The predictive risk maps are not without their limitations. When using the maps, the following limitations should be borne in mind:
- Risk in urban areas: Urban slums, with their inadequate sanitation and overcrowding, can be ideal settings for STH transmission, leading to a higher prevalence of infection in urban than in rural areas. However, this is not always the case, and prevalence can also be lower in urban areas because of improved health care and socio-economic status. It is therefore difficult to reliably predict prevalence in urban areas, and the risk maps may under-estimate or over-estimate prevalence in such areas.
- Areas without survey data: The distribution of surveys across Africa is uneven. In areas without suitable data, the ability of the models to predict infection risk may be reduced. In such cases, the model is limited to using environmental information alone, and is pulled towards the mean estimate of prevalence, which in some regions may over-estimate the true prevalence.
- Areas with few data points: In areas with only a few data points, the model gives emphasis to these data. For example, if there is a single data point with a high prevalence in an area with no other data, the models will sometimes predict a high prevalence for the surrounding area, again over-estimating prevalence.
- Areas with extreme environments: In extreme environments the relationships between infection prevalence and environmental characteristics may well be different to less extreme areas. In such cases, the model may over or under-predict the true prevalence, particularly when survey data is lacking.