The 2010–2011 measles outbreak in Katanga was one of the largest occurring during the last decade. The overall AR was higher than that reported during the previous outbreaks in Matadi or Mbuji-Mayi (DRC) in 2006 , comparable to that reported during the large outbreaks in N’Djamena (Chad) in 2005 and 2010 , but lower than those reported in Malawi in 2010  and in the city of Niamey (Niger) in 2005 .
The AR was higher in the southern health zones of Katanga, which are densely populated and well-connected, thus facilitating disease transmission. However, the variable performance of the surveillance system should also be considered, as greater efforts were made to reinforce surveillance in southern than in northern health zones, so that the number of measles cases reported from the northern part of the province is likely underestimated.
AR was higher in children <4 years (>5%) compared with older age groups (1% in 5–9 years-old and 0.2% in 10–14 years-old). This distribution of cases across age groups observed in Katanga was in between those observed in Matadi or Mbuji-Mayi and in Malawi. In Matadi and Mbuji-Mayi , ARs were respectively 2.1% and 1.4% among < 5 years and 0.4% and 0.02% among children 5 to 14 years. In Malawi, AR was 2.3%, 1.1% and 0.5% among respectively <5 years, 5 to 14 years and 15 years and over .
The age distribution of cases was consistent across health zones and throughout the outbreak. It reflected local measles epidemiology and past control strategies. Indeed, the AR was higher among children under 4 years of age, born after the last SIA and the last measles outbreak in 2007. Children aged 4 to 8 years, eligible for the 2007 SIA and exposed to the 2007 outbreak in the north of the province, and those aged 9 years and older, eligible for the 2004 SIA (not assessed during vaccination coverage surveys) and exposed to the outbreaks in 2004 and 2007, were less affected during the 2010–11 epidemics (AR < 0.5%).
Early assessment of age-specific AR should guide the choice of age groups to be targeted through vaccination campaigns. During this outbreak, the age distribution of early cases reported in Malemba-Nkulu health zone was used to restrict the targeted age group to children aged under 10 years. Early findings from surveillance in the context of limited resources might provide important insight to aid the planning of the vaccination response by defining the priority age groups to target.
Case fatality observed was lower than reported elsewhere . Deaths recorded through the surveillance system in DRC are likely underestimated as only deaths occurring in health facilities were captured. Community-based mortality surveillance may have provided a better description of measles-related mortality.
Many reasons can explain the 2010–11 epidemics in Katanga. EPI coverage among children 9–11 months was under the 90% recommended by the WHO , with some health zones under 80%. Routine coverage was also low among children aged 12 to 23 months, highlighting the need to enlarge the age range for providing the first dose of measles vaccine through EPI . Moreover, the 2007 SIA coverage was below the 95% level recommended by WHO , and the SIA planned in 2010  were not conducted.
Several limitations should be taken into account. First, only cases admitted to health facilities were registered in the surveillance system. As access to care is low in Katanga, especially in rural areas, the total number of measles cases may be underestimated. This was confirmed by the results of vaccine coverage surveys. Indeed, we asked if the children had symptoms since the beginning of the outbreak and could therefore estimate ARs. Estimated ARs were higher than those calculated through reinforced surveillance (estimated AR in Lubumbashi and Likasi cities were 5.6% [IC 95%: 4.9-6.4] and 6.3% [IC 95%: 4.7-8.5] respectively among the targeted children for example). Moreover, varying ARs in the different districts might partially reflect varying performance of the surveillance system as surveillance was reinforced in only 28 out of 67 health zones.
Second, we conducted vaccination coverage surveys in 19 health zones; our results cannot be extrapolated to other health zones, where coverage may be very different particularly in rural areas. Moreover, the percentage of vaccinated children ascertained by card was low, leading to potential misclassification, although previous studies have shown parental recall to be highly reliable . To further minimize bias, we recalled the site of injection to the participants (measles vaccine is given in the left thigh in DRC) and the period the vaccination might have taken place. This highlights the difficulty in obtaining precise estimates of vaccination coverage in settings with weak public health infrastructure as both population figures and individual data on vaccination are not available. Future work should include serologic confirmation.
Third, the last census was carried out in 1984 in DRC. Due to uncertainty in population figures, age-specific ARs may be either under or over-estimated.