mm or oedema as specified in the protocol was uneven and thus precluded any analysis of correlation between this prescription and recovery.

Conclusion
A simplified, combined SAM and MAM protocol based on early detection through household MUAC screenings, treatment with one product determined by the presence of oedema and/or MUAC < 125 mm, and progressive RUTF dose reduction had programme outcomes that exceeded national Burkina Faso and international SPHERE standards. This early detection and treatment initiation with progressively reduced RUTF ration shows promise and warrants further operational pilots in different contexts. However, randomised control trials are needed to compare the strategy to current national protocols (31) . Because of low recovery among children admitted with MUAC < 115 mm, further study is also needed to determine whether increasing RUTF dosage for the most severely malnourished will improve outcomes. isation ALIMA (Alliance for International Medical Action), primarily on improvement of maternal and child health outcomes. The scientific policy of the CORAL partnership alongside with the supervision of research projects and dissemination of results is defined by a board of directors consisting of senior representatives from both ALIMA and Inserm: Renaud Becquet (Inserm Bordeaux, methodological co-chair), Susan Shepherd (ALIMA, medical co-chair), Augustin Augier (ALIMA), Solenne Barbe (ALIMA), Marie Jaspard (ALIMA & Inserm Abidjan), Claire Levy-Marchal (ALIMA) and Xavier Anglaret (Inserm Bordeaux & Abidjan). The CORAL partnership meets annually with an external scientific advisory group to review projects and strategic orientation.
The study was funded by the European Union's humanitarian aid funds; the US Agency for International Development's Office and the Médecins Sans Frontières Foundation. The funders had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript.
This document covers humanitarian aid activities implemented with the financial assistance of the European Union. The views expressed herein should not be taken, in any way, to reflect the official opinion of the European Union, and the