Famine's standard mortality threshold was built for rural Africa - it misses mass starvation elsewhere

The global system for officially declaring famine rests on a single mortality number: two deaths per 10,000 people per day. That benchmark - embedded in the Integrated Food Security Phase Classification (IPC) as the threshold for Phase 5, the most severe designation - determines whether the international community treats a food crisis as a catastrophe requiring emergency response. A paper published in The Lancet argues that the number is wrong for many of the crises it is supposed to catch.

L.H. Lumey, a professor of epidemiology at Columbia University Mailman School of Public Health and an expert on historical famine, led the analysis with co-authors Ingrid de Zwarte (Wageningen University) and Alex de Waal (Tufts University). Their critique centers on how the IPC threshold was constructed and what it was built to detect.

A threshold calibrated to one context

"The mortality thresholds used by the Integrated Food Security Phase Classification were developed for rural African settings, not middle-income urban populations," Lumey said. "There are stark disparities in how famine mortality is assessed across contexts."

The IPC standard uses an absolute mortality rate - deaths per unit of population per day - across all age groups combined. This approach has two structural weaknesses the authors identify. First, baseline mortality rates differ substantially between populations. A two-per-10,000-per-day rate represents a very different scale of excess mortality in a country with low baseline death rates than it does in one with higher baseline rates.

Second, the all-ages aggregate can mask dramatic increases in specific vulnerable groups, particularly young children and infants. Famine characteristically strikes these groups first and hardest, but their deaths can be diluted in the total population figures and fail to move the aggregate rate past the threshold.

What the Dutch Hunger Winter shows

Lumey has studied the Dutch Hunger Winter of 1944-1945 extensively. That crisis, which struck an urban, middle-income European population under Nazi occupation, offers a documented case study of famine that may not have met the IPC's current standard.

In major Dutch cities, infant mortality in March 1945 climbed to four times its prewar level. Mortality among children aged one to four increased sevenfold. Births declined. These are the signatures of acute nutritional crisis affecting the youngest and most vulnerable. "These dramatic increases would not meet the current IPC famine threshold for children under five," Lumey observed - not because the mortality was not real, but because the absolute rate in an otherwise relatively healthy urban population did not cross the two-per-10,000-per-day line.

If the Dutch Hunger Winter would not qualify under current IPC standards, the classification system is missing something important about how famine unfolds in non-subsistence-agriculture contexts.

Mortality is a lagging signal

A broader problem is that mortality data, even when it is accurate, arrives late. People do not die of starvation immediately; they weaken over weeks to months. By the time death rates rise enough to cross a threshold, weeks or months of preventable harm have already occurred. Humanitarian response triggered by mortality data will always be delayed relative to the onset of the crisis.

"Identifying earlier indicators of famine stress could shorten the time between acute food insecurity and rising mortality," Lumey said. "A more sensitive and context-specific approach would support faster humanitarian action."

The paper also notes that mortality data is particularly unreliable in the settings where famine most often strikes. Access to conflict zones may be restricted. Governments with political incentives to minimize crisis severity may suppress or delay reporting. The IPC's reliance on this data as the primary criterion for the most severe classification creates a structural vulnerability that bad actors can exploit.

What the authors propose

The paper calls for a fundamental re-examination of how famine thresholds are defined, moving toward context-specific standards that account for baseline population health, age-stratified mortality data, and earlier indicators of nutritional stress such as declining birth weights, wasting rates in children, and reduced food intake surveys. These signals appear before deaths climb, providing a window for intervention that the current system forecloses.

The authors acknowledge that reforming international classification systems is slow and politically contested - the IPC involves multiple UN agencies and donor governments with varied interests. But the cost of the current system, they argue, is that mass starvation can persist unrecognized for months while the official designation waits for a mortality rate calibrated to a different world.