Strengthening malaria and climate research in Ethiopia

Lindtjorn-Malaria conference posterLindtjorn B, Loha E, Deressa W, Balkew M, Gebremichael T, Sorteberg A, Woyessa A, Animut A, Diriba K, Massebo F, et al: Strengthening malaria and climate research in Ethiopia. Malaria Journal 2014, 13:P56.

Poster presentation

The project “Ethiopian Malaria Prediction System” implemented from 2007 to 2012 combined new population-based malaria transmission information with climate and land use variability data to develop an early warning tool to predict malaria epidemics in Ethiopia. Scientists from Ethiopia and Norway collaborated to incorporate climate variability and forecast information for malaria epidemics.

Our study shows that the association between weather and malaria is complex. Statistical models can predict malaria for large areas. However, as malaria transmission varies and depends on local environmental conditions, we need to have good and local knowledge about each area. However, weather variability is the main driver of malaria in Ethiopia.

While the generation of precipitation depends on local ascent and cooling of the air, our research provided new data on the transport of moisture into the country that may improve weather forecasting. We developed a new classification of climate zones, have mapped drought episodes in Ethiopia during the last decades, and have improved seasonal weather forecasting. Our hydrology studies show that potential climate change differs among the Ethiopian river basins, with river flows being sensitive to variations in rainfall, and less to temperature changes.

The computer model, Open Malaria Warning, incorporates hydrological, meteorological, mosquito-breeding, land-use data, and cattle densities to find out when and where outbreaks are likely to occur. We validated the model with data for malaria transmission in the highlands and lowlands, characterizing malaria transmission over some years in both highlands and lowlands. This provided us with new knowledge on malaria transmission in Ethiopia, how intense the seasonal transmission is, and how malaria occurs in different populations and areas. Our study showed that indigenous malaria transmission during a non-epidemic year takes place above 2000 m altitude. We also showed the ideal temperature for malaria transmission is about 25°C, underlining that global warming may lead to increased risk of malaria in highland areas, and less in the lowlands with already high average temperatures. However, to validate such models, there is a need for several years of active monitoring of malaria cases and mosquito densities. Unfortunately, such data is rare in Africa, and we need to invest in long-term monitoring of malaria transmission.

Lindtjorn-Malaria conference poster

Climate, food production and malnutrition

Seifu Hagos, Torleif Lunde, Damen H Mariam, Tassew Woldehanna and Bernt Lindtjørn. Climate change, crop production and child under nutrition in Ethiopia; a longitudinal panel study. BMC Public Health 2014, 14:884 doi:10.1186/1471-2458-14-884

Background  The amount and distribution of rainfall and temperature influences household food availability, thus increasing the risk of child undernutrition. However, few studies examined the local spatial variability and the impact of temperature and rainfall on child undernutrition at a smaller scale (resolution). We conducted this study to evaluate the effect of weather variables on child under nutrition and the variations in effects across the three agroecologies of Ethiopia.

Methods  A longitudinal panel study was conducted. We used crop productions (cereals and oilseeds), livestock, monthly rainfall and temperature, and child under nutrition data for the period of 1996, 1998, 2000 and 2004. We applied panel regression fixed effects model.

Results  The study included 43 clusters (administrative zones) and 145 observations. We observed a spatiotemporal variability of rainfall, stunting and underweight. We estimated that for a given zone, one standard deviation increase in rainfall leads to 0.242 standard deviations increase in moderate stunting. Additionally, a one standard deviation increase temperature leads to 0.216 standard deviations decrease in moderate stunting. However, wasting was found to be poorly related with rainfall and temperature. But severe wasting showed a positive relationship with the quadratic term of rainfall.

Conclusions  We conclude that rainfall and temperature are partly predicting the variation in child stunting and underweight. Models vary in predicting stunting and underweight across the three agroecologic zones. This could indicate that a single model for the three agroecologies may not be not applicable.

Ownership and use of long-lasting insecticidal nets for malaria prevention in Ethiopia

Woyessa A, Deressa W, Ali A, Lindtjorn B. Ownership and use of long-lasting insecticidal nets for malaria prevention in Butajira area, south-central Ethiopia: complex samples data analysis. BMC public health 2014; 14: 99.

BACKGROUND: Despite the encroaching of endemic malaria to highland-fringe areas above 2000 meters above sea level in Ethiopia, there is limited information on ownership and use of mosquito nets for malaria prevention. Thus, this study was designed to assess long-lasting insecticidal nets (LLIN) possession and use for malaria prevention in highland-fringe of south-central Ethiopia.

METHODS: A multi-stage sampling technique was employed to obtain household data from randomly selected households using household head interview in October and November 2008. Household LLIN possession and use was assessed using adjusted Odds Ratio obtained from complexsamples logistic regression analysis.

RESULTS: Only less than a quarter (23.1%) of 739 households interviewed owned LLINs with more differences between low (54.2%) high (3.5%) altitudes (Χ2 =253, P < 0.001). Higher LLIN ownership was observed in illiterate (adj.OR 35.1 [10.6-116.2]), male-headed (adj.OR 1.7 [1.051-2.89]), owning two or more beds (adj.OR 2.7 [1.6-4.6]), not doing draining/refilling of mosquito breeding sites (adj.OR 3.4 [2.1-5.5]) and absence of rivers or streams (adj.OR 6.4 [3.5-11.8]) of household variables. The presence of ≥2 LLINs hanging (adj.OR 21.0 [5.2-85.1]), owning two or more LLINs (adj.OR 4.8 [1.3-17.5]), not doing draining/refilling of mosquito breeding sites (adj.OR 4.2 [1.3-13.6]), low wealth status (adj.OR 3.55 [1.04-12.14]), and < 1 km distance from absence of rivers or streams (adj.OR 3.9 [1.2-12.1]) of households was associated with more likely use of LLIN. The LLIN ownership was low in the highlands, and most of the highland users bought the bed nets themselves.

CONCLUSIONS: This study found a low household LLIN ownership and use in the highland-fringe rural area. Therefore, improving the availability and teaching effective use of LLIN combined with removal of temporary mosquito breeding places should be prioritized in highland-fringe areas.

Eskindir Loha defended his PhD thesis

On Tuesday September 3, 2013, Eskindir Loha defended his PhD thesis.

The title of the work is: “Variation in malaria transmission in southern Ethiopia: The impact of prevention strategies and a need for targeted intervention”.

Summary of Thesis

In Ethiopia, 60 per cent of the population is at risk of malaria. The transmission of the disease is unstable, and hence, the possibility of epidemics demanded continuous vigilance and preparedness of the health system. Meanwhile, the complexity of the transmission of the disease has become an impediment to retain the effectiveness of prevention and control strategies. Understanding factors that play role in disease transmission at different locations, the pattern of disease transmission, the impact of prevention and control strategies and challenges in control efforts were deemed crucial for the way forward.

This thesis analysed the local variations in the link between potential determinants of transmission – meteorological factors and malaria incidence. For this, we used datasets from 35 locations found in the Southern Nations and Nationalities People’s Region and registered within the period 1998 to 2007. The findings implied that the variability in the models to be principally attributed to regional differences, and a single model that fits all locations was not found. Although there is a biological link between meteorological factors and malaria transmission, the link is affected by local conditions and non-meteorological factors.

With the understanding of a need to incorporate non-meteorological factors, in an attempt to predict disease incidence, a detailed investigation was carried out in Chano Mille Kebele – one of the malarious Kebeles of Arba Minch Zuria district, Gamo Gofa zone, south Ethiopia. A prospective cohort study was conducted for two years with a weekly visit to each of 1,388 households. The findings showed that rainfall increased and indoor residual spraying with Deltamethrin reduced falciparum malaria incidence. Higher disease incidence was observed among males, children 5–14 years old, insecticide-treated net non-users, the poor, and people who lived closer to vector breeding site. Meanwhile, we identified spatio-temporal clusters of high disease rates within a 2.4 area of the Kebele.

Mass distribution of insecticide-treated nets neither showed community-wide benefit nor influenced the spatio-temporal clustering of malaria, though proved to be protective at the individual level. Further analysis on insecticide-treated nets showed that the proportion of insecticide-treated net use reached a maximum of 69 per cent despite a near universal coverage (98.4 per cent) was achieved. Sleeping under the insecticide-treated nets was influenced by gender, age and proximity to the vector breeding site. Factor compromising the usable life of insecticide-treated nets and a lack of convenient space to hang more than one net were reported.

The local variations in meteorology-malaria link, the heterogeneous risk carried by different population segments and the observed effect of prevention strategies may help to revisit the approaches towards malaria – for which I forwarded specific recommendations.

Two new malaria articles

Our malaria research group recently published two new articles about malaria in south Ethiopia:

Massebo F, Balkew M, Gebre-Michael T, Lindtjorn B. Entomologic Inoculation Rates of Anopheles arabiensis in Southwestern Ethiopia. Am J Trop Med Hyg 2013.

Woyessa A, Deressa W, Ali A, Lindtjørn B. Malaria risk factors in Butajira area, south-central Ethiopia: a multilevel analysis. Malaria Journal 2013, 12:273. 

Model to validate species distribution and seasonal variation

This article is a validation of a mathematical model described earlier.Overall, the model gives a realistic representation of seasonal and year-to-year variability in mosquito densities and it can accurately predict the distribution of An. gambiae s.s. and An. arabiensis in sub-Saharan Africa. It may be used for seasonal and long-term predictions of changes in the burden of malaria.

Lunde TM, Balkew M, Korecha D, Gebre-Michael T, Massebo F, Sorteberg A and Lindtjørn B. A dynamic model of some malaria-transmitting anopheline mosquitoes of the Afrotropical region. II. Validation of species distribution and seasonal variations. Malaria Journal 2013, 12:78

Background  The first part of this study aimed to develop a model for Anopheles gambiae s.l. with separate parametrization schemes for Anopheles gambiae s.s. and Anopheles arabiensis. The characterizations were constructed based on literature from the past decades. This part of the study is focusing on the model’s ability to separate the mean state of the two species of the An. gambiae complex in Africa. The model is also evaluated with respect to capturing the temporal variability of An. arabiensis in Ethiopia. Before conclusions and guidance based on models can be made, models need to be validated.

Methods  The model used in this paper is described in part one (Malaria Journal 2013, 12:28). For the validation of the model, a data base of 5,935 points on the presence of An. gambiae s.s. and An. arabiensis was constructed. An additional 992 points were collected on the presence An. gambiae s.l.. These data were used to assess if the model could recreate the spatial distribution of the two species. The dataset is made available in the public domain. This is followed by a case study from Madagascar where the model’s ability to recreate the relative fraction of each species is investigated. In the last section the model’s ability to reproduce the temporal variability of An. arabiensis in Ethiopia is tested. The model was compared with data from four papers, and one field survey covering two years.

Results  Overall, the model has a realistic representation of seasonal and year to year variability in mosquito densities in Ethiopia. The model is also able to describe the distribution of An. gambiae s.s. and An. arabiensis in sub-Saharan Africa. This implies this model can be used for seasonal and long term predictions of changes in the burden of malaria. Before models can be used to improving human health, or guide which interventions are to be applied where, there is a need to understand the system of interest. Validation is an important part of this process. It is also found that one of the main mechanisms separating An. gambiae s.s. and An. arabiensis is the availability of hosts; humans and cattle. Climate play a secondary, but still important, role.

Malaria model

This study highlights some of the assumptions commonly used when constructing mosquito-malaria models and presents a realistic model of Anopheles gambiae s.s. and Anopheles arabiensis, and their interaction.

We make a case that this new mosquito model, OMaWa, may improve the understanding of vector dynamics, which in turn can be used to better understand the dynamics of malaria.

Lunde TM, Korecha D, Loha E, Sorteberg A and Lindtjørn B. A dynamic model of some malaria-transmitting anopheline mosquitoes of the Afrotropical region. I. Model description and sensitivity analysis. Malaria Journal 2013, 12:28

Background: Most of the current biophysical models designed to address the large-scale distribution of malaria assume that transmission of the disease is independent of the vector involved. Another common assumption in these type of model is that the mortality rate of mosquitoes is constant over their life span and that their dispersion is negligible. Mosquito models are important in the prediction of malaria and hence there is a need for a realistic representation of the vectors involved.

Results: We construct a biophysical model including two competing species, Anopheles gambiae s.s. and Anopheles arabiensis. Sensitivity analysis highlight the importance of relative humidity and mosquito size, the initial conditions and dispersion, and a rarely used parameter, the probability of finding blood. We also show that the assumption of exponential mortality of adult mosquitoes does not match the observed data, and suggest that an age dimension can overcome this problem.

Conclusions: This study highlights some of the assumptions commonly used when constructing mosquito-malaria models and presents a realistic model of An. gambiae s.s. and An. arabiensis and their interaction. This new mosquito model, OMaWa, can improve our understanding of the dynamics of these vectors, which in turn can be used to understand the dynamics of malaria.

How bednets and indoor residual spraying influence spatio-temporal clustering of malaria

Loha E, Lunde TM, Lindtjørn B. Effect of Bednets and Indoor Residual Spraying on Spatio-Temporal Clustering of Malaria in a Village in South Ethiopia: A Longitudinal Study. PLoSONE 2012; 7(10): e4735.


Understanding the spatio-temporal pattern of malaria transmission where prevention and control measures are in place will help to fine-tune strategies. The objective of this study was to assess the effect of mass distribution of bednets and indoor residual spraying (IRS) with insecticides on the spatio-temporal clustering of malaria in one malaria endemic village in south Ethiopia.


A longitudinal study was conducted from April 2009 to April 2011. The average population was 6631 in 1346 locations. We used active and passive searches for malaria cases for 101 weeks. SatScan v9.1.1 was used to identify statistically significant retrospective space–time clusters. A discrete Poisson based model was applied with the aim of identifying areas with high rates. PASW Statistics 18 was used to build generalized Poisson loglinear model.


The total number of both types of malaria episodes was 622, giving 45.1 episodes per 1000 persons per year; among these, episodes of Plasmodium falciparum and vivax infection numbered 316 (22.9 per 1000 per year) and 306 (22.2 per 1000 per year), respectively. IRS with Dichlorodiphenyltrichloroethane (DDT) and later with Deltamethrin and free mass distribution of insecticide-treated nets (ITNs) were carried out during the study period. There was spacetime clustering of malaria episodes at a household level. The spatio-temporal clustering of malaria was not influenced by free mass distribution of ITNs; however, the time-span of the spatio-temporal clustering of malaria cases ended after IRS with Deltamethrin. The presence of clusters on the south-east edge of the village was consistent with the finding of an increasing risk of acquiring malaria infection for individuals who lived closer to the identified vector breeding site.


The risk of getting malaria infection varied significantly within one village. Free mass distribution of ITNs did not influence the spatio-temporal clustering of malaria, but IRS might have eliminated malaria clustering.


(This paper was also been presented at a conference that BioMed Central, in conjunction with its journals Malaria Journal and Parasites & Vectors, hosted. The second malaria conference “Challenges in Malaria Research: Progress Towards Elimination” was held in Basel, Switzerland from 10 – 12 October 2012).