Back to Videos

Developing and Improving Statistical Models to Predict Numbers of Homes in Scotland With Internal Pb Piping

Dr. Gail Robertson (Mathematics)/Professor Margaret Graham (GeoSciences)

Lead has been used in plumbing systems since ancient times due to its malleability, low melting point, corrosion resistance and versatility, and remains ubiquitous in UK household plumbing systems. Exposure to lead can have serious effects on human health and lead in drinking water is known to cause neurodevelopmental problems in children with prolonged exposure. The Scottish Government, working via the Drinking Water Quality Regulator (DWQR) and Scottish Water, has reduced the number of lead pipes in the publicly owned water distribution network, but a large number of houses within Scotland are thought to have internal lead piping which could contribute significantly to lead contamination in tap water. We worked with the DWQR to build a statistical model to estimate the total number of houses in Scotland with internal lead piping and identify postcodes with greater proportions of houses with lead piping. We sourced appropriate datasets from Scottish Water and the Scottish Government and used a Poisson distributed hurdle model to identify important risk factors for houses with lead piping and to predict the number of houses in each postcode with lead pipes. We used these predictions to develop a stratified sampling regime to validate our model. We found that our model fitted the available data reasonably well, although estimates in some areas showed lower accuracy when comparing model predictions with tap water sampling data. We suggest improvements that may be made to the model to account for spatial autocorrelation in the data including new data from further sampling. Our study will allow the Scottish Government to estimate the scale of work involved with removing lead piping from residential properties and the effect this will have on improving health in Scotland.