Multi-biomarker approach for estimating population size in a national-scale wastewater-based epidemiology study

Barbara Kasprzyk-Hordern; Kishore Jagadeesan; Natalie Sims; Kata Farkas; Kathryn Proctor; John Bagnall; Megan Robertson; Davey L Jones; Matthew J Wade

doi:10.1016/j.watres.2024.122527

Multi-biomarker approach for estimating population size in a national-scale wastewater-based epidemiology study

Water Res. 2024 Oct 9;268(Pt A):122527. doi: 10.1016/j.watres.2024.122527. Online ahead of print.

Authors

Barbara Kasprzyk-Hordern¹, Kishore Jagadeesan², Natalie Sims², Kata Farkas³, Kathryn Proctor², John Bagnall⁴, Megan Robertson⁴, Davey L Jones³, Matthew J Wade⁵

Affiliations

¹ Department of Chemistry, University of Bath, Claverton Down, Bath BA2 7AY, United Kingdom. Electronic address: B.Kasprzyk-Hordern@bath.ac.uk.
² Department of Chemistry, University of Bath, Claverton Down, Bath BA2 7AY, United Kingdom.
³ School of Environmental and Natural Sciences, Bangor University, Gwynedd, Bangor LL57 2UW, United Kingdom.
⁴ Wessex Water, Claverton Down, Bath BA2 7AY, United Kingdom.
⁵ Data Analytics & Surveillance Group, UK Health Security Agency, London E14 4PU, United Kingdom.

PMID: 39405625
DOI: 10.1016/j.watres.2024.122527

Abstract

This study identifies biochemical markers (BCIs) that can be used as population markers in wastewater-based epidemiology (WBE) and compares their estimates with other established population size estimation (PE) methods, including census data (PE_CEN). Several groups of BCIs (64 targets: genetic and chemical markers) were investigated in an intercity study, including 10 cities/towns within England equating to a population of ∼7 million people. Several selection criteria were applied to identify the best BCIs to provide robust estimation of population size at a catchment level: (1) excellent performance with analytical methods; (2) excellent fit of the linear regression model which indicates PE-driven BCI daily loads; (3) low temporal variability in usage; (4) human-linked origin. Only a few tested BCIs showed a strong positive linear correlation between daily BCI loads and PE indicating their low spatiotemporal variability. These are: cimetidine, clarithromycin, metformin, cotinine, bezafibrate, metronidazole and hydroxymetronidazole, diclofenac, and benzophenone 1. However, only high/long term usage pharmaceuticals: cimetidine and metformin as well as cotinine (metabolite of nicotine) performed well when tested in two independent datasets and catchments accounting for both spatial and temporal scales. Strong seasonal usage trends were observed for antihistamines, NSAIDs (anti-inflammatories), antibiotics and UV filters, invalidating them as PE markers. Key conclusions from the study are: (1) Cimetidine is the best performing BCI; (2) Chemical markers outperform genetic markers as PE BCIs; (3) Water utility PE estimates (PE_WW) align well with PE_CEN and PE_BCI values; (4) Ammonium/orthophosphate as well as viral PE markers suffer from high temporal variability, hence, they are not recommended as PE_BCI markers, and, most importantly, (5) PE_BCI calibration/validation at the country/region level is advised in order to establish the best PE markers suited for local/national needs and accounting for site/region specific uncertainties.

Keywords: Chemical and viral biomarkers; Pharmaceuticals; Population size estimation (PE); Public health surveillance; Viruses; WBE; Wastewater-based epidemiology.