El Niño Southern Oscillation as an early warning tool for malaria outbreaks in India

Originally published
View original


Malaria Journal
Dhiman and Sarkar Malar J (2017) 16:122
DOI 10.1186/s12936-017-1779-y


Background Malaria is a major public health problem in India, distributed in all the 36 States and Union Territories. The western part is outbreak prone, while the central and eastern parts are endemic to malaria. It is a proven fact that malaria is most sensitive to climatic parameters. Application of rainfall for early warning of malaria has been attempted in India dating back to 1921, which continued to be used by the Government of Punjab till 1946. Owing to the changes in malariometric indices and sociological development, the indices used for early warning are obsolete now-a-days. In the year 1994, the usefulness of ‘El Niño and the Southern Oscillation’ (ENSO) in early warning of malaria outbreaks was found for the first time in Indi. ENSO is an atmosphericcum-oceanic phenomenon that develops over Pacific Ocean. It’s warm (El Niño) and cold (La Niña) phases result into significant climatic anomalies worldwide.

Since 1994, many countries have found the relationship between ENSO and malaria outbreaks but in India, there is no revisit on the subject. ENSO has been found to have strong link with Indian Summer Monsoon Rainfall (ISMR), based on which early forecast of malaria can also be possible. India being a country with vast geographical area, it is known that all regions don’t have the same effect of ENSO on malaria. Therefore, the present study was undertaken to assess the link between ENSO, ISMR and malaria outbreaks at state level in India. The generated knowledge would be useful in early warning of malaria outbreaks for early preparedness and response by the National Vector Borne Disease Control Programme (NVBDCP) for containment of outbreaks.



Data on annual state-wise malaria incidence of India was extracted from ‘National Health Profile reports’ downloaded from the Central Bureau of Health Intelligence (CBHI for years 1994–2010, and for years 2011–2015, malaria data was downloaded from National Vector Borne Disease Control Programme (NVBDCP). Monthly Oceanic Nino Index (ONI) values at Niño 3.4 regions (170°E to 120°W longitude and 5°N to 5°S) for the corresponding period were downloaded from Climate Prediction Center, Center for Weather and Climate Prediction, NOAA, USA(CPC-NOAA). Statewise mean ISMR data (1951–2000) and monthly rainfall data (1994–2015) were extracted from ‘Annual Summary reports’ and ‘Monsoon Reports’ downloaded from Indian Meteorological Department (IMD, Pune).

Data processing

As the malaria cases across the different states of India vary greatly, ‘malaria case index’ that can bring them in single measurable scale compatible to ONI index was calculated to ease the analysis. For the purpose, the ‘malaria case index’ defined by ‘relative change of no. of cases from mean’ was calculated by: where x implies annual malaria cases of a year, and x˙ implies average annual malaria cases of the state (1994–2015).
Similarly ‘rainfall index’ defined by ‘percentage rainfall deviation from mean’ was calculated, whereby rainfall during ISMR period i.e., June, July, August and September was considered in the study as they cover 80% of rainfall received by the country,
Malaria case index = (1) x − ˙x (over) x˙ where x implies ISMR (JJAS) of a year, and x implies mean ISMR (1951–2000).

Studies undertaken in India also have found that ONI value during the pre-monsoon season do not have any predictive value for ensuing monsoon rainfalls [20]. In northern hemisphere, El Niño event peaks during winters, therefore, the winter ONI values were considered for correlation with ISMR in the following months (concurrent ISMR) in India. For the purpose of correlation with various attributes, four month average positive ONI values (+ winter ONI) from November to February were extracted.

Data analysis

In order to determine the impact of El Niño on malaria, a three-tier analysis was conducted, whereby, in step one, correlation coefficients (r) between ‘rainfall index’ and ‘+ winter ONI’ was determined, followed by, the correlation between ‘malaria case index’ and ‘rainfall index’, and ‘malaria case index’ and ‘+ winter ONI’. The resultant correlation between ‘malaria case index’ and ‘+ winter ONI’ was used on geographical information system (GIS) platform to generate spatial correlation map. ‘Natural Neighbour’ algorithm was used for spatial interpolation of the entire data set. This method allows the creation of highly accurate and smooth surface models from very sparsely distributed datasets. The ‘r’ value greater than 0.2 and less than −0.2 were considered as significant correlation at p < 0.05.