<aside>

Purpose:

This page summarises the independent cost-effectiveness assessment of IMPALA by Rethink Priorities: what they concluded about the evidence, how they discounted it, and what their model shows. GOAL 3 commissioned and funded the work but does not control its conclusions.

How to use this page:

• Read the first two paragraphs for RP's view of the evidence and its limits.
• The table shows the results against three funders' thresholds.
• For the observed results RP discounts, see IMPALA’s impact at a glance For what the discounted effect means per ward, see Funding impact: what one implementation delivers.
  • For the full report and model, see the website from Rethink Priorities: https://rethinkpriorities.org/research-area/cea-goal3-impala-system/

This page was last updated: October 6, 2026

</aside>

In August 2026, GOAL 3 commissioned Rethink Priorities (RP), an independent research group, to evaluate GOAL 3’s evidence base and model IMPALA's cost-effectiveness from an outside perspective. The report by researchers Jewell and Clare was published in October 2026. GOAL 3 funded the work and does not control its conclusions.

On the evidence, RP’s conclusion is that it points consistently towards a mortality benefit without establishing its size. They found the consistency across Malawi, Rwanda, and Tanzania harder to explain by chance than any single result, and the mechanism plausible against a weak counterfactual. They also set out the limitations: the evidence base is non-randomised and only partly controlled, several estimates cannot rule out no effect, and the largest paediatric figure depends heavily on spillover to children who were never monitored. GOAL 3 accepts this assessment.

Their model then discounts the observed mortality effects by 60% for internal validity and a further 30% for external validity, following the approach GiveWell uses for evidence of this type. That produces a modelled true effect of approximately 12% relative mortality reduction in paediatric wards and 7% in neonatal units. RP’s cost assumptions are equally conservative: maximum expected cost on every line item, no volume discount, and no credit for morbidity reduction or cost savings.

After all of those discounts, IMPALA cleared every funder threshold tested:

The finding that matters most is what drives that result. It is the low cost of the system rather than a large assumed effect. The break-even mortality reduction, below which IMPALA would stop clearing the bar, is around 1% in paediatric wards and 2% in neonatal units. Both sit far below even their heavily discounted central estimates. The system is cheap enough that the conclusion survives being substantially wrong about the effect size.

Click here for the published report on the website from Rethink Priorities.