Recent global crises have exposed the limits of a universal mortality threshold for declaring famine—an approach that can obscure how famine actually unfolds across different populations. In a paper published in the Lancet, researchers at Columbia University Mailman School of Public Health and colleagues call for a fundamental re-examination of how famine thresholds are defined.
"The mortality thresholds used by the Integrated Food Security Phase Classification (IPC) were developed for rural African settings, not middle-income urban populations," said L.H. Lumey, MD, PhD, Columbia Mailman School professor of Epidemiology. "There are stark disparities in how famine mortality is assessed across contexts."
Lumey and co-authors argue that widespread starvation can remain officially unclassified for prolonged periods because it does not meet the IPC's Phase 5 benchmark of two deaths per 10,000 people per day. As a result, mass starvation may go unrecognized until it is well advanced.
The authors also note that the IPC relies on absolute mortality rates, while overlooking sharp relative increases within specific age groups. Evidence from the Dutch Hunger Winter—an event extensively studied by Lumey—illustrates this point. The crisis was marked by declining birth weights and fewer births in the population, followed by a dramatic rise in child mortality. In major cities, infant mortality in March 1945 climbed to four times its prewar level, while mortality among children ages one to four increased seven-fold. Yet, as Lumey observed, "these dramatic increases would not meet the current IPC famine threshold for children under five."
Moreover, mortality is inherently a lagging indicator. By the time thresholds are reached, preventable starvation deaths have already occurred. The classification process can also become politicized, as access to reliable mortality data is often restricted or manipulated.
"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."
Co-authors are Ingrid de Zwarte, Wageningen University; and Alex de Waal, Tufts University.
Columbia University Mailman School of Public Health
