Rapid urban malaria appraisal (RUMA) in sub-Saharan Africa

Urbanization has a significant impact on the economy, lifestyles, ecosystems and disease patterns, including malaria [1, 2]. An estimated 39% of the population in sub-Saharan Africa (SSA) lived in urban areas in 2003 [3], 198 million Africans lived in urban malaria-endemic areas and 24–103 million clinical attacks occur annually in those areas [4]. An important message addressed in the Pretoria Statement on urban malaria was that the malaria control strategies used in rural areas cannot be directly transferred to the urban context [5]. The epidemiology of urban malaria poses a number of specific challenges: i) the first malaria infection occurs often late in childhood and the acquisition of semi-immunity is delayed [6]; ii) the intensity of the malaria risk is often heterogeneous over small distances, being subjected to the degree of urbanization of particular subdivisions [7, 8] and their proximity to possible vector breeding sites [9, 10]; iii) rural-urban migration is likely to increase the endemicity of malaria [11]; iv) agricultural and animal husbandry are important economic activities which create a favourable environment for Anopheles breeding [12, 13]; v) marginalized populations usually lack access to health care, which hampers the effectiveness of case management and the promotion of intermittent antimalarials during pregnancy [5, 14‐16]. There is now substantial private sector activity in health care provision in many cities. The private services providers are often untrained or unlicensed, but are seen as a source of inexpensive care by patients. There is not much information about the impact of the private sector on case management.

Around 235 papers related to malaria epidemiology in SSA urban settings were published from 1945 to 2004. Entomological profiles and clinical patterns are known to vary between urban, suburban and rural environments [17]. A review of other studies in SSA urban centres showed that transmission patterns vary greatly by city, season and age group. The overall prevalence of parasitaemia was 4.0% in schoolchildren in Brazzaville [18], 2.4–10.3% in Lusaka [19], 2.0% in a Gambian urban area [20] and 3.6–7.5% in Dakar [21]. It was also reported that malaria prevalence in school children varied from 3.0% to 26.4% in different areas of Ouagadougou [22] and varied from 14% in a central urban area to 65% in peri-urban areas in Kinshasa [23].

Evidence showed that the rate of clinical malaria attacks detected in urban health facilities was high and season-dependent. For example, Hendrickse et al. found that 36.8% of outpatients were parasitaemic in a hospital in Ibadan [24]. In Niamey, the parasite prevalence was 61.9% during the rainy season but only 5.4% in the dry season in 1989 [25]. In Kinshasa, malaria admissions comprised 29.5% of consultations in 1983, then 38.2% in 1985–86 [26]. In Dakar, malaria fever represented 19.7% of consultations and 34.3% of fever cases were caused by malaria in 1988 [27]; the same authors found that 5.3% (dry season) and 58.8% (rainy season) of febrile outpatients were parasitaemic in 1994 [28]. In Ouagadougou, malaria prevalence accounted for 33% of all outpatients [29], while Dabire reported 22% malaria parasitaemia among children aged 0–14 years in the paediatric ward [30].

Transmission and severity of malaria are influenced by the geographic characteristics of a town and by the socio-economic environment. The heterogeneity and seasonal variation of the entomological inoculation rate, depending on both vector densities and sporozoite rates, have been documented [31, 32]. Lindsay et al. (1990) showed a difference in the composition of vector species and the vector's adaptation in different subdivisions Banjul [20]. To improve interventions, the determinants of the diversity of transmission levels within subdivisions of a city should be understood. Concerns were raised about the association between urban agricultural activities or local irrigation systems and the creation of breeding sites for Anopheles sp. [12, 33, 34]. Peri-urban areas often lack infrastructure, including poor water supply and sanitation, which provides an ideal environment for vector breeding [35]. For example, urban Dakar has >5,000 market-garden wells which provide permanent sites for mosquito larvae [13]. An identification of vector species, regular larval inspection and larviciding activities should be implemented in the framework of urban malaria control programmes [36].

This article presents the experience of developing a rapid urban malaria appraisal (RUMA) in SSA, carried out with the support of the Roll Back Malaria Partnership. The aims were i) to develop a rapid assessment package that is explicitly evidence-based and can be carried out within a six to ten weeks timeframe; and ii) to assess how rapid malaria appraisal efforts could be best integrated into the municipal health department supervision and to inform control programmes.

One of the principal aims of the present work was to review the feasibility, perceived usefulness and consistency of the collected information. Because RUMA was a cross-sectional assessment the external validity of the findings could not be assessed. However, the internal consistency of the results was assessed.

Below, the strengths and weaknesses of each methodology are presented, bearing in mind the constraints imposed by a rapid assessment. Detailed results for each site will be provided in a series of forthcoming publications.

This assessment was accomplished in four countries within a period of six to ten weeks in the field and has proven to be a helpful tool in supporting planning of urban malaria control. An ongoing urban malaria control intervention in Dar es Salaam has been initiated on the knowledge basis provided by RUMA. With the incentive of extra-budgetary resources and technical support from STI, local partners were committed to incorporate RUMA into existing activities at the municipal level. Qualified personnel and opportunities for integration, synergy and co-ordination were identified during the meetings with local partners and the collaborations were always very successful.

The RUMA methodology is a cross-sectional design and the results are likely to change over time due to seasonality, the dynamics of urbanization and the evolution of malaria transmission. In Dar es Salaam, for example, the surveys were carried out during an exceptionally dry period and results could underestimate the true transmission intensity. Many factors such as the size of the city, the fieldwork logistics, the availability of local expertise and the coordination with local senior officers can influence the schedule and planning, as well as the outcome of such surveys.

The study highlighted the need for improved Health Management Information Systems (HMIS) in SSA urban areas. Municipal health departments routinely collect health facility data but information is rarely fed back to the districts and facilities that generate the information. The data are often not available for analysis or accessible due to false registration and under-reporting from health facilities, as well as poor filing and storage of documents at the district or municipal level. In addition, the low number of true malaria cases among fever episodes treated as "malaria" raises the issue of the validity of the collected data even further. Hence, much progress needs to be made in order to estimate more accurately the urban malaria burden and plan relevant control measures.

GIS provides a platform to display health services and geographic features in relation to population settlements. In this experience policy-makers could readily use the presented information for improved planning, re-allocation of resources and for strengthening the networking between the public and private sectors. While the GIS technology has been shown to be very useful in studying health care delivery and distribution of diseases, its application in an entomological assessment was quite difficult and costly and could only be done in conjunction with other ongoing projects. Hence it should be excluded from the process of RUMA. In contrast, the mapping of health facilities with GIS was feasible and cost-effective.

While results from the school surveys gave an indication of the endemicity range and risks in the targeted community, they cannot be considered as being representative without a wider survey. The variations of malaria risk were sometimes related to political divisions or man-made boundaries, but often were due to divergent socio-environmental factors and the degree of urbanization. Because site-specific environmental conditions lead to an aggregated distribution of vectors and different malaria risks, the sampling sites were selected taking into account the population density, the natural environment and urbanization patterns. This should improve the rough categories that previous researchers applied (centre, intermediate and periphery).

Despite a potential attendance bias, the health facility surveys allowed the determination of prevalence of parasitaemia among presenting clinical cases, and the calculation of the fraction of malaria-attributable fevers. This allowed to document clearly the high rate of malaria mis-diagnosis in the health facilities. This information is of great importance for urban malaria control.

Overall, RUMA is a first step towards understanding malaria endemicity and designing control strategies. It has exemplified the concern for mis-diagnosis of clinical malaria in SSA cities [25, 50]. A report by the Tanzania-Japan malaria control programme in Dar es Salaam mentioned that drug administration to diagnosed children was one of the essential interventions that reduced the malaria rates between 1988 and 1996 [36]. An in-depth research is now being implemented in Dar es Salaam to assess the malaria burden with a much larger sample size. The application of RUMA methodology is possible and desirable in other SSA urban areas and it should have a special focus on improved diagnosis.