Archetype-based stratification to inform sub-national tailoring of malaria interventions in Zambia

Although Zambia does not rank among the highest contributors to global malaria cases, in 2023, it had approximately eleven million cases. Furthermore, Zambia's malaria incidence distribution is spatially heterogeneous at fine scales such as health facility catchment areas (HFCAs). HFCA-level malaria incidence stratification currently guides implementation and targeting approaches. We aim to enhance the current stratification approach by incorporating archetypical characteristics associated with malaria, including environmental, access, epidemiological, and socioeconomic factors. An archetype-based stratification approach may inform the tailoring of interventions to accelerate progress toward elimination.

We used literature-informed data from 2017 to 2024 from the Zambia Health Information Management System and publicly available datasets. A zero-inflated negative binomial mixed-effects regression model was fitted to identify significant variables associated with malaria heterogeneity at the HFCA level. Significant variables were used to archetype HFCAs using Clustering Large Applications algorithm.

Significant variables included: walking travel time to nearest health facility; percentage of malaria attributed to pregnant women; percentage of malaria attributed to children under five (U5); Children U5's bednet (Long-lasting insecticide-treated nets or Insecticide-treated nets) coverage; bednet use rate; elevation; enhanced vegetation index; precipitation; and housing quality. Four new archetypes were derived. Relative to malaria incidence-based stratification, the archetype-based stratification exhibited spatiotemporal stability over time. Across all four new archetypes, elevation, bednet use rate, and housing quality emerged as the top-contributing features. However, all four archetypes exhibited distinct features defining their identity.

The study proposes embedding archetypical characteristics in the malaria incidence-based stratification approach to better inform intervention tailoring and prioritisation based on the distinct archetype features associated with malaria at the HFCAs level. Furthermore, the study provides a precursor for simulating potential intervention mixes based on malaria transmission archetypical characteristics, which can provide decision-makers with projected impact of varied investments in HFCAs with similar archetypical features.