Comparing malaria early detection methods in a declining transmission setting in northwestern Ethiopia

Area of Science:

• Epidemiology
• Public Health
• Biostatistics

Background:

• Malaria remains a significant global health threat despite progress in incidence reduction.
• Effective surveillance and early detection are crucial for timely public health interventions against malaria outbreaks.
• This study addresses the need for improved malaria outbreak detection methods.

Purpose of the Study:

• To compare the performance of selected event detection methods for identifying malaria outbreaks.
• To evaluate the potential of different algorithms in informing public health responses to malaria resurgence.

Main Methods:

• Utilized historical weekly surveillance data of Plasmodium falciparum cases from Ethiopia's Amhara region.
• Evaluated three categories of detection methods: CDC EARS, weekly statistical thresholds (WHO, Cullen), and Farrington methods.
• Assessed methods based on event detection rates and true positive alarm rates.

Main Results:

• All evaluated methods outperformed a random alarm generator.
• CDC EARS and weekly statistical methods showed high sensitivity but variable specificity.
• Farrington variants demonstrated a wide range of sensitivity and specificity, allowing for flexible performance tuning.

Conclusions:

• The Farrington improved method was most effective, achieving over 70% sensitivity and 70% specificity.
• This method's use of statistical models to control for seasonality and trends makes it highly suitable for malaria early detection.
• Model-based approaches, including the Farrington method, should be broadly considered for malaria surveillance.