Some of the most important questions we have concern large and extreme events of which we have little experience and few examples. We don’t learn nearly as much as we could from those examples: interrogate them as experiments-at-scale, look for variation in exposure and outcomes; test hypotheses against them.
The number of people in the world who are hungry is hard to estimate and volatile. Surges in food prices as in 2008 and 2011 put more people at risk of hunger. Climate change and competition for water and land are likely to trigger future food crises.
How will this affect major diseases such as HIV? Some insight is provided by a recent article which assesses the Malawi famine of 2001-03 as a country-scale natural experiment on the effect of hunger on the dynamics of HIV.
The famine and HIV
The famine was part of a wider food crisis in southern Africa. Two bad harvests back-to-back and poor policy decisions triggered the crisis in Malawi but the stage had been set by decades of underinvestment in rural areas which had left farming households dependent on a single crop, maize, and increasingly on market purchases to get from one harvest to the next. By early 2002, a kilo of maize cost up to six times what it had a year earlier – more than an agricultural labourer’s daily wage. Urban and wealthier rural households were less affected by hunger.
The study draws on existing data: the number of rural people in need of food aid at the height of the famine from a humanitarian survey and the prevalence of HIV infection in pregnant women from the antenatal surveillance rounds just prior and subsequent to the famine.
HIV increased at village antenatal sites and in proportion to the extent of hunger in the district. That was consistent with accounts from the time of women exchanging sex for food or work and with hunger undermining immune function, pushing up HIV incidence.
The trend was opposite at sites in the towns and cities. There, HIV prevalence fell through the famine and in proportion to the extent of rural hunger in the surrounding district. Surprising, but, on reflection, consistent with the many accounts of people moving in search of food or work: rural areas had lower HIV rates so the more people who moved to a town or city, the more its prevalence would fall.
Understanding the urban declines
Supporting this interpretation, the proportion of farmers among the antenatal women increased at sites in the towns and cities as rural hunger rose whereas it fell at the village sites. In other words, hunger pushed farmers into the urban areas as it took farmers from the rural. Those who left appear to have been mostly younger women. A household survey conducted shortly after the famine confirmed that there had been a surge of rural-to-urban migration during the famine, particularly by women under 25 years.
This suggests that the decline in urban prevalence may have masked an increase in infection. The village women who moved to towns and cities entered a higher risk environment than the one they had left. Socially isolated and poorly educated, they would have had few skills or resources to avoid those risks.
Another interpretation has been suggested for the decline in urban prevalence in those first years of this century: sexual behaviour change. In surveys, men reported becoming sexually active at an older age and having fewer partners. A mathematical model that included these changes fitted the urban trends better than one that didn’t. However, the study could not explain why the decline had been greater in some of the smaller towns than in the larger cities and why rural prevalence had increased at the same time. It did not consider the possible role of migration.
HIV declined in other countries in the southern African region, mostly in urban areas, in roughly the same period. Behaviour change is the generally accepted explanation. Yet none of the studies that have examined these declines seriously consider migration by people with lower HIV prevalence as a possible cause. None of them mention the food crisis that people confronted at the time.
In many countries the conditions exist that, as in Malawi, make rural out-migration a plausible contributor to urban HIV decline: prevalence is typically lower and population greater in rural than in urban areas. Migration rates are volatile, responding to changing economic conditions and livelihood opportunities.
Learning from extreme events
The implications for policy in Malawi and beyond are different if migration, accelerated by distress, was a significant cause of declining prevalence in towns and cities.
Actions that bolster, rather than undermine, food and livelihood security would enable people to avoid situations of HIV risk and escape infection where they live or when they move. Such actions would complement effective behaviour change initiatives. That wasn’t the case during the famine. Women who knew a good deal about HIV couldn’t act on it because they had to feed siblings or children. A phrase I heard which translates roughly as “screwing to die” captured their dilemma.
Natural experiments are an under-valued resource in many disciplines. In epidemiology, a number have, as in Malawi, provided insights into the effects of restricting access to food to entire populations which could not have been gained by any other means. One, the Dutch Hunger Winter of 1944-45, played out not far from where I now write: it continues to shed light on the developmental consequences, reaching into middle age, for those who were in gestation at the time and whose mothers endured different degrees of food deprivation. Such natural experiments are a people’s common property.
Michael Loevinsohn is an independent ecologist and epidemiologist based in Wageningen, Netherlands. He can be contacted at email@example.com.