Evaluating wastewater surveillance for estimating pharmaceutical use

Regional medicine use data is key for health management and in understanding many aspects of health, still such data is missing in many parts of the world. Here, we aimed to evaluate how well analyses of active pharmaceutical ingredients (APIs) in untreated municipal wastewater could be used to predict regional use. We studied 53 APIs measured at three wastewater treatment plants (WWTPs) in Stockholm, Sweden between 2004-2021 together with corresponding, comprehensive sales data from hospitals, pharmacies and other outlets as proxy for use. Conversion factors (CFs) representing recovered mass per gram of sold API were generated for each API using data from two WWTPs. Median absolute deviation normalized by the median (MADM) was used to evaluate variability of CFs over the years. While CFs ranged more than two orders of magnitude between different APIs, mass loads recovered in wastewater were equal or less than the estimated daily use for all except one API. The variability of CF estimates was below 100 % for 43 APIs. When evaluating the predictive performance of the derived CFs on data from the third WWTP, the typical error was <2-fold for 36 APIs (68 %). Further, our usage predictions aligned well for the majority of APIs when applied to data from independent international studies. Neither removal efficiency in WWTPs nor lipophilicity were good predictors of CFs. Our findings suggest that use of most APIs can be estimated from traces measured in wastewater with a typical error of <2-fold. This provides support for the development of wastewater surveillance for estimating medicine use to fill existing data gaps, but also suggests limitations in detecting small changes.