This survey revealed an alarming situation; both crude and under-5 death rates, and the prevalence of SAM and GAM, were at emergency levels in the outskirts of a refugee camp which was served by several international NGOs and UN agencies [8].
The recall period coincided with an outbreak of measles in Dagahaley, which began in June and continued until October 2011 [9], and which partially explains the increasing under-5 death rate observed; 17% of all deaths were reported to have been caused by measles. As a result of the two decades-long civil war in Somalia, vaccination coverage among all ages has declined to the point that outbreaks of infectious diseases such as measles are increasingly likely. Indeed, there was a simultaneous measles outbreak among newly-displaced Somalis in Kobe refugee camp in Ethiopia and in several places in Somalia, including the capital Mogadishu [10, 11]. Outbreaks of measles among populations in crisis are common and well-documented, such that the routine vaccination of children aged 6 months to 15 years, supplemented by mass vaccination campaigns, is widely accepted as one of the most important public health interventions for averting preventable morbidity and mortality among crisis-affected populations [8, 12–14].
Figure 4 shows a timeline of various measles vaccination interventions conducted in BB. A mass measles vaccination campaign was organised at the end of April 2011, targeting children aged between 9 months and 15 years, with a follow-up campaign in July 2011 to vaccinate those children aged under 5 who did not receive measles vaccination in April. When a registration centre was opened within Dagahaley in June 2011, all children aged 9 months to 15 years were routinely vaccinated against measles upon registration. In response to the outbreak, a reactive vaccination campaign (RVC) targeting children under 5 years was launched throughout Dagahaley camp in early August, and the target age group for vaccination at the registration centre was increased to 30 years. In September 2011, a RVC was organized targeting individuals aged 15 to 30 years. It is worth noting that the RVC launched in August 2011 ran concurrently with the survey described in the article, and therefore the decision was taken not to include measles vaccination coverage in the survey, both because the results would have had no influence over any decision to launch such a campaign, and because the coverage at the end of the survey would have been different to the coverage at the start, thereby rendering the results immediately invalid. UNHCR and partner organisations assessed measles vaccination coverage in BB shortly after this campaign, and reported a coverage of 83.9% (95% CI: 73.7 – 94.0%) among children aged 9–59 months [15].
This measles outbreak was preventable; the essential lessons from past mass displacements should have been learned, and a suitable aggressive vaccination strategy implemented at an earlier stage [16–19]. We report that, if the deaths reportedly due to measles are excluded from the calculation, death rates after arrival at BB fall from 1.0 to 0.8 per 10,000 person-days (CDR), and from 2.1 to 1.8 per 10,000 person-days (U5DR). In other words, it could be argued that it was these deaths, attributed to the measles outbreak, which elevated the death rates to above the emergency threshold. However, this supposition assumes that those individuals who died from measles otherwise had zero risk of death; as severity of measles is influenced by nutritional status, we believe that this assumption is not valid and therefore that this would be an incorrect and unfair conclusion [20, 21].
In mitigation, the population most affected was that which had recently arrived, containing a large proportion of families unregistered by camp management due to the overwhelming arrival rate of these refugees. The late establishment of the registration centre and vaccination at arrival permitted the development of a pool of susceptible individuals in BB. In addition, the measles outbreak in Dagahaley was characterized by an unusual age distribution; the median age of patients recorded in the outbreak line list was 23 years, with 75% of patients aged 15 years or older, suggesting that a wider age group could have benefitted from vaccination, an observation reported earlier following a measles outbreak in a refugee camp in Tanzania [16]. However, the current ‘one-size-fits-all’ recommendations are to vaccinate all children aged 6 months to 15 years, and do not take into account the context-specific epidemiology, which in this case included a highly immunologically-naïve population due to the breakdown of health-care services arising from the ongoing political crisis in Somalia [22, 23].
Early identification of the unusual age-distribution of measles cases would have helped guide vaccination policy in this setting. Indeed, the disaggregation of deaths attributed to measles by age and by month (Figure 3) shows that age distribution of measles cases was detectable in June 2011, at an early stage of the epidemic. However, this would have required information that was not available at the time: low health facility utilisation rates and under-resourced community-based surveillance of epidemic-prone diseases meant that most measles cases occurring before July were not detected. In July 2011, by which time an outbreak had been declared and active community-based surveillance strengthened, more data were available which led MSF to advocate for a wider target age group for the RVC planned for early August, but this advocacy was unsuccessful due to the limited resources available for that particular campaign. The target age group for vaccination at registration was, however, expanded to include all individuals aged 9 months to 30 years. Owing to the failure of the August RVC to halt the epidemic, which peaked in August and September [9], adults aged 15–30 years were the target age group for the subsequent RVC.
More recent arrivals were in a significantly worse state, which was reflected both in death rates and in nutritional status. We found trends of decreasing death rates with length of stay in BB, such that those residents who had arrived more than six months prior to the survey date had death rates well below the emergency thresholds, while those who had arrived within three months of the survey date had death rates which were above the emergency thresholds. Individuals who had been resident in BB for an intermediate length of time were found to have death rates at an intermediate level.
Similarly, we found a higher prevalence of acute malnutrition and children meeting the admission criteria for entry into the nutritional programme among those children who arrived during the three months prior to the survey than among those children who arrived earlier. The same pattern was reported in a subsequent survey conducted in BB [15].
This apparent improvement in health and nutritional status over time may be due to the assistance gained after registration and the development of coping strategies, but may also be due in part to:
-
a)
a high concentration of deaths in the period immediately prior to the survey date (in particular, due to the measles outbreak), which may have resulted in artificially elevated death rates among recent arrivals due to the relatively low number of person-days contributed by these individuals (in other words, a low denominator rather than a high numerator used in the calculation of death rates); and
-
b)
higher mortality among those children with poor nutritional status on arrival: both recovery and death have the effect of decreasing the prevalence of malnutrition by removing these children from the numerator in the calculation of malnutrition prevalence. Those malnourished children who had recently arrived had had less time in which to reach either of these outcomes.
Delays in registration and food distribution were reported by many residents, and may partially account for the high mortality observed among residents of BB. We report higher death rates among individuals after having arrived in BB than before; although this suggests that conditions are worse for individuals once they have arrived, the death rates prior to arrival are subject to selection bias. Therefore, while the death rates reported for the period within the camp should be considered to be reflective of the experience of the population while in the camp, those rates reported for before arrival should not be considered generalizable to the population of Somalia during the recall period.
One limitation of this study is that we did not use a validated verbal autopsy technique when obtaining information on cause of death as this was not a principal objective, and therefore these results should be interpreted with caution. We did not use standard case definition for measles deaths (any death within one month after rash onset); it is likely that some of the deaths associated with diarrhoea, cough or breathing difficulties and fever were in fact cases of measles. Event calendars and height sticks were used to approximate events and ages, which can lead to misclassification. However, less recall bias is expected for the most important events, such as deaths of household members [5].
Another survey conducted in September 2011 reported higher levels of malnutrition and mortality in BB than we observed [15]. This may be due to the measles epidemic, which reached its peak in August and September; it is frequently reported that levels of malnutrition are increased in the weeks following an outbreak [20].