“Mortality rates in a refugee camp in South Sudan are nearly double the threshold for an emergency, Médecins Sans Frontières (MSF) has warned” said a BBC report on 6 July 2012. “In Yida camp, [MSF data show] at least five children dying each day, most from diarrhoea and severe infections,” 2 August, MSF press release.

But where did these numbers come from? Getting good data might not seem the first priority in an emergency, but it is essential to understand and respond appropriately to the needs of the population affected. I am an MSF epidemiologist and have just returned from the Jamam refugee crisis in South Sudan. In early May 2012 we started receiving reports of refugees fleeing Blue Nile state, Sudan, and crossing the border into the world’s youngest nation, South Sudan. The refugees were escaping fighting and bombing by the Sudanese Armed Forces (SAF) targeting rebels allied with the South Sudanese (Sudanese People’s Liberation Army (SPLA)-North). These refugees, weakened by months on the run, hiding in forests and caves, and eating nothing but leaves were arriving in Upper Nile State, South Sudan, in shocking conditions, sometimes beyond medical care. And then in June 2012 the rains came, turning Jamam refugee camp into a muddy swamp.

MSF was already working in two refugee camps in Upper Nile State (Jamam and Doro) and another in neighbouring Unity State (Yida), supporting a total of 120,000 refugees since January 2012 that had fled the same fighting. Agencies were already struggling to provide enough clean water and shelter to these refugees. This was certainly a challenging environment. I have never seen children so malnourished that they are literally skin and bone, or people drinking from muddy pools. My role was to understand the scale and severity of the emergency. This is measured through the mortality rate. I needed to somehow measure what proportion of the camp population was dying every day. These data would guide the nature and size of our intervention. And ensure we gave these refugees the most appropriate help.

There are several approaches for measuring mortality such as the classic 30×30 retrospective cluster sampling survey where 30 clusters (villages, wards, districts) are randomly selected and 30 households within each cluster are interviewed. Although this technique is useful for capturing the mortality rate at a single point in time, it does not indicate whether the rate is increasing or decreasing and may not be relevant to the current situation if the recall period is large.

So we elected to set up prospective surveillance in which we visited every household within the camp weekly. As well as monitoring mortality rates, with this system we could monitor the health status of the refugees more closely. We could refer any sick people to the MSF clinic, monitor the nutritional status of children aged less than five years using mid-upper arm circumference (MUAC) measurements, and trace children defaulting from the therapeutic feeding programmes before their condition deteriorated.

Such a system relies on solid training of outreach workers, local people that are trusted by the local population and able to communicate in their language. We identified 46 such people, all from the local population, most of whom were not literate but could all count. It was at times challenging to train these outreach workers, sometimes requiring two rounds of translation (English-Arabic and Arabic-Ingassana or Magaja), and we had to re-design the surveillance forms such that they required no writing. But these teams were bright and understood quickly what they needed to do. And by forming them into teams in which there was a mix of men and women, all local languages were represented, and at least one person was literate we were able to discover quickly that the crude mortality rate was 1.8 per 10,000 per day and the under-five mortality rate was 2.8 per 10,000 per day. Both rates were well above the emergency threshold, defining this as a severe situation. Almost 3 children were dying in the camp every day and 65% of deaths were due to diarrhoea. We were able to respond to these shocking findings by decentralising clinic services, setting up oral rehydration points throughout the camp, and promoting safe water and hygiene practices at every household visited. The international media quoted our findings in headline news that demanded that governments and international agencies paid attention to this refugee crisis.

And now we have seen mortality rates dip below the emergency thresholds. There is still much to do. We need to identify and protect against the main morbidities in this population, remain vigilant against malaria and cholera for which the conditions are perfect, and moreover continue to advocate for the transfer of these refugees out of the Jamam swamp. And as long as we continue to supervise strongly and are provided with a vehicle when needed (always a major challenge in MSF projects!), the prospective surveillance system will continue to provide the real-time health data that enables us to respond quickly and provide timely care to this population who have been through so much.