Worked example · Model v0.1
Preventing malaria with mosquito nets in DR Congo
An estimate of young children’s acute malaria suffering prevented by insecticide-treated nets, using the same illness assumptions as our Chad medicine example.
Funding committed. The grant was approved in May 2026. Some repeat campaigns are conditional. Additional funding for this specific program has not been verified.
Estimated acute-malaria USU averted
≈160per $1,000 in full modeled delivery costs
malaria episodes prevented
direct malaria deaths prevented
All outcomes are per $1,000. Core program-model estimates before final grant-level adjustments. Gross acute-malaria benefit; adverse effects and long-term effects are unestimated.
Reading the result
Count the illness that could be avoided
The model estimates how much acute malaria in young children a net distribution could prevent over three years. We convert the avoided episodes into USU, using four days at intensity 3 for nonfatal illness and four days at intensity 5 for fatal illness.
These are the same provisional courses used for Chad. One USU is the stipulated kidney-stone reference of 47.25 intensity-hours. A direct death prevented identifies a complete fatal course avoided; death itself, lost life years and bereavement are not assigned USU.
Compare the two malaria examples →The calculation
From delivered nets to USU averted
Start with full delivery costs
$1,000 ÷ $4.95 per delivered net = 202.02 net equivalents. The cost includes procurement and delivery resources, not just the price of the net.
At 1.8 people per net and a 15.4973% under-five population share, that represents 56.35 young children in the source model. These are scaled program quantities, not a promise about a particular $1,000 donation.
Estimate episodes prevented over three years
The source’s expected reductions are 27.60%, 12.83% and 4.29% in years one, two and three. They already account for modeled coverage and protection declining over time.
56.3537 children × 1.006405514 episodes per child-year
× (0.2760480807 + 0.1282576019 + 0.04291177467)
≈ 25.36 episodes preventedThe annual reductions are summed across separate years. They are not added as simultaneous probabilities, and the three-year benefit horizon does not mean three years of suffering per episode.
Separate direct deaths, then assign one course per episode
The source mortality rate includes indirect deaths. We divide its adjusted rate, 0.008762998972, by 1 + 0.5687 indirect deaths per direct death. Applying the same population and annual reductions gives 0.1408 direct deaths prevented.
This decomposition assumes the mortality adjustments preserve the source’s direct-to-indirect ratio. The source uses a 1–59-month mortality rate with its under-five target population. We retain that approximation.
(25.3638 − 0.1408) × 6.0952 USU
+ 0.1408 × 10.1587 USU
= 155.17 USU, rounded to ≈160USU per illness course = days × 24 × intensity ÷ 47.25. Linear weighting, p = 1. The fatal group is removed from the ordinary group, so each avoided episode receives one complete course.
Source cells and reconciliation
Inputs are from GiveWell’s public 2026 DRC ITN model, Main CEA, DRC column K. The spreadsheet’s country columns are model alternatives; they are not additional programs to sum into this grant.
| Input | Value | Unit | Cell |
|---|---|---|---|
| Full cost per net delivered | 4.95 | USD/net | Main CEA!K28 |
| People targeted per net | 1.8 | people/net | Main CEA!H35 |
| Under-five share of target population | 0.1549727233 | fraction | Main CEA!K50 |
| Year 1 expected incidence and mortality reduction | 0.2760480807 | fraction | Main CEA!K88 |
| Year 2 expected incidence and mortality reduction | 0.1282576019 | fraction | Main CEA!K92 |
| Year 3 expected incidence and mortality reduction | 0.04291177467 | fraction | Main CEA!K96 |
| Adjusted incidence without ITNs | 1.006405514 | episodes/child-year | Main CEA!K122 |
| Adjusted direct and indirect mortality without ITNs | 0.008762998972 | deaths/child-year | Main CEA!K107 |
| Indirect deaths per direct death | 0.5687 | ratio | Main CEA!K99 |
| Mortality effect / incidence effect | 1 | ratio | Main CEA!H83 |
| Full modeled spending for the scaled example | 291,589,543.3 | USD | Main CEA!K16 |
| Modeled under-five episodes prevented | 7,395,818.973 | episodes | Main CEA!K128 |
| Modeled under-five direct and indirect deaths prevented | 64,397.05782 | deaths | Main CEA!K110 |
| Modeled under-five target population | 16,432,154.77 | children | Main CEA!K55 |
As an independent check, the source’s 7,395,818.973 under-five episodes prevented divided by $291,589,543.3 of matched modeled spending gives the same episodes per $1,000. Dividing its 64,397.05782 direct-plus-indirect under-five deaths by 1.5687 and by the same cost denominator reproduces the direct-death estimate.
The $291.59 million is the resource cost of the spreadsheet’s scaled example. It is not a claim about the full national campaign budget. Source dollar values are retained without inflation adjustment.
Why this differs from the grant’s headline impact
The grant report describes a $276 million commitment and approximately 67,000 deaths prevented across all ages under its final model. Our calculation instead uses the core delivery-cost model, isolates young-child outcomes and removes indirect deaths. The figures answer different questions.
The source’s final grant calculations also include monitoring-quality adjustments, funding displaced from other contributors and supplemental effects. We do not apply those final adjustments or GiveWell’s moral weights to this benchmark. The Chad pilot likewise uses its core program calculation. Both should be read as conditional program examples, not net marginal-donation estimates.
Core disease-risk and intervention inputs remain in the calculation, including existing net coverage, insecticide resistance and the modeled vaccine adjustment to mortality. Some underlying studies and cost worksheets are unpublished, so the public model is auditable at the selected-cell level rather than every original data record.
Shared malaria assumptions
How much could the illness assumptions change the result?
Both programs use one versioned set of illness courses. Intensity is a time average that includes relief, sleep and reduced or absent awareness, not continuous peak pain.
155 acute-malaria USU averted per $1,000
Central: nonfatal course, 4 days at intensity 3; fatal course, 4 days at intensity 5.
| Scenario | Nonfatal days × intensity | Fatal days × intensity | USU / $1,000 |
|---|---|---|---|
| Lower | 2 days × 2 | 2 days × 3 | 52 |
| Central | 4 days × 3 | 4 days × 5 | 155 |
| Higher | 7 days × 4 | 7 days × 7 | 362 |
52–362 is a mapping-only sensitivity range, not a confidence interval or full uncertainty bound. Costs, disease burden, distribution quality and effectiveness can also differ substantially.
Neither four days nor intensities 3 and 5 are measured averages for this population. We reuse the clinical context and explicit judgments documented in the Chad malaria report; we have not independently calibrated a DRC suffering profile.
Coverage and funding
What this first comparison leaves out
Benefits beyond young children
The full net-program cost is included, but suffering and deaths prevented among people aged five and older are excluded. That omission matters when comparing a household intervention with SMC, which specifically targets young children.
Indirect deaths, long-term complications, caregiver distress, bereavement and income effects are also outside this calculation. Net-related adverse effects are unestimated. This is a gross acute-malaria benefit, not a guaranteed lower bound or total program value.
Future delivery and additional funding
The model covers distributions planned for 2027–29 and counts benefits over three years after each distribution, which can extend beyond 2029. The grant includes a conditional repeat-campaign component.
This funded example does not verify that a new donation today would finance the same program or have the same effect. A current opportunity requires an unmet funding need and a credible estimate of what would change with extra funding.
Inspect and reproduce
Sources and calculation files
- GiveWell ITN CEA [DRC 2027–29], 2026 public modelMain CEA, DRC column K; selected input values retained in the download.
- GiveWell, AMF ITN campaigns in DRC 2027–29Grant approved May 2026; report published August 2026.
- Shared malaria course rationale and clinical sourcesSame assumptions as the Chad pilot, without a new local calibration.