Progress in global measles control has resulted in much higher population immunity in most parts of the world. Consequently, there has been a 78% reduction in measles mortality, from an estimated 733,000 deaths in 2000 to 164,000 deaths in 2008 [4]. Although outbreaks of measles are far less likely in many regions, interruption of measles virus transmission requires a high level of population immunity (> 90%) and measles outbreaks continue to occur in populations where such high levels of immunity cannot be maintained. Humanitarian emergencies often occur in populations with low levels of immunity, given long-term disruption of routine vaccination programs, poor infrastructure and access to health services, and therefore an increased risk of measles epidemics with consequent mortality.

Although preventive mass measles vaccination in emergency settings has not been the subject of controversy, and in fact is a part of standard international guidance to prevent outbreaks from occurring, to date there has not been a review of these interventions. Present guidance for humanitarian emergencies is largely based on a model of humanitarian relief, which is focused on camps sheltering refugees or internally displaced persons. These camps were often overcrowded, with high risk of epidemic-prone diseases such as measles, particularly during the acute phase of the crisis. Preventive mass vaccination of a targeted age group aims to reduce the risk of epidemics. However, the nature of humanitarian emergencies has changed over the past decades with increasing numbers of displaced persons and refugees now residing in urban environments and dispersed among host communities rather than just in camps [5]. Further, the coexistence of crises of differing nature and intensity in the same region renders defining the beginning and end of humanitarian crises difficult, if not irrelevant.

Measles epidemic risk may be more closely related to the characteristics of the affected population prior to the emergency, than to the precipitating event. It is thus important to consider that countries are in different stages of measles control. The Americas have seen the elimination of indigenous measles since 2002 while several other WHO regions (EURO, EMRO, WPRO and AFRO) have declared elimination goals, and SEARO region currently has a mortality reduction goal.

Our goal was to revisit the WHO-UNICEF Joint Statement on Reducing Measles Mortality in Emergencies [6] and the Sphere Project Humanitarian Charter and Minimum Standards in Disaster Response [7]. Taking into consideration the changing epidemiological landscape of measles and progress in measles control, as well as the changing nature of humanitarian emergencies, may suggest that current guidance may need to be updated. During the acute phase of an emergency, current guidance recommends a swift preventive mass vaccination campaign, along with vitamin A supplementation including all children from 6 months through 14 years of age. The WHO/UNICEF statement adds a contingency that at a minimum, children from 6 months through 4 years of age must be immunized.

Any article that mentioned: i) a measles outbreak; ii) described vaccination coverage either before and/or after an outbreak; iii) a vaccination intervention (whether or not it was implemented); and iv) occurred in a humanitarian emergency defined here as a country that submitted and received a Consolidated Appeals Process (CAP) or Flash appeal to the UN Central Emergency Revolving Fund (CERF) between January 1, 1998 and January 1, 2010 were considered "relevant". A vaccination intervention was considered to be any vaccination intervention with measles-containing vaccine beyond routine services that are normally available at healthcare facilities.

For each of these reports, some covering an outbreak in the same country, we attempted to determine objectively the impact of the measles vaccination intervention as it pertains to age range based on the data provided. Table ​Table11 describes the epidemiologic characteristics of the reviewed reports classified by region to provide context on measles control. Table ​Table22 includes details on the mass vaccination intervention noting in particular the time to the response (where reported) and if there was evidence of an impact.

In the Americas [9-12], there were no reports of preventive mass vaccination campaigns during the acute phase of a humanitarian emergency, but several reports of outbreak response immunization (ORI). An outbreak in Bolivia beginning in 1998 affected the country nationwide. A nationwide non-selective vaccination campaign, where children irrespective of their vaccination status are eligible for vaccination, was implemented four months after the first case was reported targeting children 6 m to 5 years with reported 85% coverage obtained in this age group. The following year house-to-house campaigns were performed in two departments of the country and in high-risk municipalities. In 2002, a house-to-house campaign was performed nationwide targeting children 6 m to 4 years with a reported 95% coverage and halt in transmission [9].

Similarly, in Haiti, cases were reported in Gonaives beginning on March 8, 2000. A non-selective mass vaccination campaign (single visit, house-to house) targeting children 6 m to 14 years was implemented at the end of April, 2000 with reported 95% coverage. The last case in Gonaives was reported on May 3, 2000. Subsequent campaigns were repeated in Artibonite, Port-au-Prince and Delmas after cases were reported there [10]. In Colombia, an epidemic in 2002 affected approximately one third of the country and a vaccination response was implemented door-to-door targeting children 6 months to 5 years in high risk areas. The authors posit that the prompt, although specific details of the delay are not given, door to door vaccination and surveillance may have prevented an even larger outbreak in a Colombia where routine services were limited by long-term conflict [11,12].

Reports from Asia include two non-selective mass vaccination interventions in response to natural disasters in India [13-20]. One response entailed the preventive vaccination of children in flooded areas of Bihar, where high population density and subsequent poor access to care placed the population at high risk. Non-selective vaccination of children 6 months to 14 years achieved an estimated 75% coverage. A total of 1811 measles cases were reported but there is insufficient data presented to determine the potential impact of this intervention, although the authors' qualitative analysis suggest that the campaign prevented a larger scale outbreak [13].

The second report from India describes the emergency response to the Indian Ocean earthquake and tsunami of 2004. Non-selective preventive mass vaccination for children 6 to 60 months was conducted in 58 villages of Tamil Nadu province, where one-dose measles coverage was reported to exceed 95%, beginning December 29, 2004, four days after the tsunami. A cluster of measles cases was subsequently reported in a tsunami affected area on December 30 with cases reported in non-tsunami affected areas of the province soon after. Although the overall scale of the outbreak was small (n = 101), the authors conclude that the target age range of the preventive vaccination was too restrictive as more than half of measles occurred in children between 5 and 15 years cases in both tsunami-affected villages (56.3%) and non-tsunami affected villages (60%) [14].

Two additional reports describe interventions in refugee populations [15-17]. In Afghanistan, following the fall of the Taliban, an influx of approximately 2 million refugees returning from Pakistan and other neighboring countries was anticipated in early 2002. In response, non-selective vaccination of children 6 months to 12 years was conducted throughout 2002 reaching 82%-96% of the target population by the end of 2002 [18]. The campaign initially targeted high-risk districts and cities with the largest number of susceptible children, and subsequently the most remote and inaccessible villages. A follow-up campaign was conducted in 2003, targeting children aged between 9 and 59 months. It is important to note that this campaign was prompted by the fact that an epidemic had occurred in 2001 affecting at least 7 of the 30 provinces in Afghanistan. Difficult access due to snow and mined roads and insecurity left many districts without heath services. The actual scale and scope of the 2001 epidemic is difficult to estimate, but a total of 8,762 cases were reported through the nationwide surveillance system, of which 33% of cases (n = 8762) occurred in children 5 to 12 years.

In Sri Lanka, a measles epidemic with a suspected 15,250 cases between October 1999 and June 2000 was reported [19]. The outbreak began in Colombo and progressed to becoming countrywide. Response included actively searching for and vaccinating children under the age of 10 years at the local level who did not report previous vaccination. Non-selective vaccination in "welfare centers, refugee camps, preschools, and urban slums" was also conducted without specifying the age range or whether all locations were included. The authors report that they "specifically chose not to implement outbreak response immunization as the WHO recommends such activity only under specific conditions such as refugee camps, military barracks or closed communities." The authors provide insufficient information with which to assess the potential impact of the intervention, but it is important to note that of the 3728 measles cases with sufficient detail, 40% reported having been vaccinated previously and 69.4% occurred in children over 10 years.

In the African region [21-31], two reports describe vaccination interventions in response to the nationwide epidemic in Niger in 2003-2004, where 50,138 cases were reported. A reactive campaign in the capital Niamey (n = 10,080 cases), targeted children 6 months to 5 years, 5 months after cases were reported [21]. In Mirrah District, Niger, outbreak response vaccination was restricted to outreach vaccination services in some health centers, although the extent of these efforts was not well documented [22]. The results of a retrospective household survey found two-thirds of case patients were under age 5 and 90% under the age of 10. The author's remark on the need to include children older than 5 years in vaccination activities as this may prevent deaths in infants who acquired measles from older children and also prevent deaths in older age groups, the rationale for the SPHERE recommendations. Mortality was inversely associated with the age of case patients, with the highest CFR in children under 12 months (15.7%; n = 13/83); followed by children 12-59 months (11.5%, n = 64/558); then children aged 5-14 years (5.4% (n = 14/259). In the same region, epidemics in Nigeria and Chad also occurred [23]. There was no vaccination response to the epidemic in Nigeria. A non-selective campaign targeting children 6 months to 5 years, four months after cases were reported, was implemented in N'djamena, Chad in 2005. Although subsequent SIAs in Niger, Nigeria and Chad reported obtaining high coverage among the target population, outbreaks continue to be reported in this region.

In Tanzania, a report on an outbreak among Burundian refugees in four camps noted 31% of cases were between 6 and 15 years [24]. A non-selective response targeting children 6 months to 5 years, initiated between one and five months after cases were first reported in each of the four camps reduced cases and deaths, however, the authors conclude that it would have been more effective to target a wider age range to halt transmission. A report on a measles epidemic in Gode, Ethiopia came to similar conclusion recommending that a wider age range than the 9 months to 5 years targeted in the response, which although prompt, could have contained the outbreak [25]. The authors further note the poor coverage achieved by the intervention and potentially poor vaccine efficacy due to presumed problems in the cold chain.

In Darfur, Sudan, although cases were reported throughout the Darfur region, non-selective vaccination targeting children 9 months to 15 years was conducted only in North Darfur, reaching a reported 93% of the accessible population, but an estimated 77% of the total target population [26,27]. Measles cases continued to occur after the intervention. The authors report difficulties accessing a population that was continually moving to avoid violence with the repercussion that new returnees to the camps were not vaccinated.

Of the few well-documented epidemic descriptions in humanitarian emergencies, the age range of cases is not limited to under 5 year olds. Combining all data, both from preventive and outbreak response interventions, about 59% of cases in reports with sufficient data reviewed here remain in children under 5, 18% in 5-15 and 2% above 15 years (Figure ​(Figure1).1). In instances where interventions targeted a reduced age range, several reports concluded that the age range should have been extended to 15 years, given that a significant proportion of cases occurred beyond 5 years of age. Non-selective mass vaccination of children 6 months to 15 years remains the most prudent option for reducing measles morbidity and mortality in emergencies. In some cases, vaccination of age groups greater than 15 years may need to be considered based on a risk assessment of the area including whether the country has a mortality reduction or elimination goal. Recent epidemics in Burkina Faso and Malawi, although not in the context of a humanitarian emergency, reported more than one third of cases over the age of 15 years. In humanitarian emergencies, particularly in protracted crises, routine services may be compromised for many years and thus older age groups may not have been routinely vaccinated. Older age groups continue to be left out as the routine program targets children under 5, again highlighting the importance of mass campaigns and increasing the target age group for mass campaigns to 15 years.

However, we recognize that in some cases, target age groups may need to be reduced due to lack of medically trained staff, extreme security situations or limited vaccine supplies. The current ongoing epidemic in the Democratic Republic of Congo, spanning a large geographic area and population presents a serious challenge in terms of a rapid response and in this case if it is only possible to vaccinate a portion of children at risk, children 6-59 months should be prioritized. We recognize that extenuating circumstances may necessitate allocation of scare resources and less optimal strategies put in place.

While mass vaccination for measles in humanitarian emergencies remains necessary, the best and most cost-effective approach is to prevent epidemics entirely by ensuring high first dose routine vaccination coverage and a second opportunity for measles vaccination for all children. Humanitarian emergencies are overlaid onto contexts with differing level of pre-existing population immunity, which influence the risk of an outbreak. Countries in the Americas, where measles has been eliminated due to a high quality and sustained effort, have smaller scale epidemics occurring in a setting where routine vaccination remains the core of the control effort. Conversely, larger scale epidemics in countries like Afghanistan, where routine services have been interrupted for more than 20 years and insecurity curtails both preventive and reactive vaccination, have continued for long periods. There is a critical need to consider the epidemiology of measles within the context of the crisis in order to provide an adapted response.

This review has important limitations. First, we restricted our review to the published literature. Although we did search the grey literature through the collection of international agency and NGO documentation, conclusions from the grey literature are severely limited. Reports and databases focus often on delivery and rather than on an epidemiologic analysis of the intervention. This is due largely to the fact that formal documentation of emergency response is not a part of the standard operating procedure of many emergency organizations. It is not a routine part of the professional culture, and when reports exist, they relate to a single organization's response and are often for internal use or limited distribution. By relying only on the published literature this review suffers from a clear publication bias. Reports relating to responses in humanitarian emergencies are rare as the necessary and important aspects of publication are not often met (ethical approval, study protocol, logistic constraints and poor awareness of the publication process) and documentation of events may be low on the list of priorities in often overwhelming situations where the primary goal is to deliver and provide aid to a population in order to reduce mortality and morbidity. Nevertheless, we chose to review the published literature, as however scanty, it still remains the reference for evidence-based guidance. An additional limitation to focusing on the published literature that there are scarce reports of reactive or preventive vaccination campaigns in emergency settings where no measles outbreak occurred. This is an important part of evaluating the impact of current recommendations; however, such situations are even less likely to be published.

A third and related limitation is the choice of our definition for emergencies. We chose to use the definition of countries applying for a CAP or Flash appeal during the period of our review. We also included countries that had ever applied for assistance during the time period, whether this coincided with the reported epidemic or not. As a result, humanitarian emergencies were not included if they occurred in a country that did not apply for CAP or Flash appeals.

As the landscape of emergencies changes, epidemics in countries not undergoing armed conflict or natural disaster, but rather political instability, dire poverty and displacement from trans-boundary or regional conflicts become increasingly frequent sites for emergency interventions. Displaced persons, whether escaping violence or seeking employment and assistance, increasingly seek refuge in cities, as reflected by the number of large urban outbreaks included in this review. Alternately, rural and remote areas with dispersed populations may become a more frequent site for intervention as care provision and maintaining sufficient vaccination coverage in these areas is difficult. Responding to the risk of a measles outbreak in rural areas bears closer similarities to an emergency response than in a stable setting. Measles outbreak responses in humanitarian emergencies are predominantly campaign-based, the population denominator is often unknown or unreliable and the response is often done in coordination or partnership with UN agencies and disaster relief agencies. This is contrasted with a stable setting where the response may be undertaken predominantly through fixed sites and the national infrastructure.