Screening of human antibiotic substances and determination of weekly mass flows in five sewage treatment plants in Sweden

Twelve antibiotic substances for human use, including trimethoprim and representatives of the fluoroquinolone (FQ), sulfonamide (SA), penicillin (PE), cephalosporin (CE), nitroimidazole (NI), tetracycline (TC), and macrolide (MA) groups, were subjected to a screening study at five Swedish sewage treatment plants (STPs) during one week in 2002 and one week in 2003. The analytes were extracted from raw sewage water, final effluent, and sludge by solid-phase extraction (SPE) or liquid-solid extraction (as appropriate) and then identified and quantified by liquid chromatography/tandem mass spectrometry. The mostfrequently detected antibiotics in the matrices considered in this study were norfloxacin, ofloxacin, ciprofloxacin, trimethoprim, sulfamethoxazole, and doxycycline. The other analytes were only detected in a few samples. Analysis of the weekly mass flows through each STP showed that FQs were partly eliminated from the water during sewage water treatment and the highest amounts of these substances were found in sludge. Sulfamethoxazole and trimethoprim were mainly found in raw sewage water and final effluent, but these substances had balancing mass flows, indicating that they too can withstand sewage water treatment. The mass flow patterns for doxycycline were more complex, with high amounts occurring in sludge in some cases, suggesting thatthe behavior of this analyte may be more strongly influenced by the treatment process and other variables at individual STPs. The environmental load (the sum of the amounts in the final effluent and sludge) normalized to the number of inhabitants in the catchment area of each investigated STP compared with theoretical predictions based on consumption data (in parentheses) showed good correlations: norfloxacin, 0.8 (0.9); ofloxacin, 0.3 (0.2); ciprofloxacin, 1.3 (3.5); sulfamethoxazole, 0.2 (0.4); trimethoprim, 1.1 (1.0); and doxycycline, 0.7 (0.4) mg per person per week. The results show that reasonably accurate predictions of environmental load of these antibiotics can be time-effectively derived from consumption data without additional measurements.