A diagnostic supply chain that has never run out of test kits has usually erred the other way: it has thrown away boxes that expired on a shelf before anyone opened them. Both failures trace back to the same weak link — quantification, the unglamorous forecasting work that decides how many rapid diagnostic tests (RDTs) and reagents a program actually orders. Programs that source and procure validated diagnostics for field use are, in practice, running a demand-planning exercise as much as a procurement one, and the two rarely get equal attention.
Quantification sits upstream of every other procurement decision covered elsewhere in this space — vetting a supplier, checking a shelf-life date, confirming cold-chain compatibility. None of that matters if the order quantity itself was wrong.
Why Diagnostic Test Kit Forecasting Is Its Own Discipline
Quantification and forecasting is the process of estimating how many units of a health product a program will need over a fixed period, then translating that estimate into a procurement and delivery schedule that keeps stock available without keeping it too long. For medicines with stable, well-understood use, this is close to routine inventory math. For diagnostics — especially RDTs deployed against a fluctuating disease signal, an outbreak, or a seasonal campaign — the underlying demand itself is the hard part to pin down. A malaria RDT order sized for an average transmission season can be badly wrong the year transmission spikes or a neighboring program reallocates its own surplus.
The Methods Programs Actually Use
Global health procurement has settled on a small set of forecasting methods, each suited to a different data situation. None is universally correct; the choice depends on what a program actually knows about its own consumption and caseload.
| Method | Best used when | Key limitation |
|---|---|---|
| Consumption-based | Past usage data exists and testing protocols are stable | Inherits any past stockouts or reporting gaps unless explicitly adjusted for |
| Morbidity-based | A new program or product with no consumption history | Depends on accurate disease-prevalence and attendance estimates, which field programs may not have |
| Proxy consumption | No reliable local data at all | Borrows assumptions from a different program or country context |
| Service-level forecasting | The number of planned tests or services is already fixed | Only as accurate as the service target itself, which can shift mid-cycle |
In practice, most programs blend two of these — using consumption data where it exists and falling back to a morbidity or proxy estimate for a new test type or a newly reached population.
The Tools Behind the Numbers
Because the arithmetic gets complicated fast — multiple products, multiple sites, staggered delivery lead times — quantification is rarely done on a spreadsheet alone. John Snow, Inc. (JSI) developed PipeLine, one of the longer-standing tools used to track usage rates, monitor stock on hand, and flag an emerging shortfall or surplus before it becomes a crisis; Management Sciences for Health built a comparable tool, Quantimed. Disease-specific calculators exist as well — the Clinton Health Access Initiative’s Simple Tool produces a morbidity-based forecast over a multi-year horizon, and an Advanced HIV Disease Commodity Calculator supports quantification for that particular diagnostic and treatment bundle. None of these tools removes the underlying data problem; they organize it so a program officer can see where the numbers stop making sense.
Where the Forecast Breaks Down
Even a well-run quantification exercise fails in a handful of predictable places, and this journal’s own coverage of cold-chain and field-diagnostic sourcing keeps circling back to the same root causes:
- Expiry-linked wastage. Overstocking a diagnostic with a limited shelf life is not a safe hedge — it is a slower-motion stockout, because the excess is unusable by the time anyone reaches for it.
- Long, uneven lead times. The gap between placing an order and a shipment clearing into a field site can run to months, which means the forecast has to be right well before the need is visible.
- Thin or inconsistent data. Reliable consumption and prevalence figures are often the scarcest resource in the settings that need them most, which pushes programs toward proxy estimates with wider error bars.
- Demand that will not sit still. An outbreak, a policy change, or a shift in which test a national guideline recommends can invalidate a forecast built only months earlier.
Underneath all four is a structural issue: diagnostic demand planning is typically financed and procured in a fragmented way, without the steady demand signal that would make forecasting easier — a problem more visible in diagnostics than in commodities with longer histories of centralized planning.
Building a Buffer Without Building a Landfill
The practical response is not to forecast harder in isolation but to pair the forecast with a deliberate buffer strategy: a safety stock sized to cover lead-time variability, a reorder point triggered before the buffer is exhausted, and — critically — a review cycle short enough to catch a forecast going stale before the next order locks it in. Programs that treat quantification as a one-time exercise at the start of a grant cycle tend to be the ones surprised by either outcome; programs that revisit it quarterly, against actual consumption, are the ones that avoid both a shelf-life write-off and a stockout in the same year.
Conclusion
Diagnostic test kit forecasting will never be an exact science in a field setting — disease incidence, funding cycles, and supply-chain lead times all move independently of each other. But the discipline of choosing a defensible method, matching it to the data a program actually has, and revisiting it against real consumption is what separates a shelf that is quietly stocked from one that is either empty or full of expired boxes. For programs procuring validated test kits and reagents, that discipline belongs at the start of the sourcing process, not as an afterthought once the order has already shipped.