Building health surveillance that lasts

A health surveillance system that does not outlive the project that funded it has, in truth, never really existed. The real success of a public-health intervention is measured not by the quality of the dashboard delivered to the donor, but by what is still working three years after the consultants have left. Yet in West Africa, this test of time is rarely passed. Behind flattering indicators during the project phase, many systems stop dead the moment the final tranche of funding is disbursed. The problem is not technical, the tools exist and work: it is structural. It lies in how systems are designed, financed and handed over. This article sets out to look squarely at what makes surveillance hold, or collapse, once external money withdraws.

The foundation already exists: the region is no longer a blank page

We should start with good news, too often left unsaid. West Africa is no longer devoid of surveillance infrastructure: it has built one, and it is far from trivial. Two building blocks form its backbone. The first is Integrated Disease Surveillance and Response (IDSR), the strategy championed by the WHO for the African Region: 43 of its 47 member states have adopted the technical guidelines, 37 produce regular epidemiological bulletins, and more than 23,000 public-health workers have been trained in it since 2010. The second is DHIS2, the health information management software developed by the HISP Centre at the University of Oslo: it now equips the health ministries of 80 low and middle-income countries, covering countries home to 3.2 billion people, or 40% of the world's population. Benin, like most of its neighbours, has moved its national health information system onto DHIS2. The region therefore already has the rails. The question is no longer how to lay them, but how to keep the trains running once the funding that launched them stops.

The structural flaw: a system funded from outside does not stop with the project, it stops with the donor

If so many systems collapse, it is not through any negligence of the teams, but because they rest on a financial base that is not their own. The share of health spending funded by external aid is the most telling indicator here, because a surveillance system paid for by a project lives and dies to the rhythm of that project. In Benin, external aid covers 38% of current health expenditure in 2023 (World Bank, after WHO), one of the highest proportions in the region and more than double the sub-Saharan average (15%). By comparison, Ghana and Mali depend on aid for just 9% and 6% respectively. This dependence is no accounting detail: it means that in Benin, almost four francs in every ten spent on health come from a decision taken outside the country, and therefore revocable outside the country. A surveillance system anchored to that share mechanically inherits its fragility.

The second Achilles heel: too few hands to keep the system alive

A surveillance system is not a piece of software: it is first a human chain. Week after week, someone must count the cases, fill in the form, check the entry, transmit, analyse, raise the alert. Yet in most countries of the region this chain rests on a workforce that is rare to the point of scarcity. The WHO sets at 4.45 doctors, nurses and midwives per 1,000 people the threshold below which a health system struggles to perform its essential functions. Benin has roughly 0.8, Senegal 0.5, Niger and Mali fewer than 0.5. Only Ghana, at around 4.4, comes close to the threshold. This shortage is the most underestimated sustainability factor: you can train workers on a new collection tool for the length of a project, but if each worker already carries a crushing clinical load, the extra surveillance task is the first to be dropped as soon as external support ends. Workforce density is not a separate human-resources matter: it is the very condition for a data system to survive.

Framing the problem: surveillance is not about accumulating, it is about reporting on time

Before reaching for remedies, we must name precisely what is expected of surveillance. Its value lies not in the volume of data collected, but in two modest yet demanding qualities: completeness (all units report) and timeliness (they report on time). The WHO sets a benchmark of 80% for both indicators. Below that, an outbreak can smoulder for several weeks before it is seen. This distinction is decisive, because it shifts the success criterion. A system that collects fifty indicators with 50% completeness is in fact less useful than a system that tracks ten with 95% completeness. The temptation of projects runs the other way: multiplying indicators to impress, at the cost of a reporting chain that seizes up and ends up producing nothing reliable. Sobriety here is not a fallback, it is the condition of reliability.

The most useful health data is not the data that impresses the donor, but the data the field worker understands, knows how to produce on time, and keeps producing when no one is asking for it any more.

This demand for timeliness also explains why ownership matters so much. A system designed to satisfy a donor's logframe produces data turned outward, toward the report. A system owned by national teams produces data turned toward local decision-making: the nurse at the health centre who sees a cluster of malaria cases come up that very week does not wait for the annual review to act. It is this short loop, between the data and frontline action, that anchors use and, through it, guarantees continuity. You do not sustain a system through the constraint of reporting: you sustain it by making it useful to those who feed it.

Three conditions of robustness, tested in the field

The experience of systems that survive, set against that of systems that fade out, traces three conditions consistent enough to be stated as principles.

• A tight set of indicators. Ten indicators tracked faithfully, with completeness above the 80% threshold, are worth more than fifty abandoned along the way. Every added indicator carries a hidden cost of collection, verification and transmission that weighs on already scarce teams.
• Tools integrated into existing national systems, not parallel ones. A project system that duplicates the national circuit (DHIS2, IDSR) creates an additional burden destined to be abandoned. Integrating into the national system means leaning on an infrastructure that is, by design, meant to outlast the project.
• Training conceived as a transfer of skills, not a session. To train is not to gather workers for a three-day workshop: it is to equip national trainers, document procedures and organise the handover, so that competence stays in the country when the consultants leave.

The cost of inaction: paying dearly for surveillance that serves only once

The cost of a short-lived system is not limited to the lost investment. It is paid in blindness. A surveillance system that stops means an outbreak spotted too late, a measles or Lassa-fever flare-up detected only once it has already spread, a vaccination campaign calibrated on stale figures. In a region where health spending per person remains very low, 47 dollars in Benin and 26 in Niger in 2023 (World Bank), against 79 on average across sub-Saharan Africa, every franc counts twice over. Building a system that does not outlive the project means converting a prevention investment into money down the drain, then paying a second time, in crisis management, for what surveillance would have allowed to anticipate. The economic logic is clear: a durable system, however modest, costs less over ten years than a succession of ambitious systems that fade one after another. The real economy is not to spend little, but to spend once and for good.

On top of this depleted public spending comes a burden that falls directly on households. In Benin, 42% of health spending comes out of patients' pockets in 2023; in Togo, 64%; in Nigeria, 72% (World Bank). This high out-of-pocket share is not only an equity issue: it also weakens surveillance, because a patient who forgoes care for lack of means is a patient who appears in no statistic. Surveillance data is always, by implication, access data: you only see what presents itself to the health system.

What national averages conceal

One last point, decisive for action: national completeness and timeliness indicators are averages, and averages hide what matters most. A country may post national completeness of 85% while harbouring entire districts where reporting runs at 40%, because the health centre is remote, the network absent, the lone worker overwhelmed. Yet it is precisely in these under-reporting zones that an outbreak is most likely to smoulder unseen. The average reassures the donor; it lulls the decision-maker. Robust surveillance does not settle for a national rate: it maps its own holes, district by district, and concentrates its strengthening effort where the chain is weakest. Without this disaggregated reading, surveillance investment risks watering the zones already covered and leaving in shadow the ones that matter most.

This is the conviction that guides CRAD's work in public health. The firm has supported multi-year monitoring systems, from HIV to nutrition by way of malaria, for partners such as the Global Fund, ENABEL or Plan International. These engagements confirm a simple rule: a surveillance system is judged by its ability to produce reliable data repeatedly, without rising overhead, and carried by the national teams themselves. That is why our approach places ownership and the training of trainers before the sophistication of the tool. An elegant system that no one in the country can maintain is a deferred failure; a sober system, integrated into the national circuit and run by trained workers, is a lasting gain. Durability is not added afterwards: it is designed into the first line of the terms of reference, by asking a single, nagging and salutary question: what will still be running here the day we have gone?

Measure to decide, not to archive

Ultimately, surveillance only makes sense if it leads to action. Dashboards readable by a nurse and a district director alike, regular feedback to the teams that produced the data, feedback loops that turn a figure into a decision: this is what moves a database from the status of archive to that of a living tool. Data that is never returned to those who collect it ends up no longer being collected. The sustainability of a surveillance system feeds on its felt usefulness, in the field, to the very people who supply it. This is the most consistent lesson of two decades of engagements: you do not save a surveillance system by perfecting it, you save it by making it indispensable to those who keep it alive.

← All insights