Summary
A global analysis of 58,319 outbreak events across 32 diseases found shared landscape associations for several vector-borne infections, but less consistent patterns for directly transmitted zoonoses. The study also found that outbreak reporting was strongly associated with access to healthcare.
A study published in Nature on 23 September 2026 analysed 58,319 recorded outbreak events across 32 emerging infectious diseases and 169 countries. It found that outbreak patterns were associated with landscapes where people and livestock live alongside forests and fragmented ecosystems, but the relationships varied across diseases. Access to healthcare also had a strong influence on where outbreaks were recorded.
The researchers assembled geolocated outbreak and case data spanning 1910 to 2022, then used Bayesian geospatial logistic regression to examine associations with 16 social and environmental factors. Their models compared outbreak locations with population-weighted background locations and accounted for spatial patterns and local factors that could affect detection and reporting.
Landscape associations vary by disease
Across the combined dataset, outbreak risk was associated with greater forest cover, more fragmented landscapes and higher livestock density after adjustment for detection-related factors. The authors describe the resulting pattern as a landscape mosaic in which people, livestock and ecosystems meet. The findings are consistent with the idea that contact among people, domestic animals and wildlife can shape disease risk, but the models estimate geographic associations rather than the causal effect of any one landscape change.
The disease-by-disease results were not uniform. Vector-borne diseases, including infections spread by mosquitoes and ticks, showed more consistent associations with ecosystem characteristics and long-term climate drying. The study linked drying trends to outbreak risk in several vector- or water-borne diseases, including dengue, Rift Valley fever, melioidosis and Japanese encephalitis.
By contrast, directly transmitted zoonoses—including Ebola, MERS and mpox—shared few consistent geographic drivers in the analysis. The authors point to differences in pathogen ecology and human exposure, as well as sparse outbreak records for some diseases, as reasons that patterns may vary. The results suggest that prevention strategies based on ecosystem change are more likely to be useful when tailored to particular diseases and landscapes than applied as a single approach to all emerging infections.
Healthcare access shapes the outbreak record
A prominent finding was the relationship between reported outbreaks and proximity to healthcare. For most diseases with suitable data, the odds of an outbreak being reported fell as motorized travel time from a health facility increased. Across 22 disease models, the median decline was 32% for each additional hour of travel; estimates ranged from 1.2% to 96.7%.
This pattern likely reflects the role of clinics, diagnostic laboratories and other health-system resources in identifying and reporting infections. It also means that maps of recorded outbreaks partly show where detection is more likely, not only where infections occur. The authors argue that improving access to diagnosis and care, alongside surveillance at human–animal interfaces, can help detect outbreaks and support response, particularly in underserved and remote communities.
The analysis draws on presence-only outbreak records: it contains locations where cases were detected, rather than confirmed disease-free locations. Its statistical framework addresses several sources of geographic reporting bias, but the study is a spatial analysis of recorded events. Its findings identify patterns and candidate drivers; they are not direct experimental measurements of the effects of interventions.