The bacteria, Paenibacillus thiaminolyticus, has been identified as contributing to neonatal sepsis and subsequent post-infectious hydrocephalus (PIH) in Ugandan infants. The absence of infection in mothers suggests that the infants must have been exposed to the bacteria within the first days of life. The spatiotemporal distribution of patients with PIH appears to be inhomogeneous, however the distribution of the bacteria in the environment is currently unknown. Based on an apparent seasonal increase in cases observed by hospital workers when the rains come in, it is hypothesized that the prevalence of the bacteria is related to weather conditions.
This work uses data collected over a 20-year period by the CURE Children’s Hospital of Uganda (CCHU) in Mbale on infants with hydrocephalus. The data includes cases of PIH and non-post infectious hydrocephalus (NPIH). By using NPIH as the control population we estimate the relative risk (RR) of PIH. The point pattern data given by the spatiotemporal coordinates of the PIH and NPIH cases are assumed to be realizations of underlying inhomogeneous Poisson point processes. By examining the ratio of their intensities, we can fit a logistic model to the data.
Our model identifies areas of elevated RR which can be utilized to inform diagnostics and treatment at point-of-care. We demonstrate increased RR i) spatially: in the area north-west of lake Kyoga throughout the study period and ii) temporally: for the years 2006-2012, across the whole of Uganda. By incorporating information on environmental variables, including rainfall, temperature, and the Standardized Precipitation Evapotranspiration Index, we can explain some of this increased RR. In particular, we find evidence of a significant positive association between RR of PIH and rainfall. This information can be used to predict times, locations, and environmental conditions with increased risk of PIH and to inform preventative measures. The results are being used to instruct further investigation into the distribution of Paenibacillus thiaminolyticus, through soil sampling at relevant locations.