Results of an interlaboratory comparison of analytical methods for contaminants of emerging concern in water

An evaluation of existing analytical methods used to measure contaminants of emerging concern (CECs) was performed through an interlaboratory comparison involving 25 research and commercial laboratories. In total, 52 methods were used in the single-blind study to determine method accuracy and comparability for 22 target compounds, including pharmaceuticals, personal care products, and steroid hormones, all at ng/L levels in surface and drinking water. Method biases ranged from <10% to well over 100% in both matrixes, suggesting that while some methods are accurate, others can be considerably inaccurate. In addition, the number and degree of outliers identified suggest a high degree of variability may be present between methods currently in use. Three compounds, ciprofloxacin, 4-nonylphenol (NP), and 4-tert-octylphenol (OP), were especially difficult to measure accurately. While most compounds had overall false positive rates of ≤5%, bisphenol A, caffeine, NP, OP, and triclosan had false positive rates >15%. In addition, some methods reported false positives for 17β-estradiol and 17α-ethynylestradiol in unspiked drinking water and deionized water, respectively, at levels higher than published predicted no-effect concentrations for these compounds in the environment. False negative rates were also generally <5%; however, rates were higher for the steroid hormones and some of the more challenging compounds, such as ciprofloxacin. The elevated false positive/negative rates of some analytes emphasize the susceptibility of many current methods to blank contamination, misinterpretation of background interferences, and/or inappropriate setting of detection/quantification levels for analysis at low ng/L levels. The results of both comparisons were collectively assessed to identify parameters that resulted in the best overall method performance. Liquid chromatography-tandem mass spectrometry coupled with the calibration technique of isotope dilution were able to accurately quantify most compounds with an average bias of <10% for both matrixes. These findings suggest that this method of analysis is suitable at environmentally relevant levels for most of the compounds studied. This work underscores the need for robust, standardized analytical methods for CECs to improve data quality, increase comparability between studies, and help reduce false positive and false negative rates.