Seasonal Impacts on pfhrp2-deletions









Since 2014, false negative rapid diagnostic test (RDT) results due to pfhrp2/3 gene deletions have been reported in 10 countries in sub-Saharan Africa (SSA) (WHO Threat Maps). In February 2018, the WHO issued guidance advising national malaria control programme managers to investigate suspected false-negative RDT results. This included a protocol for estimating the prevalence of false negative RDTs due to pfhrp2/3 deletions by sampling from symptomatic P. falciparum patients seeking treatment at public health facilities, with sampling to be completed within an 8-week interval. The resultant prevalence is then to be used to determine if a switch to a non HRP2-based RDT should be made. The specific interval chosen, however, could lead to bias in the observed prevalence of pfhrp2/3 deletions as a result of seasonal variation in the intensity of malaria transmission.

In response we have extended our original methods to characterise the impact of seasonal variations in transmission intensity on the detected prevalence of clinically relevant pfhrp2-deleted parasites. We conducted 200 simulations for each each level 1 administrative region in SSA with ongoing malaria transmission. For each 8-week interval the observed percentage of clinical cases yielding a false negative RDT (microscopy positive but RDT negative due to pfhrp/32 deletions) was summarised. The true annual proportion of false negative RDT results due to pfhrp2/3 deletion was then subtracted from these summaries to produce the predicted bias for each 8-week interval, which is shown in the first plot. These dynamics are presented when we assume sampling is conducted from all the population and when sampling only from children under the age of 5 years old. These dynamics can then be used to identify the 8-week interval that yields the least biased estimate of the true percentage of clinical cases yielding a false negative RDT, and are designed to assist national malaria control programs when planning their surveillance strategies. The black box shows the optimum 8-week interval for both sampling schemes.

In the second plot, we have shown the raw data from the 200 simulations. Simulations in which the confidence interval included the true value are shown in blue, and those that do not are shown in red. The resultant percentage of simulations that are correct (sampling accuracy) is then presented in the side panel on the left. (For adminstrative units with very low malaria prevalence, a large degree of variation is observed in the raw datau. This is the expected behaviour, with the low prevalence of malaria allowing for more rapid changes in the frequency of pfhrp2/3 deletions to occur).