Incidence of malaria among children living near dams in northern Ethiopia: community based incidence survey

Introduction

In 1974 and 1984 severe famines occurred in Ethiopia, attributable to drought, erratic rainfall, and dependence on traditional rainfed agriculture coupled with many years of war. Tigray region in northern Ethiopia with a population of 3.2 million, of whom 85% live in rural areas, was one of the worst affected.

To deal with these problems the regional government of Tigray initiated a postwar rural development programme called “Sustainable Agricultural and Environmental Rehabilitation in Tigray” (SAERT), the main objective of which was to minimise dependence on rainfed agriculture and improve food production by constructing microdams and introducing irrigation systems.

Alongside potential advantages of increased agricultural productivity, development of water resources may have both positive and negative impacts on health.1^–3 Diseases that rely on the presence of water for transmission—such as malaria and schistosomiasis—may increase in incidence. As a result of such large scale development in Sri Lanka, malaria became prevalent in areas where it was once declared to have been nearly eradicated.4 In the Rusizi valley, Burundi, parasite prevalence rates were substantially higher than outside the irrigation scheme,5 and in Sudan transmission changed from a seasonal to a perennial pattern in irrigated areas.6 Cross sectional pilot studies in Tigray suggested that malaria was more prevalent in villages close to existing microdams compared with villages where dams were not yet built.7 8

In Tigray region, malaria follows a highland, hypoendemic pattern with seasonal transmission, mainly from September to November, after rain from June to August. Of the entire population, 60% live in malarious areas, and malaria is a major public health problem, being a significant cause of morbidity and mortality. As the overall incidence of malaria is modest in this highland area, acquired immunity also tends to be low and so epidemics can be serious. The major vector is Anopheles arabiensis, and around 90% of infections in the high transmission season are Plasmodium falciparum. Extrapolations from studies of malaria and irrigation in other environments are therefor tenuous, particularly for variations specific to altitude and season. Against this background, possible effects of the current water resources development programme on malaria in Tigray have been investigated in a 1 year longitudinal study, which examined whether dams increase the risk of malaria in established hypoendemic situations, extend established seasonal patterns of malaria transmission, and facilitate transmission of malaria at altitudes otherwise free from the disease.

Methods

Eight microdams in the Tigray region (in rural areas, with established permanent water) were identified, at altitudes of 1800 m to 2225 m. By using established data on the flight patterns of anopheline mosquitoes 9 10 rural communities within 3 km of the dams were identified as potentially at risk, and villages at a similar altitude (within 200 m) and 8 to 10 km from the dams were randomly identified as controls. Villages within 8 km of other major water bodies (impounded water persisting perennially) or which had been spray treated were excluded from further consideration. To monitor the incidence of malaria in the community longitudinally four surveys were planned in a 1 year period among children under 10 years of age. The study was designed to detect a 1.33 relative rate of incidence of P falciparum malaria (20% in the at risk group and 15% in the control group) with a drop out rate of 20% and a design factor of 3 (to allow for the dam effect operating at village rather than individual level) with 80% power and 95% confidence, leading to a total requirement of almost 7000 children. House to house census surveys were undertaken in identified at risk and control communities, giving rise to slightly over 7000 children under 10 years of age (age determined by interview with parents or guardians); no further sampling was therefore required. The census data were entered on a dBase computerised system that enabled tracking of individuals and vital events throughout the survey.11 The mean altitude of the at risk communities was 2013 m and of the control communities 1991 m, while the dams themselves were situated at similar altitudes to the corresponding communities (mean 1997 m).

Field work was undertaken by enumerators specifically recruited and trained, both at headquarters and on site. Each quarterly survey was undertaken in two phases. Firstly, blood films were sought from all children under 10 years, and those with clinical malaria (defined as 1 or more days of fever, chills, sweating, headache) were treated with chloroquine (25 mg/kg over 3 days) and another blood film taken after 14 days. Secondly, further blood films were sought from all children as close to 30 days as possible after their initial film Overall compliance with blood taking was 93% in both groups. During the intervening period any child falling sick was encouraged to consult the project enumerator in their village for passive case detection, at which time a further blood film was taken and chloroquine given if appropriate. Surveys were conducted on this basis simultaneously in all villages four times during 1997: in January-February, April-May, July-August, and October-November. All blood films were initially read in local malaria control laboratories. Films positive for parasites and a 10% sample of films negative for parasites were subsequently checked in the regional office; when discrepancies remained, a third independent reading was made and the majority opinion taken.

The computer system at the regional headquarters was used to facilitate longitudinal monitoring of the surveyed population and to process the results of blood films.11 Days at risk were calculated in untreated children from date of first film to date of passive case detection (if treated or positive) or date of second film, and days at risk for treated children were similarly calculated from time of negative follow up film to passive case detection or second film Children whose first film yielded positive results for P falciparum were excluded even if results were negative on follow up (14 days) as late recrudescence could not be ruled out.12 As children with P vivax infections were not treated with primaquine, relapsing vivax infections could not be ruled out; thus second and subsequent P vivax infections in individual children were not taken as incident cases Incidence rates were calculated from numbers of incident cases and child months at risk, the latter being calculated on a daily individual basis from the database system.

Incidence of malaria (with 95% confidence intervals) among children under 10 years in highland Ethiopia in at risk communities (close to dams) and control communities (away from dams) during 1997 and rainfall, which governs the seasonal pattern of malaria in this environment

• Download figure
• Open in new tab
• Download powerpoint

Statistical analyses of incidence are based on rates of cases per child month at risk, with 95% confidence intervals of incidence rates and rate ratios when appropriate.13

Ethical approval for this study was given by the Ethiopian Science and Technology Commission and the Tigray Regional Health Bureau and local consent obtained through community meetings.

Results

Over the 1 year period of surveillance 160 incident cases were detected in at risk communities during 11 392 child months at risk, giving a rate of 14.0 episodes/1000. For control villages, 23 cases in 11 938 child months gave 1.9 episodes/1000, giving a 7.3-fold increased incidence close to dams. This is a highly significant difference, with a 95% confidence interval of 4.7 to 11.3. Of the total 183 cases, 107 were diagnosed as P falciparum, 73 as P vivax, and three as mixed. The proportion of P vivax infections was not significantly different between at risk and control groups. Six children experienced second P vivax episodes, which were excluded.

The seasonal variations in incidence are shown in the figure, together with rainfall patterns. The overall results are presented stratified by altitude and season in the table. For the dams below 1900 m the overall incidence for the year in at risk communities was 60.6 episodes/1000 child months at risk compared with 7.0 in controls; a rate ratio of 8.7 (95% confidence interval 4.2 to 17.9). Similarly for the higher dams overall rates were 6.0 and 1.2; rate ratio 5.0 (2.7 to 9.0). The difference in incidence between the two altitude bands was significant for the at risk communities (rate ratio 10.1; 7.3 to 14.0) and the control communities (5.8; 2.4 to 13.9). In two high altitude dams (2100 and 2225 m) no cases were detected in either at risk or control populations from a total of 7388 child months at risk.

Discussion

It seems clear from these results that microdams close to villages have the potential to increase the incidence of malaria substantially among children living nearby. Malaria at all of these study sites is hypoendemic as they are at altitudes where malaria transmission is marginal. The effect of altitude on incidence is clearly important; our results suggest an appreciably greater amount of malaria in villages near dams at altitudes below 1900 m than above. At both altitudes, however, there was also a similar difference between at risk and control communities, suggesting that the dams affect different intensities of existing transmission. The seasonal pattern of incidence in both at risk and control communities generally follows the rainfall pattern, as shown in the figure, with the peak incidence slightly later than the peak rainfall. In April to May, when transmission would be expected to be at its lowest, the incidence ratio between at risk and control communities was similar to that in the peak transmission season, suggesting that the dams are increasing established patterns of transmission throughout the year. This is important because it leads to greatly increased levels of malaria at the end of the transmission season, as seen in other areas of low exposure to P falciparum.14 The observation of no cases in either group at two of the higher dams, however, does not support the possibility that dams might facilitate transmission in areas otherwise free from malaria.

Malaria is probably the major threat to health and life in Africa and so any increased risk has to be considered carefully. In this case, compared with many locations in Africa, the risk is still moderate—even near dams only around 16% of children might be expected to be infected each year. On the other hand, these villages generally consist of children who have not been continually challenged by the disease from infancy and who are therefore unlikely to have developed protective immunity. The dams thus represent a risk of serious illness or death to children in these communities. This can reasonably be extrapolated in terms of economic loss and also risk to older members of the community, who are also not immune or only partly so Effective malaria control strategies therefore need to be implemented in communities exposed to these risks.

Our findings also raise serious concerns about the potential effects of ecological and agricultural development on public health While in the light of past severe droughts in this region the need to conserve and store water is crucial for human life, animal husbandry, and agriculture, we have shown that an obvious solution such as the construction of dams may also have unforeseen detrimental effects. Many more dams are planned in the Tigray region. These plans will need to be carefully reviewed in the light of our findings so that, when possible, future dams will be sited at the highest feasible altitudes and as far away as possible from villages.

In the next phase of our research we plan to offer bednets to villagers living close to existing dams to test the hypothesis that using bednets can counteract the increased risks associated with the dams At the same time the economic benefits of the dams will be assessed and compared with the cost of bednets. Bednets will be provided under a revolving fund, cost recovery basis and the sustainability of this evaluated. The effects of these dams on schistosomiasis is also being investigated.

The importance of continuing intersectoral collaboration in assessing the health impact of the dam programme and in working together to maximise the benefits of the dams while also minimising the risks is clearly crucial. The community based malaria control programme already in place15 and the strong commitment shown by the regional government are important elements in this endeavour. This is also an important principle for similar development programmes elsewhere.