Research note | Open | Published:
Spatial modeling, prediction and seasonal variation of malaria in northwest Ethiopia
BMC Research Notesvolume 12, Article number: 273 (2019)
The aim of this study was to determine the spatial modeling, seasonal variation of malaria and making prediction map of malaria in northwest Ethiopia.
The overall average cumulative annual malaria incidence during the study period was 30 per 100 populations at risk. The highest proportion (29.2%) was observed from June 2015 to October 2016. In temporal analysis of clusters, the epidemic was observed from 2015/7/1 to 2016/12/31 throughout the study period in all districts. Hotspot areas with high clusters (p < 0.001) were observed in Metema district it accounts 18.6% of the total malaria cases. An area of high median predicted incidence proportion (> 50%) was seen in the southwest part of the region. Most of the northern part of the study area was predicted to have a low median incidence proportion (< 10%).
Malaria is a mosquito-borne infectious disease of humans caused by the genus Plasmodium, which are introduced into the circulatory system by the bite of an infected female anopheles mosquito . It is one of the major public health challenges undermining development in the poorest countries [1,2,3,4].
Malaria is largely seasonal in Ethiopia, with the major incident occurring during the rainy.
Season from April to October . In Ethiopia 75% of the country is malarious with about 60% of the total population living in areas at risk of malaria. That is, 50.6 million people are at risk from malaria, and four to five million people are affected by malaria annually [1, 6].
The study area is one of the lowland malarious regions in Ethiopia. The dominant species of malaria in the region are both P. falciparum and P. vivax . It accounts 19% of the national malaria burden [7, 8]. Currently, the study area accounts 31% (1.3 million cases) of Ethiopia’s malaria burden .
Malaria mosquitos are aggregated over large areas and time periods. There are few studies examining the extent and drivers of local variation in malaria exposure.
In geographical location, close proximity share common exposures which influence the disease outcome. Ignoring the potential spatial correlation in neighboring areas due to common exposure could result in incorrect model estimates. A little research was conducted by considering spatial correlation. But, using a geostatistical modeling with Bayesian framework takes into account spatial clustering by introducing location-specific random effect parameters in the covariance matrix of a function of distance between locations.
The findings of this study can be used to increase the evidence for targeting control measures and will contribute to the development of models capable of predicting future malaria scenarios.
A repeated cross-sectional study was conducted from 2014 to 2017 in the north Gondar zone, northwest Ethiopia, which is located in the north western part of Ethiopia (Additional file 1).
Topographically, the study area contains lowlands, 552, to highlands, 4620 m above sea level . The study area has had 28.7 °C; 14.1 °C and 0% to 87.5% mean maximum and minimum temperature and relative humidity respectively. The study area also had and two rainy seasons: the main one is from June to September, followed by a shorter one from March to May. The dry season ranges from October to February .
Malaria data were obtained from monthly reported surveillance forms of north Gondar zone health offices between early 2014 and late 2017. The data were collected from the study population, that is patients who visit health institutions in the study area and were aggregated at woreda levels. The spatial coordinates (the latitudes and longitudes) for each woreda were obtained from the Ethiopian demographic and health survey (EDHS) GPS data reference.
For clustering and spatial pattern was detected by using ArcGIS and Sat Scan™ software, version 9.1  using the Kulldorf method.
For computing hypothesis (Ho) test and the presence spatial autocorrelation we used global and local Moran’s I test statistic . Significant clustering, variable or dissimilar patterns and random patterns were declared when the mean Moran’s I values are positive, negative and zero respectively. Z score values were used to see local clustering of malaria cases. Z score values above 1.96 and below negative 1.96 were used to show hotspot and coldspot areas respectively and any value between the two shows random distribution of cases. Interpolation was done using Empirical Bayesian kriging for predictions and smoothing.
About 916,204 malaria cases which were reported from the study period, 69.8% were plasmodium falciparum, 26.8% plasmodium vivax and 2.2% of the case were mixed infected. 71.78% of total malaria cases were observed in adults aged above 15 years old and 8.5% were childrens under 15 years old.
We have identified two major malaria transmission periods, the major transmission time starts from mid-April with highest occurrence in June and July this falls to its lowest in December. The second high transmission was from October to November (Fig. 1).
Even though the cumulative annual malaria incidence was 30 per 100 populations at risk, it has showed a significant variation across districts and months. The highest incidence (44.8%) occurred in Metema district from July to November 2016. Whereas, lowest incidence (5%) occurred in Debark Town administration district from late November to the end of December 2014. From the total malaria cases reported the largest percentage (29.2%) occurred from June 2015 to October 2016 and the lowest (19.1%) observed in 2017 (Additional file 2).
Spatital clustering of cases was detected; Gendawuha Town Administration and Metema districts being the most likely clusters (LLR = 184,968.375, p < 0.001). Secondary cluster was West Armachiho district (LLR = 8936.64, p < 0.001) and the third cluster was Quara district (LLR = 36,249.66, p < 0.001) (Table 1).
High rates of spatio-temporal malaria clusters were identified in Dembia district (LLR = 1628.51, p < 0.001) from 2014/01/01 to 2017/12/31. Secondary clusters were identified in the East Belesa district (LLR = 1469.21, p < 0.001) from 2014/1/1 to 2017/12/31 (Additional file 3).
Purely temporal high malaria clusters were observed from 2015/7/1 to 2016/12/31 (LLR = 66, 53.42, p < 0.001) throughout the study period across all districts.
Hot spot areas with high cluster of malaria transmission and cold spot areas with low level clusters were identified. Hotspot areas with high clusters (p < 0.001) were seen in Metema district it covers 18.6% of the total malaria cases (Fig. 2).
Districts that had lower malaria incidence proportion in the region are indicated by high bright green color on the map and they were clustered around the North and east part of the study area, while the high malaria incidence proportion are indicated by high dark and less dark red color which are located in the Western part of the study area.
The maximum peak, where spatial clustering highly pronounced is at a distance of 135.778 kms with corresponding Z score of 9.81 (p-value < 0.00). This distance band is used for the analysis of hot spot and cold spot clusters (Additional file 4).
The global Moran’s index statistic for the malaria incidence per 100,000 population was 0.31 (p-value = 0.007), indicating the presence of significant positive spatial autocorrelation over the whole study area.
In this analysis the z-score with corresponding p-values showed that there is statistically significant autocorrelation at 0.1, 0.05 and 0.01 level of significance (Additional file 5).
An area of high median predicted incidence proportion (> 50%) was seen in the southwest part of the region. Most of the northern part of the study area was predicted to have a low median incidence proportion (< 10%). The area of highest predicted incidence proportion within the study area corresponds with the majority of high incident districts. The black ramp color indicates the predicted malaria high risk areas and red ramp color indicates less risk areas of malaria (Additional file 6).
In study area, the overall average cumulative malaria incidence proportion was 30%. Thus, results were higher than studies conducted in other parts of Ethiopia (9.7%) , (4.3%) , (6.8%) , (10.4–13.5%), (7.6–14.1%), (5.4%)  and (4.5%) . But, lower than studies conducted in 2011 (43%) and in 2014 (33%) in Tigray regional state . Higher malaria incidence in the study area might be the result different climatic variables [16,17,18,19,20]; the study area has a long rainy season and temperature variability .
The dominated species plasmodium falciparum were covered 69.8% of the total malaria cases. Other studies also agreed that plasmodium falciparum were the dominant species in the northwest parts of Ethiopia [15, 22] and both Plasmodium falciparum and Plasmodium vivax is common in Amhara region . The reason may relate to temperature; temperatures more than 18 °C for p. Falciparum and more than 15 °C for vivax are suitable for the growth of these two species. Obviously, in the most part of the country, especially in northwest Ethiopia, temperature is greater than the specified minimum scale . The above idea was supported by researchers in Tanzania .
The 71.8% of malaria victims were adults aged above 50 years. This is in agreement with other studies [12, 14]. However, other studies reversely stated that malaria infection decreased with increasing age group (70); it reasoned out higher body temperature individuals had more susceptible to malaria, which are incapable of clearing parasites more effectively as adults. Even though, the study area is an active investment and agricultural hot spot areas in Ethiopia whereby there are frequent adult immigration from another part of the country.
The 29.2% cases from the total cases were recorded from July 2015 to December 2016. It is supported by other studies in Ethiopia [9, 15, 25], because of high malaria burden in high-risk geographic areas .
The major transmission period were observed from the beginning of mid-April and peaked in June and July, declining at the end of December. The second peak of was observed between October and November. Other studies agreed that in Ethiopia season there are two peak malaria transmission periods, but the occurrence months were varied [27, 28]. However, whatever the starting point all agreed that malaria transmission were high at the first wet season (summer rains) while the latter followed the end of the wet season .
The spatio-temporal models indicated that the most likely cluster was located in Dembia district and in East Belesa. All these identified clusters are closely related to a specific geographical area and share similar geographical parameters, such as altitude and weather conditions.
Spatial clusters were detected in Metema, Tach Armachiho and Gendawuha town administration districts, all of which are located in the Ethiopia–Sudan border regions of the country. These districts are located far from referral hospitals and contain a high number of seasonal migrants due to the presence of agricultural investments and malaria is poorly controlled in these travelers and hard to reach populations. This highlights the risk of cross-border transmission of malaria in Ethiopia and Sudan, particularly in predominantly male migrant populations.
The highest predicted malaria incidence was observed in the region of Dembia, Takusa, Metema district (North West part of the zone) this region can be attributed to sesame (local name “selit”) growing which is a predominant economic activity. The sesame paddies are suitable habitats for malaria vector breeding. In southeastern region and central North part of malaria is low largely due to its location in the highlands .
Malaria is still a major public health problem in the study area with an overall average cumulative annual incidence of 30 per 100 populations at risk. This malaria incidence proportion in the region was higher than the country statistics. Malaria incidence has also showed a significant spatio-temporal clustering. Most of the northern part of the study area is predicted to have a low median incidence.
As the data were obtained from passive surveillance system all clinical records did not fully capture the level of malaria transmission in the districts because some people do not report to the formal clinics, either using traditional therapists or self-diagnose and purchase their own drugs.
Availability of data and materials
The data upon which the result based could be accessed based on a reasonable request to the corresponding author
Communicable Disease Control
Central Statistical Agency
Federal Minister of Health
geographical information system
geographical positioning system
World Health Organization
WHO. World malaria report 2015. World Health Organization; 2016.
Hung Q, Vries PJ, Giao PT, Nam NV, Binh TQ, Chong MT, et al. Control of malaria: a successful experience from Viet Nam. Bull World Health Organ. 2002;80(8):660–6.
Tilaye T. Assessment of malaria prevalence and knowledge, attitude and practice (KAP) in relation to malaria prevention and control in Gondar town, northern Ethiopia. 2005.
WHO. World malaria report. 2017.
Alemu A, Abebe G, Tsegaye W, Golassa L. Climatic variables and malaria transmission dynamics in Jimma town, South West Ethiopia. Parasit Vectors. 2011;4(1):30. https://doi.org/10.1186/1756-3305-4-30.
Dereje B, et al. Spatial modelling of malaria prevalence and its risk factors in rural SNNPR, Ethiopia: classical and bayesian approaches. Am J Theor Appl Stat. 2017;6(6):254. https://doi.org/10.11648/j.ajtas.20170606.11.
Public Health Emergency Manegment (PHEM). A.N.R.S.H.B.P.H.E.M. Annual malaria report. 2012.
WHO. Global malaria programme: updated WHO policy recommendation: single dose primaquine as a gametocytocide in Plasmodium falciparum malaria. Geneva: World Health Organization; 2012. http://www.who.int/malaria/publications/atoz/who_pq_policy_recommendation/en/.
Élodie AV. Assessing the risk factors associated with malaria in the highlands of Ethiopia: what do we need to know? Trop Med Int Health. 2017;2:4.
Kulldorff M. SaTScan user guide for version 9.0. 2010.
Blossom JC, Finkelstein JL, Guan WW, Burns BFJ, et al. Applying GIS methods to public health research at Harvard University. J Map Geogr Libr. 2011;7(3):349–76. https://doi.org/10.1080/15420353.2011.599770.
Alemu K, Worku A, Berhane Y, Kumie A. Spatiotemporal clusters of malaria cases at village level, northwest Ethiopia. Malar J. 2014;13(1):223. https://doi.org/10.1186/1475-2875-13-223.
Adhanom T, et al. Epidemiology and ecology of health and disease in Ethiopia. Addis Ababa: Shama Books; 2006. p. 556–76.
Ethiopia P.S.M.I. Ethiopia malaria operational plan FY. 2017. http://www.pmigov/docs/default-source/default-document-library/malaria-operational-plans/fy14/ethiopia_mop_fy14pdf?sfvrsn=14Accessed.
Teklay H, et al. Pattern and trend of malaria morbidity and mortality in Tigray Region, Ethiopia from 2011/12–2014/15. Bioanal Biomed. 2017;9:2.
Smith MW, Macklin MG, Thomas CJMM, Thomas CJ. Hydrological and geomorphological controls of malaria transmission. Earth Sci Rev. 2013;116:109–27. https://doi.org/10.1016/j.earscirev.2012.11.004.
Hardy AJ, Gamarra JG, Cross DE, Macklin MG, Smith MW, Kihonda J, et al. Habitat hydrology and geomorphology control the distribution of malaria vector larvae in rural Africa. PLoS ONE. 2013;8(12):e81931. https://doi.org/10.1371/journal.pone.0081931.
Shaman J, Day JF. Reproductive phase locking of mosquito populations in response to rainfall frequency. PLoS ONE. 2007;2(3):e331. https://doi.org/10.1371/journal.pone.0000331.
Paaijmans KP, Wandago MO, Githeko AK, Takken W. Unexpected high losses of Anopheles gambiae larvae due to rainfall. PLoS ONE. 2007;2(11):e1146. https://doi.org/10.1371/journal.pone.0001146.
Martens WJ, Niessen LW, Rotmans J, Jetten TH, McMichael AJ. Potential impact of global climate change on malaria risk. Environ Health Perspect. 1995;103(5):458–64. https://doi.org/10.1289/ehp.95103458.
Midekisa A, Beyene B, Mihretie A, Bayabil E, Wimberly MC. Seasonal associations of climatic drivers and malaria in the highlands of Ethiopia. Parasit Vectors. 2015;8(1):339. https://doi.org/10.1186/s13071-015-0954-7.
Alemu K, Worku A, Berhane Y. Malaria infection has spatial, temporal, and spatiotemporal heterogeneity in unstable malaria transmission areas in northwest Ethiopia. PLoS ONE. 2013;8(11):e79966. https://doi.org/10.1371/journal.pone.0079966.
BradfieldLyon E. Temperature suitability for malaria climbing the Ethiopian Highlands. Environ Res Lett. 2017;12:064015. https://doi.org/10.1088/1748-9326/aa64e6.
Bodker E. Relationship between altitude and intensity of malaria transmission in the Usambara Mountains, Tanzania. Med Entomol. 2003;40:706–17.
Deribew A, et al. Incidence, prevalence and mortality rates of malaria in Ethiopia from 1990 to 2015. Malar J. 2017;16:271.
Berhanu K, Poulton C. The political economy of agricultural extension policy in Ethiopia: economic growth and political control. Dev Policy Rev. 2015;33:s2.
Tamiru MA, et al. Malaria outbreak investigation in Mecha, Dera and Fogera Districts, Amhara Region, Ethiopia. Am J Health Res. 2014;2(4):182–7. https://doi.org/10.11648/j.ajhr.20140204.23.
Odongo-Aginya E, Ssegwanyi G, Kategere P, Vuzi PC. Relationship between malaria infection intensity and rainfall pattern in Entebbe peninsula, Uganda. Afr Health Sci. 2005;5(3):238–45.
Yeka. Malaria in Uganda: challenges to control on the long road to elimination: I. Epidemiology and current control efforts. 2016. http://www.sciencedirect.com/science/article/pii/S0001706X11000611.
We would like to thank North Gondar zone health office and west Amhara meteorology agency for providing the data.
Ethics approval and consent to participate
The study was approved by the Institutional Review Board of the University of Gondar. The research was done based on record review without contacting patients. Support letters were obtained from north Gondar health offices for retrieving retrospective malaria data from records. All the information was kept confidential and no individual identifiers were collected.
Consent for publication
The authors declare that they have no any competing interests.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.