Impact of Jittering on Malaria Risk Estimates in Cameroon
Author Information
Author(s): Massoda Tonye Salomon G., Wounang Romain, Kouambeng Celestin, Vounatsou Penelope
Primary Institution: Swiss Tropical and Public Health Institute, Basel, Switzerland
Hypothesis
How does jittering of DHS cluster locations affect geostatistical model-based estimates of malaria risk?
Conclusion
Jittering of cluster locations modified the interpretation of the relationship between environmental predictors and malaria transmission, but the risk estimates remained comparable to those generated using true locations.
Supporting Evidence
- Jittering increased prediction error but did not significantly alter the parameter estimates of socio-economic factors.
- Distance to water bodies and presence of forest were the most affected predictors by jittering.
- Altitude and vegetation index were the least affected predictors.
Takeaway
When researchers move the locations of survey clusters a little bit to protect privacy, it can change how we understand the factors that affect malaria risk, but the overall risk estimates stay similar.
Methodology
The study generated one hundred datasets by jittering the original MIS data and applied Bayesian geostatistical variable selection to identify important climatic predictors and malaria intervention coverage indicators.
Limitations
The study focused on a single dataset from Cameroon, which may limit the generalizability of the findings to other contexts.
Participant Demographics
{"total_population":24000000,"urban_population_percentage":51,"children_age_range":"6 to 59 months","malaria_prevalence":{"overall":"33%","rural":"43%","urban":"19%"}}
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website