| Organization | Hospital for Sick Children, Toronto, Canada |
| Location | Kenya, Bangladesh, Uganda, and Malawi |
| Amount | $1.2 million |
| Approval date | February 2026 |
| Funded by | All Grants Fund. This grant is co-funded with the Children’s Investment Fund Foundation (CIFF) and Eleanor Crook Foundation (ECF). |
Summary
This grant co-funds a phase III randomized controlled trial (RCT) testing whether a lower-calorie therapeutic milk formula (F35) can reduce mortality in hospitalized children with severe acute malnutrition (SAM) relative to the current standard-of-care therapeutic milk formula (F75).1
Published: August 2026
The challenge
SAM affects approximately 19 million children under 5 globally and contributes to roughly 400,000 deaths per year.2 Of these cases, 15–20% require inpatient treatment,3 where mortality commonly exceeds 10%.4 Despite decades of effort, inpatient outcomes have improved little.5
Our understanding is that the current standard of care for SAM, therapeutic milk containing 75 kcal per 100ml (F75), is based on expert opinion, not on clinical trials.6 Results from a large RCT in high-income settings indicate that providing fewer calories during the initial treatment phase may lead to improved outcomes in critically ill children.7 However, to date, there is limited evidence about whether reduced-calorie treatments would provide similar benefits in low-income settings.
How this grant could help
Identifying improved treatment for children who are critically ill with SAM could save many lives. In the Phase I/IIa trial, results indicated that F35, containing 35 kcal per 100ml, was safe and was not associated with increased hypoglycemia or serious adverse events.8 This Phase III trial will enroll severely malnourished children aged 6 to <59 months with two or more clinical danger signs of severe illness across six sites in Malawi, Kenya, Uganda, and Bangladesh.9 Children will be randomly assigned at a 1:1 ratio to receive either F35 or F75 during the initial stabilization phase, transitioning to higher-calorie foods once stabilized.10 The primary outcome measure will be 14-day mortality.11 We expect the trial to enroll participants from October 2026 through October 2028.12
Our reasoning
This trial could lead to the widespread adoption of an alternative therapeutic milk formula that could substantially reduce mortality rates for a large number of children receiving inpatient treatment of SAM. We are uncertain about the likelihood of this trial's success, but our cost-effectiveness analysis using reasonable assumptions suggests that the grant is a good bet. This analysis serves as a quantitative expression of our thinking about the grant rather than a precise estimate.
Our analysis is based in part on uncertain probabilities of success at multiple points along our theory of change. However, we think that a) there is a reasonably high chance of success at each stage and b) the size of the predicted impact is large enough to make this grant a strong bet.
- We think there is an approximately 50% chance that the trial concludes that F35 is superior to F75. We think the hypothesis is well-justified based on current evidence (see above). Additionally, the trial's design is rigorous. This is a partially blinded RCT, minimizing bias, and five of six sites have participated in previous trials and demonstrated high adherence to study protocol.13
- The trial is multi-country and so would provide evidence across diverse settings, and the design of the trial means a positive result would come with high confidence in the superiority of F35.14 Therefore, we guess that with a positive result, there is a 70% likelihood that WHO would update its guidelines for SAM treatment.
- In a best-case scenario where the trial found that F35 was superior to F75, WHO updated its recommendations, and those recommendations were fully adopted, we estimate that this grant would accelerate the adoption of F35 by three years, and that roughly 330,000 children would receive updated treatment annually, averting roughly 12,000 deaths per year. Our BOTEC makes downward adjustments to this best case scenario to arrive at our cost-effectiveness estimate of 47 times our benchmark.
The results of this RCT could also inform a major grantmaking area for GiveWell: community-based management of acute malnutrition (CMAM). We are interested in ways to lower treatment costs for CMAM programming without compromising effectiveness, and F35 is expected to be cheaper than F75.15 Most of our CMAM grants involve some inpatient treatment of complicated SAM, and we believe that a substantial share of mortality occurs in children with complicated SAM. Thus, this trial has the potential to impact an important component of programs we would fund.
Additionally, co-funding with CIFF and ECF deepens our engagement with aligned funders, which may surface other high-value opportunities in the future.
What we could be wrong about
Despite its rigorous design, we think the trial may be statistically underpowered,16 and a false negative may be taken as definitive evidence against F35. However, the researchers have the option to extend the trial after interim analyses if it looks like it does not have enough power, and they have used an adaptive Bayesian design to balance power with efficiency and ethical considerations. While extending the trial would require additional funding, we think it adequately addresses this concern.
Our cost-effectiveness estimate relies on several highly uncertain parameters. However our estimate clears our funding bar17 by a wide enough margin that we think it’s relatively unlikely that plausible errors in these parameters, even large ones, would change our decision.
This page summarizes our research at the time of approval and was reviewed by the grant recipient.
Sources
- 1
- A research team from the CHAIN Network conducted the Phase I and IIa trials and will conduct the Phase III trial supported by this grant, with Robert Bandsma as lead researcher.
- "Building on these findings [of Phases I/IIa], we now propose a Phase III, multi-center, randomized controlled trial to assess the efficacy of the reduced-caloric formulation F35. Approximately 1200 acutely ill, severely malnourished children will be randomly assigned (1:1 ratio) to receive either F75 (standard of care) or F35 during the stabilization phase of hospitalization. The primary outcome will be reduction of mortality." Childhood Acute Illness and Nutrition (CHAIN) Network, Proposal for reduced calorie feeds in the early management of ill severely malnourished children, p. 2-3 (unpublished)
- 2"Severe acute malnutrition affects an estimated 19 million children under 5 years of age worldwide and is estimated to account for approximately 400,000 child deaths each year." WHO, Identification of severe acute malnutrition in children 6–59 months of age, 2023
- 3"Facility-based treatment, however, is required for complicated cases; approximately 15 percent to 20 percent of SAM cases require such treatment." Disease Control Priorities, 3rd Edition, Reproductive, Maternal, Newborn, and Child Health, 2016
- 4The trial's participating hospitals report a baseline mortality rate of approximately 13.9%, per the power calculation. We expect it may be higher in the average hospital. CHAIN, Proposal for reduced calorie feeds in the early management of ill severely malnourished children, p. 3 (unpublished)
- 5"Despite protocolized management of these highly vulnerable patients, mortality rates have remained high over the past decades and reported to range between 10-30%, depending on population, setting and duration of follow up." CHAIN, Proposal for reduced calorie feeds in the early management of ill severely malnourished children, p. 3 (unpublished)
- 6"The current F75 composition is only based on expert opinion." CHAIN, Proposal for reduced calorie feeds in the early management of ill severely malnourished children, p. 5 (unpublished)
- 7"In critically ill children, withholding parenteral nutrition for 1 week in the ICU was clinically superior to providing early parenteral nutrition." Fivez T et al. 2016
- 8"To address this, we conducted a Phase I/IIa clinical trial to determine the lowest calorie formulation of F75 (F50 and F35) that maintains safety, during early stabilization in ill severely malnourished children. The primary outcome for this initial Phase-I trial was the occurrence of at least one episode of hypoglycemia (≤3.0 mmol/l) during the stabilization phase. Results demonstrated that both F50 and F35 were safe, and not associated with increased hypoglycemia or serious adverse events compared to F75." CHAIN, Proposal for reduced calorie feeds in the early management of ill severely malnourished children, p. 1 (unpublished)
- 9“Inclusion Criteria:
- Age 6 to <59 months (age range for which WHO guidelines were developed and where children are not expected to be exclusively breastfeeding).
- Admitted to hospital with an acute, non-traumatic illness and having received a maximum of two F75 feeds at the time of enrolment.
- Severe malnutrition (WHZ <-3 z-scores of the median based on WHO growth standards and/or MUAC <115 mm and/or nutritional edema).
- Accompanied by caregiver able to provide written or witnessed informed consent.
- Presence of two or more features of clinical danger signs as outlined in Table 1."
CHAIN, Proposal for reduced calorie feeds in the early management of ill severely malnourished children, p. 12 (unpublished)
- 10"Approximately 1200 acutely ill, severely malnourished children will be randomly assigned (1:1 ratio) to receive either F75 (standard of care) or F35 during the stabilization phase of hospitalization…Once clinically stabilized, on average after 3-5 days, children are transitioned from F75 to ready-to-use therapeutic foods (RUTF) that provide a substantially higher amount of protein and calories to promote rapid weight gain." CHAIN, Proposal for reduced calorie feeds in the early management of ill severely malnourished children, p. 1-3 (unpublished)
- 11"The primary outcome of this trial will be a 14-day mortality." CHAIN, Proposal for reduced calorie feeds in the early management of ill severely malnourished children, p. 14 (unpublished)
Note that the proposal also lists occurrence of severe adverse events as a primary outcome, but that is outdated. "Initially we thought about using a composite outcome, but later decided against that and went for a more clean mortality only outcome and calculations are based on this." Email from Robert Bandsma, January 27, 2026 (unpublished).
- 12
- Email from Robert Bandsma, Lead Researcher, February 6, 2026. (unpublished)
- The trial uses an adaptive design wherein, at prespecified points throughout the trial, the researchers will determine whether to continue, extend, or stop early depending on whether there is enough evidence to conclude that F35 is better or worse than the standard of care or if there is unlikely to be enough statistical power to detect a significant benefit. This could mean that the timeline is longer or shorter than expected.
- "A series of interim analyses will be conducted to assess trial progress and determine whether to continue or stop the trial for inferiority, futility, or superiority. These analyses will be performed according to the prespecified schedule (Figure 5), and decisions regarding continuation will be guided by prespecified Bayesian statistical triggers." CHAIN, Proposal for reduced calorie feeds in the early management of ill severely malnourished children, p. 19 (unpublished)
- 13
- The trial will be partially blinded: The team preparing materials will know what product the child received, but the clinical care teams, patients, families, and statisticians will be blinded.
- Meeting with Robert Bandsma and F35 trial group, January 30, 2026 (unpublished)
- 14"The trial will be stopped for superiority if the posterior probability of superiority, P(relative risk < 1.0), is greater than the threshold of 0.98." CHAIN, Proposal for reduced calorie feeds in the early management of ill severely malnourished children, p. 10 (unpublished).
- 15Email from Robert Bandsma, November 15, 2025 (unpublished).
- 16
- “If F35 treatment would reduce absolute risk of 14-day mortality by 3.7%, i.e., a relative rate reduction of 26.5% compared to standard of care, with planned interim analyses as detailed above, the estimated average sample size would be 1211 for a power of 73.4%, where 39% of trials would require less than 750 patients and 53.4% of trials would require less than 1000.” CHAIN, Proposal for reduced calorie feeds in the early management of ill severely malnourished children, p. 10 (unpublished).
- With an intended enrollment of 1200, the trial has 73% power to detect an effect of the size they assume, lower than the typical 80%.
- 17Our funding threshold as of February 2026 was 8 times our benchmark.