Preventing malaria with AMF

A practical use

Put a possible benefit beside the cost

Under our central assumptions, $1,000 received by AMF could prevent roughly 16 episodes of malaria in young children. The suffering avoided during those illnesses adds up to about 100 USU. Direct malaria deaths prevented are shown separately.

This gives a donor a concrete starting point for asking what their support might achieve and which assumptions matter most. It does not yet establish which charity is best. A fair comparison needs the same outcome coverage and cost basis.

One USU is a stipulated six-hour kidney-stone pain reference of 47.25 intensity-hours. It is a reporting unit, not an empirically established equivalence between malaria and kidney-stone pain. How the unit works

Explore the estimate

Try a donation amount and scenario

Lower, central and higher scenarios change the money-to-health estimate as well as the illness course. The inputs below are our explicit choices, not measured averages for AMF’s future portfolio.

Why these choices?

DRC is an unusually high-burden setting. We reduce the central reference effect by 25% when extending it to unspecified future programs. The exact discount is a judgment, not a measured country adjustment. The lower scenario uses half the reference effect with 25% higher costs. The higher scenario retains the reference effect with 20% lower costs.

“Relative effect” combines possible differences in disease burden, realized prevention and allocation. We apply it after the source’s adjustments. It is not another estimate of the same monitoring shortfall. Holding the illness course at its central values, these cost and effect choices alone give about 51, 96 and 160 USU per $1,000.

Illness assumptions are shared with the Chad malaria analysis. They are rough full-course averages, including relief, sleep and reduced awareness. The combined scenarios are not confidence intervals, and actual impact can fall outside them.

Follow the calculation

From a documented grant to a future scenario

1. Use one recent, transparent reference

   GiveWell’s public 2026 DRC model estimates 7,395,818.973 under-five malaria episodes prevented in its core calculation. We apply its 22.0653% program-output reduction for grantee quality and other-funder effects, then divide by the matching $276 million grant.

   7,395,818.973 × (1 − 0.2206534809) ÷ $276,000,000 × $1,000
   = 20.8837 reference episodes per $1,000

   Applying this output reduction proportionally to episodes is a WSI assumption. GiveWell does not publish this as a USU estimate. This reference grant is already funded and is not the destination promised for a new donation.

2. Keep direct deaths separate

   The core model’s 64,397.05782 under-five deaths include indirect deaths. We divide by 1 + 0.5687 to isolate direct malaria deaths, then apply the source’s separate 22.8447% mortality reduction.

   64,397.05782 ÷ 1.5687 × (1 − 0.2284469461)
   ÷ $276,000,000 × $1,000
   = 0.114758 reference direct deaths per $1,000

   This assumes the adjustments preserve the direct-to-indirect ratio. The source uses an under-five population with a mortality rate for ages 1–59 months. We retain that approximation.

3. Make the future-program assumption visible

   Multiply reference outcomes by relative effectiveness, then divide by relative cost. The central scenario uses 0.75 ÷ 1.00, giving 15.6628 episodes and 0.086068 direct deaths prevented per $1,000.

4. Assign one complete course to each avoided episode

   For the central course, nonfatal illness is 4 days at intensity 3 and fatal illness is 4 days at intensity 5. A fatal course replaces the ordinary course for each direct death prevented.

   [(15.6628 − 0.086068) × 4 × 24 × 3
   + 0.086068 × 4 × 24 × 5] ÷ 47.25
   = 95.8 USU per $1,000

   Death itself, lost life years and bereavement receive no USU. Linear weighting, p = 1. We include only the acute illness suffering avoided.

What the dollars mean

A donation still depends on delivery partners

The earlier DRC worked example reports about 160 USU per $1,000 of full modeled delivery costs before final grant adjustments. This page instead uses grant dollars, includes the source’s adjustments, and adds explicit future-program assumptions. The two numbers should not be ranked against each other.

GiveWell’s DRC grant report explains how additional funding interacts with delivery spending and nets that other funders would otherwise provide. Our denominator is the $276 million grant matched to its model, not the retail price of a net or the entire nationwide campaign budget.

AMF says public donations buy nets, while other funding supports delivery and monitoring. We provisionally assume a new dollar has the same grant-equivalent role after the modeled co-funding and displacement effects. That equivalence has not been verified for AMF’s next allocations. We do not multiply the result again to credit cheaper net-only purchases.

If partner delivery funding does not materialize, nets arrive late, or another donor would have paid anyway, the result can change substantially. The scenarios illustrate some variation; they do not resolve these unknowns.

Funding snapshot · 11 October 2026

AMF reports about $44 million in immediate need

AMF’s funding page listed a $43.83 million immediate funding gap when checked. Its detailed table placed the remaining gap in two provisional programs whose countries were not disclosed. This is evidence of a reported organization-level need, not confirmation of a named open campaign.

The source is marked “Currently updating” and includes older background text. We use its dated immediate-gap table, retain the reported amount as a snapshot, and do not attach the DRC impact model to either unnamed program. We have not obtained direct confirmation from AMF.

What this estimate covers

Inspect and improve

Sources and reproducible inputs

1. GiveWell’s public 2026 DRC modelRecent documented reference, retrieved 11 October 2026. One campaign model, not a measured portfolio average.
2. GiveWell’s DRC grant reportApproved May 2026, published August 2026. Funding, counterfactual and co-funding context.
3. AMF’s funding pageFunding snapshot checked 11 October 2026. Reported need is separate from the modeled impact.
4. AMF, allocating donationsPublic net donations, distribution allocations and reassignment.
5. Malaria course assumptions and clinical contextShared provisional WSI profiles v0.1.

Future updates can replace the reference with campaign-specific or portfolio evidence, update costs, or refine the shared illness courses. Each change will preserve the previous version and explain its effect.

Prepared 11 October 2026 · AMF opportunity model v0.1 · Funding recheck due 10 November 2026 or sooner after a material change · Suggest a correction