The probability of Plasmodium vivax acute illness following primary infection and relapse in Papua New Guinea

Inoculation with Plasmodium vivax malaria parasites can lead to blood-stage infections from the primary infection and relapses from liver-stage parasites or non-circulating merozoites. Understanding the risk of clinical illness following primary infection and relapse would inform surveillance and intervention strategies, but the probabilities are uncertain in people living in endemic areas. A major difficulty lies in the inability to distinguish primary infections and relapses. In this study, we estimate the probabilities of clinical illness using the different seasonal patterns of primary infection and relapse. Children aged one to three years in Ilaita, Papua New Guinea, were followed up over 16 months for illness (fever with ≥500 parasites/µl) with fortnightly active and passive case detection, and for blood-stage infection every two months. Estimates of the number of primary infections and relapses for each two-month time-period, age-group, village and ITN use category were derived from previous analyses using genotyping data. In this study, we use a Bayesian statistical model to relate the number of observed P. vivax clinical cases in each covariate category to the expected numbers of primary infections and relapses. We include the cumulative number of primary infections experienced since birth as a proxy for acquired immunity. To reflect uncertainty, we use varying assumptions about whether relapses can cause illness in different circumstances. The probability of illness decayed exponentially with increasing cumulative numbers of primary infections experienced. The estimated probability of illness following relapses was lower than that for primary infection, how much lower depended on how they were defined. Later relapses within the same brood tended to have lower probabilities than earlier ones. Varying seasonally, relapses were estimated to contribute half of P. vivax illness in this cohort despite accounting for 80% of the force of blood-stage infection. The results can inform estimates of the burden of P. vivax and provide building blocks for mathematical models for predicting the impact of interventions. Interventions triggered by clinical cases would focus on more recent infections and age-groups with less acquired immunity.