Can the early spatial distribution of HIV explain today's heterogeneity in HIV prevalence across Malawi? A flexible individual-based mathematical model

Abstract Background HIV prevalence in Malawi varies greatly across the country, ranging between 4% and 22% in the 28 administrative districts in 2010. We built a mathematical model to study if this heterogeneity could be explained by the distribution of the early cases of HIV across the country. Methods We developed an individual-based stochastic mathematical model for HIV transmission, incorporating a geographical structure and an unrestricted number of individual-level factors. We ran the model with five different geographical divisions for Malawi between years 1975-2030: (I) no geographical structure; 28 administrative districts including (II) only permanent relocations between districts, (III) permanent relocations and between-district casual sexual relationships, or (IV) permanent relocations between districts and to/from abroad and between-district casual sex; and (V) a grid of 10×10km2 cells, with permanent relocations and between-cell casual relationships. We assumed that HIV was present in 1975 in the districts where the prevalence was >10% in 2010. We calibrated the models to national and district-level prevalence estimates. Results Reaching the national prevalence estimate required the entire adult population to have at least 20 sex acts with irregular partners during each year until year 1990. Models with district- and cell-level division performed equally well, reproducing the observed heterogeneity in prevalence if casual relationships between geographical districts were either excluded (Model II) or restricted to minimum (Models III, V). Inclusion of long-distance casual partnership mixing (Models III-V) led to a more rapid mitigation of the differences in prevalence between districts; international migration evened out the differences almost completely (Model IV). National prevalence is expected to continue to decrease to 4-6% by 2030. Conclusion Our model could sustain the differences in HIV prevalence across Malawian districts from 1975 until present, assuming that the casual relationships between districts are kept on a sufficiently low level. This suggests that HIV may have been introduced first to the Southern part of Malawi. The role of sociobehavioural factors needs further investigation.