[N.b. I used Claude to help format this post; all arguments are mine]

A response to Rethink Priorities HSS report  & F/U, with some inspiration from @NickLaing, @Berke and others. 

--

There is a category error sitting at the centre of the EA Forum conversation on community health workers (CHWs); workers who bring care straight to their neighbor's doorsteps. Rethink Priorities' HSS report models "CHW programs" as an intervention, scores them at roughly 1,000x  in Coefficient Giving terms, and notes they do not meet the 2,000x bar. The conclusion that follows is that CHWs are an interesting but second-tier bet.

But CHWs are not an intervention. CHW programs are the delivery platform through which most of the interventions EA already funds—bed nets, vitamin A, deworming—reach billions.

Modelling the platform against its own throughput suggests the model for the throughput wasn’t built to deliver.

A platform, not an intervention

A community health worker is not an intervention any more than a nurse is. Whether a CHW program is cost-effective depends, like any other delivery mechanism, on a) what it delivers and b) how it is designed.

What the platform delivers

A modern, professional CHW program  is a primary care delivery point, not a single-disease vertical. The WHO competency-based curriculum—the closest thing the field has to a current job description–sets out the portfolio: childhood illness (diarrhoea, pneumonia, malaria), maternal and newborn health, family planning, nutrition screening, immunisation support, TB and HIV case-finding, NCD screening (hypertension, diabetes), mental health first response, neglected tropical disease case management, and health promotion.

The same worker, in the same household visit, delivers whatever subset the program is designed for.

Any cost-effectiveness analysis that attributes the full fixed cost of that worker to one of those service lines is going to produce a misleading number.

Facility-based care alone cannot reach everyone. Hospitals struggle to serve remote communities, and in most high-burden countries the realistic counterfactual to a CHW is not a clinic visit — it is no contact with the health system at all.

Where researchers have compared CHW-delivered care head-to-head with facility-based care, the CHW route has been consistently cheaper. Most CHW programs evaluated for HIV, TB, malaria and for reproductive, maternal, newborn and child health were cost-effective against the facility comparator.

The median annual cost to deliver primary care via a CHW platform—across 380 scenarios in 130 studies—is US$0.59 per capita. That is not cost per DALY, and not cost per visit. It is the full annual unit cost, per person per year. 

The median cost per beneficiary is $10.03, but of course the cost varies by service package. Integrated horizontal platforms—many service lines through one worker—sit at a median of just $6.02 per capita per year to reach those not close to a facility. ^[1]

HIV, TB, malaria and RMNCH range more widely; in both areas, CHW programs were cost-effective in over 80% of scenarios assessed.

Bundling is where the savings compound. When Liberia folded NTD case management into an existing CHW platform, cost per diagnosis fell up to tenfold versus running NTDs as a standalone vertical.

The mechanism is platform economics. A CHW's fixed costs—salary, supervision, supplies, transport, data—exist whether she delivers one service line or ten. Vertical analyses load the whole stack onto a single disease; horizontal analyses spread it across everything the worker delivers in one visit.

How the platform is designed

Pool salaried, supervised, well-supplied, skilled CHWs together with volunteer, unpaid, under-trained ones, and the average tells you little. The "critical components" literature—including the work by yours truly that Rethink cites—has been clear for a decade: compensation, training, supervision, supply-chain integration, and career progression are not optional. Averaging two different things: poorly supported volunteer initiatives and high-performing professional platforms isn't consistent with the documented 4x variation in ROI.

Contrary to what Nick Laing's piece from April '25 implies, this isn't a workforce that is "often not cost-effective." Like nurse-led clinics—which Nick rightly champions!—it is a platform that is cost-effective when designed properly and not cost-effective when it isn't. 

The relevant question is not whether to fund "CHWs" but which design pattern to fund, and at what level of the system.

Rethink's way forward

For all the buzz about CHWs, the design pattern with the strongest evidence is the most neglected. A professional CHW platform is salaried (not volunteer), skilled (competency-based training to a published curriculum), supervised (relevant supervisor-to-CHW ratio), supplied (the commodities the job requires), and built into the public primary care system.

So the smart bet is the one Rethink itself names in its own report:  "Help fund and establish a public health program until it is transitioned to national government ownership."

For the whole vertical portfolio EA has spent fifteen years building—bed nets, SMC, vitamin A, deworming, child immunisation—that pathway is the scale opportunity. 

How countries can get there

How does a country move from volunteer CHWs to a paid, professional, government-owned workforce? Three conditions hold the problem in place, and an EA funder can act on each one.

Guidelines. Until recently, a health minister had no authoritative international consensus describing a well-designed CHW program That has changed. The WHO 2018 guideline, the aforementioned curriculum, and proCHW guidance co-authored by  Africa CDC and other norm-setters now give a minister both the cover to make a major workforce decision and the spec to make it well.

Start up capital . Even a willing government faces a chicken-and-egg problem: standing up a salaried workforce needs startup capital before the recurrent domestic budget line can absorb it. The Global Fund and others move money at that scale, often catalyzed by private actors.

Domestic constituency. Professional CHW associations are what turn a presidential commitment into a recurrent budget line that outlasts the president who made it.

Funding a delivery NGO to ship commodities saves lives now, and it is essential. Funding the workforce, the policy uptake and the government adoption that let a country deliver those commodities itself is the complementary lever that outlives any single donor.

This is a bet not dissimilar to hits-based giving: a large, durable return  (i.e. a permanent, domestically-financed platform that scales all the verticals and outlives any single donor) with outcomes hard to attribute to any one funder. 

It also has the one thing a hit almost never does: a decade of cost-effectiveness evidence already behind the intervention. The only open risk is whether a government finishes the handover, but if it does it pays for the whole portfolio. 

A closing thought

Berke framed the underlying choice well on the Forum: 

"Does success for EA look like hundreds of cost-effective NGOs distributing health commodities across South Asia and Sub-Saharan Africa until 2100 — or governments able to keep their people from dying of preventable conditions without depending on funders far away?"

For any funder serious about scale, it's the second.  And professional CHW platforms are the pathway.

1. ^{^}

   To my point in this article: the $0.59,$10.03, $6.02 figures sit on quite a bit of heterogeneity! They are medians across 130 studies, with wide variation in setting, package, and study quality. I think the central tendency is robust enough to act on. I don't think any single number should be quoted without that caveat, and I'd welcome a re-analysis that weights by study quality.