Leprosy Mailing List – September 28, 2026
Ref.: (LML) Last mile outcomes in Leprosy care, pushing the value chain to the edge. Part 2.
From: Arie de Kruijff, Kijabe, Kenya
____________________________________________________________________________
Note editors:
Last week, September 23, 2026, we published the first part of Arie de Kruijff’s essay “Last mile outcomes in Leprosy care, pushing the value chain to the edge”. We thought this excellent and thoughtful but lengthy contribution better be divided in two parts to improve readability.
Last mile outcomes in Leprosy care, pushing the value chain to the edge. PART TWO OF TWO · CONCLUSION
Paying for activity, hoping for outcomes
The money sits underneath all of this, and it is where the model is most visibly misaligned. Leprosy control is almost everywhere funded as activity: a training delivered, a campaign mounted, a supervision visit completed. Government budgets cover the staff establishment and little else. The operational money that does flow comes largely from NGOs, often from non-designated funds rather than institutional line items, and it is billed against outputs — people trained, people screened — while the outcome everyone actually wants — a case found before disability, a contact examined, a treatment course completed — goes unmeasured and unpaid.
Several structural consequences follow. Output is paid for while outcome is merely hoped for. Short funding cycles produce short horizons, in a disease that demands multi-year follow-up and surveillance measured in decades. Activity funds, when they stop, leave no budget line and no institutional memory behind them. And the sector is caught in a loop of its own making: outcomes are not funded because they are not verifiable, and they are not verifiable because the information system is paper-based and geographically dispersed.
That information system is the clearest illustration of the whole problem. Leprosy is notifiable, yet in many countries the primary record remains a paper clinic card held at a facility, with only aggregate numbers travelling upward. Four consequences follow. Verification is expensive, because confirming a register entry means a physical visit — a 400-kilometre drive to flip through a card box — and therefore happens rarely. The data is not case-based at higher levels, so no one can follow an individual through treatment. It is hard to map, which makes hotspot identification guesswork. And confidence in the numbers is limited, including among the health officials who must use them.
The more instructive reading is that the information system is not merely a reporting problem. It is the mechanism by which leprosy service quality is made visible — or not — to the people who allocate money. That is a business-model question as much as a technical one.
What the new technology could change
This is where artificial intelligence becomes relevant, though the field is over-supplied with speculative language and under-supplied with field evidence. The genuinely practical capabilities are unglamorous: reading a paper form accurately from a phone photograph, extracting and geocoding a location, checking a record for internal inconsistency, and co-ordinating a small set of follow-up actions between named people.
Applied to leprosy, the concrete possibilities are these. Case-based notification can become a by-product of the clinical encounter rather than a separate administrative task performed months later: a health worker photographs the card, the software extracts the data, the worker verifies and corrects it, the address is pinned to a map and the record is confirmed into the national structure. If notifications carry location, hotspot identification becomes a standing capability rather than a special study. The system can flag what a busy clinic cannot — a missing disability grade, an ulcer with no follow-up, a reaction recorded without a treatment plan, gaps in a monthly treatment record. And if the trail of service delivery is captured digitally, verification stops being an expedition and becomes a trace of the work itself.
The concerns here are very real: Connectivity remains a real constraint in the places leprosy concentrates, and offline-capable workflows is a requirement, not a feature. The accuracy of card-reading software is not yet proven across countries’ differing forms, and should be measured before any operational claim is made. Leprosy is a stigmatised disease, so data protection is not a compliance detail but a condition of doing the work at all. Digital verification has to earn the trust it claims; a visible digital layer over a weak data foundation can manufacture false confidence rather than real assurance. And none of this removes the need for human judgement — the realistic design keeps a clinician at the helm, with software supplying the memory and the co-ordination.
The honest characterisation is that technology plays the platform role, not the doctor role. It can make peripheral work visible and therefore financeable. It cannot perform the work, and it should not be asked to pretend otherwise.
The lessons are already on the record
None of this ”pay for verified results at the edge” thinking is new in public services. Results-based financing has a two-decade history, and its examples are worth recalling mainly to establish that the idea is not exotic. Argentina’s Plan Nacer programme, launched in 2004, tied part of its funding for maternal and child health to a set of verified clinical indicators, with payments flowing onward to the facilities that produced the results. Performance-based financing has run in health systems across dozens of countries. A development impact bond for maternal and newborn care in India tested the structure that lets private investors pre-finance delivery and be repaid only on verified outcomes. And outcome-linked funding of community health worker networks has proved, at least in principle, that a peripheral network can be the contracted party.
The accumulated lessons from that literature are hard-won and bear repeating, because each one maps onto leprosy with unusual precision. Government ownership decides whether a scheme survives; those embedded in national financial management persist, while those living in donor project units collapse when the project ends. Metrics distort behaviour — pay narrowly and providers tunnel-vision onto the paid indicator and cherry-pick the easy cases. Outcomes must sit within the provider’s control and be measurable in a reasonable window, which for a slow disease argues for rewarding verified service events rather than distant epidemiological shifts. Verification cost decides feasibility. And pre-financing decides participation: small facilities and volunteer networks cannot bankroll months of work awaiting payment, so the poorest actors are exactly the ones who need capital provided up front.
That literature neither proves such an approach would work in leprosy nor proves it cannot. What it establishes is that the mechanisms exist, that their failure modes are well understood, and that nobody has yet made the serious attempt to adapt them to a neglected disease whose defining problem is the last mile.
Where the argument will meet resistance
A candid account must name the tensions, because pretending they do not exist is how good ideas die quietly.
Shifting value and decision-making toward the last mile touches established roles, reporting lines and budget control, and the reasonable question from a national programme is not whether the idea is good but who is accountable when it goes wrong. A decentralised model that leaves the centre without the information it needs will not survive its first supervision visit; the objective is not to route around central structures but to make their job easier while the work is done closer to the patient. Incentives, if they reach frontline actors, have to be transparent and able to survive a change of government. The question of what happens when external funding stops is the one that determines whether any of this is real — a design that only works while the grant lasts is a project, not a service model. Data that makes health workers and patients more visible also makes them more exposed, and consent, ownership and the limits of automated judgement must be settled before, not after, deployment. And technology cannot substitute for capability: no application delivers disability care where no one has been trained and no supplies exist.
Questions for reflection
What remains is less a proposal than a set of questions the sector has not yet answered, and which may usefully be argued over rather than resolved.
Which outcomes are sufficiently meaningful, measurable and within a provider’s control to be worth paying for — and which should explicitly not be reduced to a metric? What concrete mechanisms would keep value at the edge, where the work happens, rather than letting it be captured by the centre? Which functions genuinely improve when pushed outwards, and which must stay central for safety and equity? What independent evidence would be needed before trusting automated verification in a national programme? And what would a health department need to see — in cost, evidence and exit routes — to pilot a different model in a single district, and to keep it standing if every external funder left?
The ambition behind the questions is deliberately modest in means and demanding in aim: to give the people who already do the work the tools, the information and the recognition that actually reach them, and to let the value they create finally reach the people affected by the disease.
Sources
World Health Organization, Leprosy (Hansen's disease) fact sheet; and Global leprosy (Hansen disease) update, Weekly Epidemiological Record.
World Health Organization, Towards zero leprosy: global leprosy (Hansen's disease) strategy 2021–2030 (2021); and WHO technical guidance on contact tracing and post-exposure prophylaxis.
World Bank evaluations of Argentina's Plan Nacer / Sumar programme (Gertler et al., 2014, and subsequent literature).
Documentation of results-based financing and development impact bond programmes, including the Utkrisht maternal and newborn care bond (India) and outcome-linked community health worker funding (Living Goods).
Grover, D., et al., “Using supervised learning to select audit targets in performance-based financing in health”, PLOS One (2019), on machine-learning targeting of verification audits.
____________________________________________________________________________
LML - S Deepak, B Naafs, S Noto and P Schreuder
LML blog link: http://leprosymailinglist.blogspot.it/
Contact: Dr Pieter Schreuder << edit...@gmail.com
You received this message because you are subscribed to the Google Groups "Leprosy Mailing List" group.
To unsubscribe from this group and stop receiving emails from it, send an email to leprosymailinglist+unsubscribe@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/leprosymailinglist/4cfda758-9d4a-45a0-8779-e4f0fb045aa7n%40googlegroups.com.