Andy Baker

Last week I received an e-mail from a former industry partner from around two decades ago, kindly alerting me to this paper: ‘Fluorescence characteristics of dissolved organic matter (DOM) in bottled drinking water of different countries: A potential risk to public health’. It was just published in Water Environment Research by researchers from Jahangirnagar University, Dhaka, Bangladesh and Southern Cross University, Australia. The paper is Open Access. You can read it at: https://onlinelibrary.wiley.com/doi/10.1002/wer.70064

So what is this ‘potential risk’? The authors have analysed some bottled water samples from countries in SE Asia and the Pacific, including Australia, and they suggest that most are at risk from faecal coliform contamination. To put it crudely, shit in the bottled water.

First off, this is very bad science and this is not true. Especially so for the Australian bottled waters analysed.

Why was I alerted? Well, the authors don’t actually measure coliforms in any water bottles. Instead, they use a relationship between water fluorescence and coliform numbers from a paper that did measure faecal coliforms. And that was one that I led with Sue Cumberland, John Bridgeman and our South African water industry collaborators, where we did measure coliform numbers and water fluorescence in a large number of sewage impacted river systems. (That paper: Baker et al 2015, To what extent can portable fluorescence spectroscopy be used in the real-time assessment of microbial water quality?https://www.sciencedirect.com/science/article/pii/S0048969715301728)

So what is the problem? The authors of the WEM paper analyses bottled water sample and quantified ‘peak T’ fluorescence using an unnamed fluorescence spectrometer and use a modelling method (PARAFAC) to report different fluorescence signatures in relative units. So far, so good as a standard approach, although the analytical detail is lacking.

The authors then convert that relative fluorescence to a parts per billion (ppb) tryptophan concnetration (their Table 2), but don’t say anything about how this was determined. I plotted the first two columns in Table 2, fluorescence in relative units and ppb, and they have a gradient of 1, so the authors have an equation from somewhere that relates fluorescence in relative units to a ppb equivalent. But where does this come from. It has to be incorrect.

The authors then they use this ppb value, and use the relationship between fluorescence (in ppb) and faecal coliform numbers published from South African contaminated rivers in 2015 (Baker et al., 2015, Science of the Total Environment). Again, I am that lead author. They suggest that bottled water has a medium to high risk of having faecal coliforms.

I suggest that there is no way that this is possible. The authors should demonstrate how they converted from relative fluorescence units to ppb concentrations for their instrumental set-up, and provide QC/QA checks by determining the colifrom concentrations in the bottled water samples.

A comment has been submitted to the community PubPeer review system. Your can read that here https://pubpeer.com/publications/3AFE6E22F2E35D99A66A467778EDDF