Appreciate it, Toby. Have done so! Thanks for the LEEP rec. My outside impression was that the gravity pulls hard toward the measurable end over the further-out bets (mass movements, narrative, constituency-building). Am curious who is thinking about health delivery policy change and welcome all such leads.

Would also love to explore the spectrum idea. My sense is there a) are a couple of axes that correlate: how upstream the lever is (service delivery → rules → institutions) and how easily you can measure and attribute the result (measurable/attributable → diffuse/long-horizon).  B) That the  distribution is lopsided…but I am interested in outliers! Particularly for reasons of scale and/or permanent norm change (à la abolition example) 

Thanks, Justin for the question about Liberia. Two responses, and a question for you/where I think we converge:

1. "Domesticated" requires policy ownership + budget commitment + workforce embedded in government. Liberia has all three. Co-financing ≠ failure of domestication---every health system is co-financed. Countries routinely face fiscal-space problems and (post-2025) are in one now. That the gov't is doubling down on the program in the face of this stress + trying to find a way forward would seem to prove the ownership and durability point.  
2. Liberia is evidently not the only example. Next door in Cote d'Ivoire cost per person fell 20% per year post proCHW policy adoption, then a Prime Ministerial directive opened fee-free care to 13 million people, World Bank covered the first months, domestic financing took over (and all of this during the current contraction). Kenya is another recent example: 100k CHWs onto monthly stipends plus insurance, under domestic commitment, post-aid-cut announcement. Ethiopia is decades of domestic commitment etc etc.
3. Agreed on your taxonomy and would suggest we probably we want both: meet direct needs now via NGOs and ensure we can meet them even more cost effectively in the future via policy change. The latter requires modelling handover probability;  a long time horizon on the benefit side (i.e. credit DALYs averted across the program's full multi-decade lifetime, post-handover incl, discounted for durability risk); and (probably) unit cost modelled as a variable that integration can drive down? Do you know of anyone working on similar models in other issue areas?

Thanks so much, Justin, for taking the time to read and engage w/these great questions! A couple of thoughts:

1) Re: example of ntl level evidence: I used to live/work in Liberia and it's a great example of such a transition. They adopted proCHWs based on advocacy + pilot evidence from Konobo district post ebola in 2016. Malaria prevalence then fell from28% (2011) to 10% (2022) across nationally-representative surveys. Multi-driver (nets, seasonal malaria chemoprevention, artemisinin therapies, the Liberian program all contributed), the platform is the delivery channel. David Walton of PMI said it was one of the fastest declines they ever documented. 

On a more meta scale: 15+ years ago only a handful of countries had proCHW policy (Brazil, Ethiopia). People used to laugh at the suggestion :) Now 55+ countries have such policies following coordinated, multi-country 10 yr campaign targeting international guidelines, funding, and ntl policy. Contribution, not attribution etc etc but after 100 yrs of basically no movement whatsoever on this, it's something.

2. Predictors: Literal PhDs written on this! (Finnemore & Sikkink on norm diffusion = excellent). We at CHIC have two realist synthesis papers on this in process (one on how CHW associations organize to shift government behavior, e.g. mechanisms = a combined legal/association/protest pressure in a political opening), one on the political economy of domestic CHW financing, e.g. mechanisms = design choices at program inception etc.) 
Some common ingredients: a ministerial champion;  a unified professional voice/constituency (e.g. see Chenoweth & Stephan on the 3.5% threshold);

3. Counterfactual: Not sure I agree. The 2015 Konobo baseline (from the Liberia example above): a near-total dose-response collapse with distance to facility...odds of even one antenatal visit dropped to 0.04 in the farthest quartile. i.e. the hard-to-reach half of rural sub-Saharan Africa is closer to "no formal contact" than to "competent ministry already delivering care."

4. A broader observation: The donor's operative counterfactual is policy stuck vs policy moves: paper policy plus volunteer workforce plus indefinite donor dependence, OR a 5–7 year catalytic grant plus the four predictors gets a country onto a professional workforce on a recurrent domestic budget for 30+ years. Liberia made the transition; the 32% to 10% drop is the outcome.

We are all well served by remembering: 
a) Unit cost is a lever, not a constant: antiretroviral therapy went from 10k to 90 bucks per patient per year in a decade - due to funded, coordinated efforts 
b) Domestic financing is also a lever: the proCHW dashboard I linked above shows proCHW policy does not track w/gdp per capita (e.g. if Liberia did it...!).

Yes to the EA-style cost-effectiveness analysis! Would love help building it. If anyone is interested in this or wants to offer a view on what it needs  , I'd love to connect. 

Many thanks again and hopefully talk more!

Thanks, Nzube, for taking the time to read and share your experience. Would love to hear more of your thinking re: offsetting social losses

Appreciate it, Albert! The point your making about health being political and not just technical is well taken

Thanks so much, Nick! We published a massive multi-paper cost-effective review only weeks after you shared your original post! You can dig into all the papers/numbers/methodology here: https://joinchic.org/resources/cost-effectiveness/ 

TL;DR: The numbers are medians across 380 scenarios in 130 studies. As I said in the footnote, data is heterogeneous (mostly due to platform design differences, including salaries). Ranges are big and depend on context, but would say a) your $0.95 BOTE is a fair floor for a professional program at modest catchment, but b) that your earlier take (CHWs are "often not cost-effective," they treat "few patients per month," they can only treat "a handful of conditions in young children") is not reflected in these data.

Re: DALYs: you'll see in the papers that vertical evidence is already in EA terms (e.g. there are DALY numbers for iCCM) but none of the 42 horizontal integrated scenarios in the BMJ GH piece report cost per DALY. The research  I'd most want EA to fund is a prospective platform-economics evaluation: does bundling 6–8 service lines on one salaried CHW compound DALYs and lower cost-per-DALY the way fixed-cost-sharing predicts? (or to your point, maybe not?)

My bigger point though is that maybe this is already a good enough bet based on the vertical DALY evidence we both already accept (i.e. why not apply hits-based giving on the handover: catalytic financing to professionalize a salaried, skilled, supervised, supplied workforce via national, government). Esp, as this has what a hit almost never has, which is a decade of cost-effectiveness evidence already behind it. If the handover lands, the donor pays once and the government runs the whole portfolio forever (and the vertical $/DALY figures are the floor). 
Have added 2 footnotes re:the above.

D'oh! Thank you, I've added to the first sentence :)