Challenges of surveying wastewater drug loads of small populations and generalizable aspects on optimizing monitoring design’


This scientific article was one of the 2015 EMCDDA scientific award winners, which celebrates scientific writing and distinguishes high-quality research in the field of illicit drugs


This abstract is provided here as a convenience only. Check the publisher's website (if available) for the definitive version.

Quantifying illicit drug loads through wastewater analysis (WWA) is an alternative approach to estimating population drug use. This study investigated the variability of daily drug loads in wastewater and their relationships to environmental factors over an extended period to: (i) explore the suitability of WWA in small populations and (ii) optimize the monitoring design for future studies.

Design, Setting, Participants
Daily wastewater samples (n = 1369 consecutive days) from a German village with approximately 7160 inhabitants.

Samples were analysed for cocaine and benzoylecgonine with liquid chromatography‐tandem mass spectrometry. Time‐series analysis was used to explore the effects of weather and other factors on daily cocaine loads. Subsampling was used to assess monitoring design.

Cocaine loads [mean = 652 mgCOC/day, standard deviation (SD) = 498 mgCOC/day] increased over the study period, with higher values during winter and spring. Despite high day‐to‐day variation, loads were significantly higher during weekends [+161 mgCOC/day, 95% confidence interval (CI) = 115–207 mgCOC/day, P < 10−4] and days with frost (+114 mgCOC/day, 95% CI = 6–223 mgCOC/day, P = 0.039) or snow (+150 mgCOC/day, 95% CI = 46–253 mgCOC/day, P = 0.005). Annual means estimated from 1‐week periods were subject to approximately 60% relative error. Increasing sample size and changing sampling from consecutive days to stratified random decreased this uncertainty.

Day‐to‐day variation and seasonality of drug loads from the few long‐term wastewater studies available to date suggest that up to 56 stratified random samples are required to obtain reliable (expected uncertainty around 10%) annual estimates of drug loads. Successfully assessing changes in consumption patterns or relationships to external factors requires larger sample sizes than estimating annual means, which holds true for high‐prevalence drugs in small communities and low‐prevalence drugs in big cities.