Famine is still being measured as if the world has not changed. By relying on fixed mortality thresholds, today's classification systems risk recognising mass starvation only when it is already too late. Ingrid de Zwarte argues that this approach systematically underestimates famine in modern conflicts and calls for a fundamental rethink of how we define and measure extreme hunger.
In a Correspondence in The Lancet, De Zwarte and peers set out why current benchmarks fail to capture famine's true impact. Mortality figures may look objective and decisive, but they are blunt instruments. They often reflect the final stage of a crisis rather than its onset. By the time death rates cross an official threshold, months of deprivation may already have passed.
Gaza exposes structural weaknesses
The recent case of Gaza illustrates the problem. In August 2025, following months of repeated warnings from humanitarian organisations and medical professionals, the Integrated Food Security Phase Classification (the global system that categorises food insecurity into five phases) determined famine in Gaza Governorate. In December, that classification was reversed when key indicators dropped below the official threshold.
For De Zwarte, Gaza has revealed structural weaknesses in the current classification system. The IPC Phase 5 threshold for famine requires mortality to exceed two deaths per 10,000 people per day. In a lower middle income, highly urbanised area such as Gaza, overall mortality would have to increase more than twentyfold to reach that level. In rural Somalia or South Sudan, a six or sevenfold rise would suffice. The same numeric standard thus produces starkly different practical consequences across contexts.
Such disparities stem from the origins of the system. The mortality benchmarks were originally designed for rural African emergencies. Applying them universally assumes that all societies have similar baseline mortality patterns, demographic structures and health systems. That assumption does not hold.
Mortality comes late
De Zwarte draws on historical evidence to show that mortality is a late indicator of famine. During the Dutch Hunger Winter of 1944-45, food rations fell below subsistence levels months before death rates began to rise. Starvation unfolded gradually, first affecting body weight, strength and health long before it was visible in mortality statistics.
Other early warning signals, such as declining birthweights, are absent from current classification systems. Birthweight is a sensitive indicator of maternal malnutrition. When babies are born smaller, it often reflects nutritional stress that has been building for months. Ignoring such indicators narrows our field of vision.
Age patterns in famine: lessons from history
Historical famines also challenge assumptions about which groups are most at risk. Drawing on cases from the Netherlands, Ukraine, Greece and Sudan, De Zwarte shows that relative mortality increases are just as important to take into account as absolute mortality rates.
"The IPC focuses primarily on children under five, using a threshold of more than four deaths per 10,000 children per day. Yet in many historical famines, mortality among children aged five to fourteen rose more sharply in relative terms. During the Dutch Hunger Winter, infant mortality increased fourfold and mortality among children aged one to four increased sevenfold. Even so, in absolute terms the Dutch famine would not have crossed today's IPC famine threshold for under-five mortality".
Looking only at absolute mortality numbers can therefore obscure how sharply conditions are deteriorating within specific groups. Relative increases, which measure how much mortality rises compared with normal levels, may provide a more context-sensitive signal.
The risk of politicisation
In conflict settings, data are rarely neutral. Demographic information may be incomplete, restricted or manipulated. When famine recognition depends strictly on mortality thresholds, governments and armed actors may have incentives to contest or delay the data that would trigger international action.
De Zwarte warns that this creates a structural risk of under-recognition. If classification systems rely on indicators that are slow to move and politically contested, large scale starvation can remain officially invisible.
Towards more sensitive indicators
For that reason, De Zwarte and her co-authors advocate the systematic collection of more sensitive famine indicators. Earlier and more nuanced measures would provide a timelier and more accurate diagnostic tool for humanitarian intervention.
The correspondence in The Lancet was written together with Alex de Waal and L H Lumey. De Zwarte is Associate Professor in the Economic and Environmental History group. By bringing historical evidence into a contemporary debate, she challenges policymakers to reconsider what it means to declare famine and, crucially, when to act.
