Setting the malaria epidemic threshold in the Central Health Region of Burkina Faso using historical data

Abstract

Introduction: Malaria has been endemic in Burkina Faso. Setting epidemic thresholds is then crucial for early detection and responses. We compared  three methods for epidemic detection in the Central Health Region of Burkina Faso using historical data between 2013-2016.

Methodology: Monthly malaria data from 2013 to 2015 were used as the baseline to set the thresholds. Three methods were applied: quartiles, mean + 2  Standard Deviation (SD), and cumulative sum (C-sum). The median and third quartile, as well as the mean and the mean + 2 SD were calculated per  month, for the baseline period, and plotted in a graph with the monthly malaria cases of 2016. For each month, the number of cases of the previous and  following months between 2013-2015 was summed up and divided by 9. These monthly average numbers were refined with the 1.96 SD and plotted with  the 2016 monthly malaria cases. Any time that the 2016 line crossed the quartiles, the mean and the C-sum thresholds, an unexpected increase in malaria cases was caught.

Results: Cases were higher every month of 2016 compared to the corresponding months of the previous three years. The  Quartiles method detected the whole 2016 year as unusual. Using the mean + 2SD method, malaria cases raised unusually in 2016, except for August,  whereas the C-sum + 1.96 SD method did not detect outbreaks in July.

Conclusions: Not dependent on extreme values, the quartiles method seems more reliable to capture an abnormal rise in malaria cases. This increase in cases would be due to the free healthcare policy for children and pregnant women  launched in 2016. Abnormal rise should always be investigated before confirming an epidemic.