Inferring Lead Risk from Drinking Water Compliance Data

In order to meet the EU requirement of compliance with the 10 µg/l lead standard utility companies in England and Wales implemented a large-scale plumbosolvency treatment programme based around the addition of orthophosphate. This was largely delivered by the end of 2003. This solution has resulted in a major improvement in compliance (circa 99%) as measured through random daytime sampling. A consequence of this substantial improvement is a desire to re-optimise treatment for efficiency, minimise potential environmental impacts and maximize benefits from targeted lead pipe replacement programmes. The extent to which compliance data can be used to provide evidence-based justification for these objectives is examined for supplies in the north-west of England. In the twelve years since the treatment solution was implemented lead non-compliance has steadily fallen from 3.0 % in 2004 to less than 0.3 % over in 2016, despite minimal changes in treatment, indicating that long term structural changes in the corrosion layer are occurring over this timescale. Only by combining regulatory and operational data collected over 5-year period to form a large data set (N=62,810) did significant non-random deviations emerge. The resulting zonal compliance distribution was characterised as a zero-inflated overdispersed Poisson. A large part of the overdispersion was accounted for by the zonal variation in the proportion of properties with lead service pipes. This variation also accounted for the zero-inflated aspect which relates to a substantial number of zones having few or no lead service pipes. Such areas may provide economic opportunity for small lead pipe replacement programmes as an alternative to plumbosolvency treatment. At a regional level compliance varies with property age peaking in the period 1919 and 1940. This is accounted for by the historical selection of lead has the preferred material for service pipes. The analysis shows that in order to assess zonal lead risk from random daytime data it is essential to have quantitative information about the proportion of properties with lead service pipes. Incidentally, it was found that the risk of non-compliance in properties with lead service pipes was nearly thirty times that in properties without lead service pipes. Analysis of compliance data by supply provided only limited justification for re-optimisation of treatment. This issue will get more difficult with further improvement in compliance.