Economic evaluation of a reduced dosage of ready‐to‐use therapeutic foods to treat uncomplicated severe acute malnourished children aged 6–59 months in Burkina Faso

Abstract Ready‐to‐use therapeutic foods (RUTF) used to treat children with severe acute malnutrition (SAM) are costly, and the prescribed dosage has not been optimized. The MANGO trial, implemented by Action Contre la Faim in Burkina Faso, proved the non‐inferiority of a reduced RUTF dosage in community‐based treatment of uncomplicated SAM. We performed a cost‐minimization analysis to assess the economic impact of transitioning from the standard to the reduced RUTF dose. We used a decision‐analytic model to simulate a cohort of 399 children/arm, aged 6–59 months and receiving SAM treatment. We adopted a societal perspective: direct medical costs (drugs, materials and staff time), non‐medical costs (caregiver expenses) and indirect costs (productivity loss) in 2017 international US dollar were included. Data were collected through interviews with 35 caregivers and 20 informants selected through deliberate sampling and the review trial financial documents. The overall treatment cost for 399 children/arm was $36,550 with the standard and $30,411 with the reduced dose, leading to $6,140 (16.8%) in cost savings ($15.43 saved/child treated). The cost/consultation was $11.6 and $9.6 in the standard and reduced arms, respectively, with RUTF accounting for 56.2% and 47.0% of the total. The savings/child treated was $11.4 in a scenario simulating the Burkinabè routine SAM treatment outside clinical trial settings. The reduced RUTF dose tested in the MANGO trial resulted in significant cost savings for SAM treatment. These results are useful for decision makers to estimate potential economic gains from an optimized SAM treatment protocol in Burkina Faso and similar contexts.

A third of the RUTF world production is produced by Nutriset (Malaunay, France) with local productions in several African countries, which sometimes struggle to attain sufficient quality standards (Segrè, Liu, & Komrska, 2017). Therefore, the RUTF is usually imported, cleared through customs and distributed by UNICEF in conjunction with the local ministries of health and the health districts. This leads to high costs and multiplicity of actors that can create supply chain disruption threatening the treatment sustainability for countries with weak health systems and scarce resources. It reinforces external dependence of low-and middle-income countries on donors to purchase the nutritional products (United Nations Standing Committee on Nutrition, 2011).
Furthermore, there is a lack of evidence regarding the RUTF optimal dosage during SAM treatment. A study conducted in Myanmar in 2015 showed a high recovery rate in children treated with a reduced dosage of RUTF following a stock-out (James et al., 2015). The MANGO trial investigated the non-inferiority of a reduced dosage of RUTF after the third week of treatment on the recovery of 801 children with SAM aged 6-59 months without medical complications. This randomized trial provided a rigorous assessment of the clinical impact of a reduced dose and presents an opportunity to assess its economic consequences as well.
To our knowledge, no economic evaluation of a reduced dosage of the RUTF within the SAM outpatient treatment protocol has been undertaken. Hypothesizing that a reduced dose would result in considerable savings that will allow countries to better support the economic burden of SAM treatment, we aimed to perform an economic evaluation alongside the MANGO clinical trial. We seek to quantify the economic impact of the transition from the standard RUTF dose to the new reduced dose tested in the trial. The district has 48 health centres, 10 of them were chosen to participate in the study based on the number of SAM children registered per month (minimum of seven new SAM cases per month), their accessibility and a suitable schedule to couple study visit days with routine growth monitoring days. The children included in the study are those aged 6 to 59 months who in accordance to the national protocol for SAM management in Burkina Faso had a weight-for-height z score <−3 and/or a mid-upper arm circumference (MUAC) < 115 mm, no oedema, who passed an appetite test and presented no medical complications. Other exclusion criteria included children having received SAM treatment in the last 6 months and caregiver planning to travel or unable to comply with the weekly check-up schedule, peanut or milk allergy or disability affecting food intake (Kangas et al., 2019).

| MANGO protocol
The children included in the MANGO trial were treated with strict compliance to the national protocol for SAM management in Burkina Faso (Burkina Faso Ministry of Health, 2014). After admission, chil-

Key messages
• Ready-to-use therapeutic foods (RUTFs) used to treat children with severe acute malnutrition (SAM) in community-based setting are costly, and the prescribed dosage has not been optimized.
• The MANGO trial implemented in Burkina Faso demonstrated the non-inferiority of a reduced RUTF dosage in community-based treatment of uncomplicated SAM compared with the standard dosage.
• We conducted a cost-minimization analysis alongside this trial, which has shown that reducing the dosage leads to 16.8% or $15.43 of cost savings per child treated.
• These results are useful for decision makers to estimate potential economic gains from an optimized SAM treatment protocol in Burkina Faso and similar contexts. dren were received for weekly follow-up consultations until discharge.
Children could be discharged from the study as recovered (reaching recovery anthropometric criteria), referred (presenting medical complications, having more than 5% weight loss within 3 weeks and having no more than 100-g weight gain over 4 weeks), defaulted (missing three consecutive visits but confirmed alive or transferred to health centres not involved in the study), non-responder (still not attaining recovery criteria by 16 weeks of treatment) or died. Each consultation consisted of the study staff taking the child's anthropometric measurements and a nurse performing a clinical examination to identify co-morbidities and potential medical complications and finally, if the child passed the appetite test, prescribing RUTF according to the weight, treatment week and trial arm of the child. The caregivers were also provided sensitization messages on child feeding and the use of the RUTF. As part of SAM treatment, children received systematic medical treatment consisting of the administration of an antibiotic (amoxicillin) on the day of admission, anthelmintic (albendazole) at the second visit and antimalarial drugs (arthemeter + lumefantrine) at any visit if the malaria rapid diagnostic test was positive. During the clinical examination, any diagnosed illnesses like malaria, diarrhoea and acute respiratory infections were also treated. Children who presented complications (hypoglycaemia, acute dehydration, septic shock, gastric dilation, heart failure, hypothermia, severe anaemia, fever and corneal ulceration) during a visit were referred for inpatient care and did not return in the trial. As per national protocol, children who missed a programmed visit were contacted by telephone or actively searched in the community by the community health workers to encourage them to continue the weekly follow-up visits.

| Perspective
As this study aims to assess the economic impact of a change of protocol for Burkinabe health decision makers, a societal perspective was preferred. It includes the costs borne by ACF and its partners, the health system and communities (Russell, Fryback, & Sonnenberg, 1999;Weinstein, Siegel, Gold, Kamlet, & Russell, 1996).
The portion of global costs borne only by institutions (excluding community costs) was also assessed.

| Strategies compared
A total of 402 children with SAM were treated with standard RUTF dose (control arm) and 399 children with reduced RUTF dose (intervention arm). In order to be able to make a comparison based on two arms with identical number of participants, we randomly removed three children from the arm that received the standard dose. We therefore performed the economic evaluation with 399 children per arm.
Socio-economic characteristics of children recruited in the two arms were not different (Kangas et al., 2019). All children were treated in accordance with the national community-based management of acute malnutrition (CMAM) protocol (Ministère de la santé BF, 2014). However, in the intervention group, children received a reduced dose of RUTF from the third week of treatment onwards, prescribed according to the weight of the child (see Table 1). The direct non-medical costs are related to transport and other outof-pocket expenses supported by the beneficiaries while seeking SAM treatment. Indirect costs are defined as costs related to temporary absence from work due to illness, reduced working capacity due to illness and disability or lost productivity due to early death (Kirch, 2008). Further details on each cost component are available in the technical appendix. The evaluation criterion was the difference between the costs involved in each arm.

| Data collection
Data used in this study were collected through two data field collec- Seventeen face-to-face semi-structured interviews were conducted as well as three focus group discussions and structured observations in 2017 in 8 out of 10 health centres participating in the trial. Responses were triangulated with different interviewees to minimize bias. Further information on data collection can be found in the appendix.

| Cost estimation and analysis
The costs for each step in the trial were proportionally allocated based on actual usage during the implementation. When possible, an ingredients-based approach was used to estimate unit costs. Following the guidelines, all resources that may substantially influence overall costs were collected and all big item costs that are differential between the two arms were assessed (Ramsey et al., 2015).
This study does not include the costs that had no impact on the intervention itself such as research costs (team  Caregiver's loss of productivity (or opportunity cost) was calculated as the potential income lost due to the follow-up consultation, as if they had otherwise performed an activity with an economic value (Drummond et al., 2015;Puett et al., 2013). Caregiver characteristics collected in the baseline questionnaires of the trial showed that 95% reported 'housewife' as their status, and therefore, housework was considered as their main occupation.

| Multivariate sensitivity analysis
We conducted multivariate sensitivity analysis whereby various parameters of the model were varied simultaneously to assess the joint uncertainty of all the model parameters over different scenarios.
First, a worst and best case scenario were explored to assess the minimum and the maximum saving incurred by the reduced dosage. Informed consent to participate in the study was obtained from participants prior to any data collection. Permission to record interviews was collected verbally from the participants and the necessary steps were taken to ensure their confidentiality. No incentives were provided to study participants.

| Base case results
The patient's socio-demographic and clinical characteristics were previously reported in the trial efficacy paper (Kangas et al., 2019).
Children in both arms were on average 13 months of age at admission. Only 14% and 15% of the beneficiaries were living in an urban area, 63% in the reduced arm and 61% reduced dosage arms lived more than 30-min return trip to the health centre, The main unit costs are presented in Table 2 and a full list can be seen in the appendix. The average cost per sachet prescribed during the trial was $0.33, the average cost of drugs delivered per consultation was $0.2 and human resource cost was $1.9 per treatment consultation. The average total costs per treatment consultation was $11.6 and $9.6 in the standard and reduced arms, respectively, with RUTF and related transport and storage costs accounting for 56% and 47% of the total consultation. Community opportunity costs were equivalent to 11.4% and 13.7% of the overall treatment cost in the standard and the reduced dosage arms, respectively.

| Univariate
A tornado analysis was performed to identify the key cost drivers.
Results showed that main cost drivers were related to RUTF cost per

| Multivariate
In the worst and best case scenarios, the global cost reduction between the two protocols ranged from 16.7% to 20.2% corresponding to savings per child treated between $11.4 and 22.5$.
The 'real-life' scenario showed a lower cost per consultation: $7.7 for the standard dose and $6.2 for the reduced one. Under the assumption of that scenario, global costs were $24,322 for the standard arm T A B L E 3 Global treatment costs, base case analysis, expressed in 2017 dollar using purchasing power parity (1 euro = USD $0.80 PPP) Note. #The institutional perspective includes cost borne by Action Against Hunger and the health system. The societal perspective includes institutional perspective costs and community costs. Abbreviation: RUTF, ready-to-use therapeutic food.
F I G U R E 2 Univariate sensitivity analysis: Tornado diagram of global incremental cost between arms. The horizontal axis is our main outcome (the cost reduction between the compared protocols); along the vertical axis, parameters are arrayed, and horizontal bars represent the outcome range associated with each specified parameter's tested range in the sensitivity analysis. The outcome point estimate corresponding to base case values is indicated by the vertical line cutting through all horizontal bars. The two most influential variables on the difference between the two arms were the provenance of RUTF then followed by the average cost of drug administrated during a treatment consultation. Other variables had limited impact on incremental cost between the two dosages. RUTF: ready-to-use therapeutic food and $19,762 for the reduced one. It led to a 19% cost reduction corresponding to $11.4 per child treated of (vs. $15.4 in the base case). Further results of the sensitivity analysis can be found in the technical appendix.

| DISCUSSION
Previous results of the MANGO trial showed that a reduced RUTF dose did not affect children's recovery from SAM, their length of stay, their vitamin A or iron status nor their body composition (Kangas et al., 2019; (Wilford et al., 2012).
As in the other studies, the largest cost driver in the MANGO trial was the RUTF, but our results highlighted that the sourcing of the product was very influential. The product accounted for 56.2% and 47.0% of overall costs in the standard and reduced dosage arms, respectively. This proportion is higher than that reported by Isanaka et al. (2017) in Niger (44%). The RUTF cost estimations vary depending on whether or not storage and product management costs (that are rarely taken into account) are included. Its chosen mode of transportation affects its cost (airfreight being expensive compared with shipping) as well as whether international transportation is needed. Therefore, the RUTF share in the treatment costs depends largely on the RUTF supply chain in the country concerned. In the MANGO trial, only 17% of the prescribed RUTF was locally supplied, and it was 26% cheaper than the one supplied abroad. If all the RUTF used had been supplied through the Ministry of Health's usual supply circuit-which is the cheapest-the RUTF would have represented 48.6% of the cost per child treated with the standard protocol and lead to a cost reduction of 14% in the total consultation costs in that arm. In our study, the locally produced RUTF was cheaper than the one bought abroad and sent by air, but this may not be the case in other contexts. Indeed, producing RUTF locally can lead to very high production costs, especially when ingredients and manufacturing resources are rare and expensive. In low-income countries, setting up a highly industrialized production system can be extremely costly. In addition, running costs of the production site (like electricity) can also lead to higher expenses that can then be reflected in the final product costs and make it more expensive than an imported product.
Our study further shed light on the importance of the costs borne by the community seeking treatment that are largely assumed to be free of charge in Burkina Faso. These costs for the household repre- is not always followed. This is a strong limitation, but the durations estimated can be more accurately valued in the future provided that reliable data become available. However, inversely, we considered that the costs of transport from homes to health centres were negligible and were not included. Second, the time allocated to the various activities was collected on the basis of recall; therefore, response bias may exist. Third, the number of staff, their wages and their strict compliance to the protocol steps were not reflective of real-life practice and the trial-associated costs were likely greater than in real life.
However, this impacted both arms similarly and therefore was nullified in the cost difference calculation; it did not affect the final percentage of cost reduction. Furthermore, the impact of these limitations was extensively explored in the sensitivity analysis and did not significantly change our main results. could be used to treat 7,212 more SAM cases. These estimates would need to be further assessed through a budget impact analysis, but they are likely to represent interesting savings without lowering the current effectiveness of the SAM treatment.

| CONCLUSION
Our cost-minimization analysis shows that the adoption of a reduced dosage of RUTF in the treatment of SAM might lead to substantial cost saving in the Burkina Faso context. The resources saved may be more efficiently used for malnutrition prevention and treatment activities.

ACKNOWLEDGMENTS
We thank the staff of Action Contre la Faim in Ouagadougou and Fada NGourma, the research team of the MANGO trial and caregivers for participating in this study.
The study was funded by Children's Investment Fund Foundation (grant CIFF03: 'Reinventing community-based management of malnutrition').