We evaluated the cholera surveillance system of the region of Maradi and found the system sufficient to appropriately identify endemic and high risk areas for potential cholera vaccination.
Overall, the evaluation of the surveillance system suggests that the system is reasonably sensitive and that collected data are representative of the distribution of cholera cases in the region. Initial cholera cases were missed, but most were subsequently identified and recorded. Furthermore, the systematic realization of stool cultures in suspected cholera clusters and the high transmissibility of the disease made it unlikely that an epidemic of cholera would evade the surveillance system.
Few missing values were found. This may be due to the regional centralization of the surveillance system during cholera epidemics where data are checked daily at the regional level. This increased communication and feedback loop between local and regional levels helps to improve data quality. Interviews with surveillance system officers also asserted that seeking health care in case of diarrhea during a cholera epidemic was common. Thus, for people older than 5 years of age, and especially in areas with health care facilities, non-reported severe cholera cases should have been rare. The data issued from Maradi’s surveillance system could be improved, but the quality sufficed to identify high-risk areas with confidence.
Several limitations should be kept in mind when interpreting the results of the surveillance system assessment. First, the evaluation was retrospective and limited to a four year observation period during which only two had a large number of reported cases. Second, due to logistical constraints, information about cholera surveillance outside Maradi city was filled by regional heads only, while within the city all the heads of health facilities were also interviewed. In all cases, however, it was possible to contrast findings from interviews with surveillance system reports. Third, as there was no alternative surveillance system for comparison, the evaluation was based solely on a retrospective review of MoH records, reports and personal interviews. Consequently sensitivity was assessed qualitatively. Fourth, initial clinically suspected cases were laboratory confirmed, but as testing ceased after the epidemic was confirmed, a specific study would be necessary to fully evaluate the system’s PPV. Lastly, the surveillance system was designed to capture cases 5 years of age and older. While this follows WHO surveillance recommendations, it does necessarily result in an underestimation of cases in this age-group.
As an estimated 40% of the population resides more than 5 km from a health facility [14], under-reporting from these more distal areas could lead to inaccurate epidemiological estimations. In fact, sensitivity at the onset of epidemics appears to have been lower in rural areas, especially in remote villages and health centres. Due to cholera’s severity and high transmissibility, however, and supported by the results of our evaluation, it appears that most originally missed cases were retrospectively recorded by health authorities. Conversely, an overestimation of cholera cases may have occurred, particularly in urban areas due to over-attribution of non-cholera diarrheas. The clustering of cholera cases in Maradi, in addition to most neighborhoods reporting no cases suggests that over-reporting did not play a major role either.
An additional limitation applies to the calculation of incidence rates. Population data from the 2001 census would likely underestimate the population figures during the period 2006–2009, resulting in an overestimation of incidence rates in the study period. Cholera caseload could, however, as noted in Bangladesh be several-fold higher than the figure presenting at hospitals [15]. Furthermore, the system does not routinely report cases under 5 years of age, which in some Asian and African settings present with higher attack rates and are more prone to be hospitalized during cholera epidemics [16, 17]. Consequently it is more likely that cholera incidence rates are higher than presented here, despite the population underestimation.
Our evaluation suggests that the cholera surveillance system in place during the reviewed period is reasonably sensitive and could be used for detection of areas at risk for cholera epidemics. WHO criteria to implement a vaccination campaign [8] were met in the neighborhoods of Bagalam and Yandaka in the city of Maradi. The city of Maradi is the major transport trade and agricultural hub of the region. These neighborhoods are also among the poorest of the city, over-crowded and are susceptible to flooding. These factors may explain why the city took a major part in the spread of cholera epidemics, particularly in the overcrowded, poor sanitation neighborhoods of Bagalam and Yandaka. Following this evaluation and identification, these two neighborhoods are being considered for a cholera vaccination campaign. Of the two currently available prequalified oral cholera vaccines, Shanchol does not require a buffer or water for administration, is less expensive and potentially promising for use in contexts with limited sanitation infrastructures [18]. In overcrowded, poor sanitation neighborhoods in the city of Beira, Mozambique, Jeuland et al estimated that a vaccination campaign targeting the whole 1–14 year-old population would be very cost-effective [19], despite the logistical difficulties to its implementation. The implementation of a similar long-term oral cholera vaccination campaign in the identified neighborhoods of Bagalam and Yandaka may present similar results, until associated water and sanitation infrastructure are improved.