Limitations of malaria reactive case detection in an area of low and unstable transmission on the Myanmar-Thailand border

Background: Reactive case detection is an approach that has been proposed as a tool for malaria elimination in lowtransmission settings. It is an intuitively justified approach based on the concept of space-time clustering of malaria cases. When an index malaria clinical case is detected, it triggers reactive screening and treatment in the index house and neighbouring houses. However, the efficacy of this approach at varying screening radii and malaria prevalence remains ill defined. Methods: Data were obtained from a detailed demographic and geographic surveillance study in four villages on the Myanmar-Thailand border. Clinical cases were recorded at village malaria clinics and were linked back to patients' residencies. These data were used to simulate the efficacy of reactive case detection for clinical cases using rapid diagnostic tests (RDT). Simulations took clinical cases in a given month and tabulated the number of cases that would have been detected in the following month at varying screening radii around the index houses. Simulations were run independently for both falciparum and vivax malaria. Each simulation of a reactive case detection effort was run in comparison with a strategy using random selection of houses for screening. Results: In approximately half of the screenings for falciparum and 10% for vivax it would have been impossible to detect any malaria cases regardless of the screening strategy because the screening would have occurred during times when there were no cases. When geographically linked cases were present in the simulation, reactive case detection would have only been successful at detecting most malaria cases using larger screening radii (150-m radius and above). At this screening radius and above, reactive case detection does not perform better than random screening of an equal number of houses in the village. Screening within very small radii detects only a very small proportion of cases, but despite this low performance is better than random screening with the same sample size. Conclusions: The results of these simulations indicate that reactive case detection for clinical cases using RDTs has limited ability in halting transmission in regions of low and unstable transmission. This is linked to high spatial heterogeneity of cases, acquisition of malaria infections outside the village, as well missing asymptomatic infections. When cases are few and sporadic, reactive case detection would result in major time and budgetary losses.