Urban wastewater treatment plants as hotspots for the release of antibiotics in the environment: A review

Urban wastewater treatment plants (UWTPs) are among the main sources of antibiotics’ release into various compartments of the environment worldwide. The aim of the present paper is to critically review the fate and removal of various antibiotics in wastewater treatment, focusing on different processes (i.e. biological processes, advanced treatment technologies and disinfection) in view of the current concerns related to the induction of toxic effects in aquatic and terrestrial organisms, and the occurrence of antibiotics that may promote the selection of antibiotic resistance genes and bacteria, as reported in the literature. Where available, estimations of the removal of antibiotics are provided along with the main treatment steps. The removal efficiency during wastewater treatment processes varies and is mainly dependent on a combination of antibiotics’ physicochemical properties and the operating conditions of the treatment systems. As a result, the application of alternative techniques including membrane processes, activated carbon adsorption, advanced oxidation processes (AOPs), and combinations of them, which may lead to higher removals, may be necessary before the final disposal of the effluents or their reuse for irrigation or groundwater recharge. a 2012 Elsevier Ltd. All rights reserved.


Introduction
During the last years, it is recognized that antibiotics constitute a new class of water contaminants of emerging concern with adverse effects on the aquatic life (Kolpin et al., 2002;Kü mmerer, 2009;Fatta-Kassinos et al., 2011a). The generic term "antibiotic" is used herein to denote any class of organic molecule that inhibits or kills microbes by specific interactions with bacterial targets, without any consideration of the source of the particular compound or class (Davies and Davies, 2010). Investigations for the occurrence of various antibiotics in wastewater effluents have been conducted in several European countries (Jones et al., 2001;Heberer, 2002;Miao et al., 2004;Batt et al., 2007;Gulkowska et al., 2008;Kü mmerer, 2009;Fatta-Kassinos et al., 2011a). Because of the intensive use of antibiotics for human (domestic and hospital use), veterinary and agriculture purposes, these compounds are continuously released into the environment from anthropogenic sources, such as urban wastewater treatment plants (UWTPs), which are considered as one of the main 'hotspots' of potential evolution and spreading of antibiotic resistance into the environment (Hirsch et al., 1999;Diaz-Cruz et al., 2003;Brown et al., 2006;Kü mmerer, 2009;Czekalski et al., 2012;Le Corre et al., 2012). The presence of antibiotics in environmentally relevant concentration levels has been associated to chronic toxicity and the prevalence of resistance to antibiotics in bacterial species (Schwartz et al., 2006;Kü mmerer, 2009). The number of studies focusing on the chronic toxicological assessment of antibiotics in the environment is constantly increasing with the aim to bridge the various knowledge gaps (i.e. relevant endpoints to be considered in chronic bioassays) associated with these issues. Boxall (2004) and Kü mmerer (2009) represent two comprehensive review articles regarding the ecotoxicity of antibiotics. Thomulka and McGee (1993) determined for example the toxicity of a number of antibiotics (e.g. novobiocin, tetracycline, chloramphenicol, nalidixic acid, ampicillin, streptomycin) on Vibrio harveyi in two bioassay methods. Almost no toxic effects were found after short incubation times when luminescence was used as an endpoint. However, in a long-term assay using reproduction as the endpoint, a toxic effect in environmentally relevant concentrations was detected for almost all the examined antibiotics. These results are in accordance with the observations of Froehner et al. (2000) concerning chloramphenicol, nalidixic acid and streptomycin. The chronic toxicity of several groups of antibiotics toward Vibrio fischeri is also presented in a study by Backhaus and Grimme (1999). The chronic bioluminescence inhibition assay was shown to be sensitive against many of the high volume antibiotics used for veterinary purposes and in aquaculture. Furthermore, exposure to antibiotics may have adverse effects on the reproductive system in the early life stages of different organisms like the freshwater flea Daphnia magna and the crustacean Artemia salina (Macrì et al., 1988;Wollenberger et al., 2000). In the study by Kim et al. (2007), sulfonamides (i.e. sulfamethoxazole, sulfachloropyridazine, sulfathiazole, sulfamethazine, sulfadimethoxine), and trimethoprim, were examined for their acute aquatic toxicity by employing a marine bacterium (V. fischeri), a freshwater flea (D. magna) and the Japanese medaka fish (Oryzias latipes). In this study, D. magna was in general the most susceptible in terms of effective/lethal concentrations-E/ LC 50 , among the test organisms.
Moreover, the extensive use of antibiotics has contributed to the development of antibiotic resistance genes and bacteria, reducing the therapeutic potential against human and animal pathogens (Kemper, 2008). The consequences are particularly worrying as bacteria in the aquatic environment can be continually exposed to antibiotic residues (Rosal et al., 2010). The biological treatment process creates an environment potentially suitable for resistance development and spreading, because bacteria are continuously exposed to w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 9 5 7 e9 9 5 environmentally relevant levels of antibiotics. However, it remains unclear where most of the resistant bacteria have been selected, and in particular if the low antibiotic concentrations that are present in natural environments or in human/animal body compartments during therapeutic use, are important for the selection and enrichment of resistant mutants (Gullberg et al., 2012). The extent to which human activities contribute to the development of resistant bacterial strains is still poorly understood (Auerbach et al., 2007). The number of studies, focusing exclusively on wastewater treatment systems regarding the removal of antibiotic resistance, is still however limited. Gao et al. (2012) investigated the relationship between concentrations of tetracyclines and sulfonamides and the number of antibiotic resistance genes and antibiotic resistant bacteria in a conventional UWTP located in Michigan. Significant reductions (2e3 logs) of antibiotic resistance genes and antibiotic resistant bacteria were observed between raw influent and final effluent whereas no apparent decrease was observed in the concentrations of tetracycline resistance genes (tetO and tetW ) and sulfonamide resistance gene (sulI ) by chlorine disinfection. Moreover, Dodd (2012) provide a comprehensive overview on the significance of antibiotic resistant genes (ARG) and bacteria occurrence in environmental systems, and a discussion on the role that commonly used water and wastewater disinfection processes may play in minimizing ARG transport and dissemination. Zhang et al. (2009) reported the impact of the wastewater treatment process on the prevalence of antibiotic resistance in Acinetobacter spp. in the wastewater and the possible spread of antibiotic resistance to receiving water bodies. It was found that the prevalence of antibiotic resistance was significantly higher in the downstream samples than in the upstream samples, with the higher values occurred for trimethoprim (97%), followed by rifampin (74%). Other studies have reported that the prevalence of resistant bacteria in sewage may significantly vary, depending on the plant (initial quality characteristics of sewage, type of treatment, plant operation, etc.), the target bacterial population, and the antimicrobial agent under study, as well as on the methods and the breakpoint values used to determine antimicrobial resistance (Guardabassi et al., 2002).
Another issue related to the use of reclaimed wastewater for irrigation is the plant uptake of antibiotics. The accumulation may or may not affect the growth and development of plants; however, the uptake into plants may represent an important exposure pathway of these compounds to humans and other biota (European Medicines Agency-EMEA). Migliore et al. (2003) determined the phytotoxicity of enrofloxacin on crop plants Cucumis sativus, Lactuca sativa, Phaseolus vulgaris and Raphanus sativus in a laboratory model. Between 50 and 5000 mg L À1 , enrofloxacin induced hormetic effect in plants, with a dose-dependant stimulation or toxicity on the length of primary root, hypocotyl, cotyledons and the number/length of leaves. There are also new concerns that antibiotics decrease the biodegradation of leaf and other plant materials, which serves as the primary food source for aquatic life in rivers and streams (Richardson and Ternes, 2011).
The aim of the present paper is to introduce a critical review on the removal efficiency of various antibiotics in wastewater treatment during the application of different processes, namely biological processes, advanced treatment technologies and disinfection. An effort to include as many studies as possible was made in order to highlight important findings and present the knowledge currently available on the removal efficiency of antibiotics from wastewater through a variety of treatment processes.

Fate of antibiotics in UWTPs
The conventional wastewater treatment generally consists of a primary, secondary and sometimes a tertiary stage, with different biological and physicochemical processes available for each stage of the treatment. Primary treatment intends to reduce the solid content of the wastewater (oils and fats, grease, sand, grit and settleable solids). This step is performed entirely mechanically by means of filtration and sedimentation and is common at all UWTPs. However, the secondary treatment, which typically relies on a biological process to remove organic matter and/or nutrients with aerobic or anaerobic systems, can differ substantially. Several biological treatments are being used in modern municipal UWTPs, but the most common method is conventional activated sludge (CAS). Membrane bioreactors (MBR), moving bed biofilm reactor (MBBR), or fixed bed bioreactors (FBR) are less common. Activated sludge plants use dissolved oxygen to promote the growth of a biological floc that substantially removes the organic material and nitrogen at given conditions. In the final step, tertiary wastewater treatment processes can be applied to remove phosphorus by precipitation and particles on a filter (Batt et al., 2007). In some UWTPs the effluent is also disinfected before it is released into the environment, typically by chlorination or ultraviolet irradiation. The effect of biological treatments, membrane filtration, activated carbon adsorption, advanced oxidation processes (AOPs), and disinfection on different classes of antibiotics has been widely investigated in the last years; several of these studies are presented in the subsequent paragraphs.

Effect of biological treatment on antibiotics' removal
Elimination and transformation of antibiotics during the biological treatment is the result of different processes. These processes can be biotic (biodegradation, mainly by bacteria and fungi) and non-biotic or abiotic (e.g. sorption, hydrolysis, photolysis). The removal of antibiotics mainly depends on their sorption on the sewage sludge and their degradation or transformation during the treatment. Hydrolysis can play a role for some compounds, while photolysis is not very likely to occur due to the low exposure of the substances to light during the wastewater treatment.
Hydrophobic (or non-polar) antibiotic residues are expected to occur at higher concentration in primary and secondary sludge than hydrophilic ones because they have a greater affinity to solids and hence, concentrate in the organic-rich sewage sludge (Le-Minh et al., 2010). Antibiotics can also be removed from aqueous solutions onto solid w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 9 5 7 e9 9 5 particulates by ion exchange, complex formation with metal ions and polar hydrophilic interactions (Diaz-Cruz et al., 2003). Antibiotics that are sorbed to flocs, suspended solids and activated sludge, are removed from the aqueous phase by sedimentation and subsequent disposal of excess sludge. The affinity of antibiotics sorbed to sludge is most often represented by sludge sorption constants K d (L kg À1 ). The higher K d values the higher sorption of the compounds to sludge. A review on K d values of several antibiotics is provided in Kovalova et al. (2012). It is important to note that the sludge is often used as fertilizer on agriculture fields, but in several European countries this is forbidden and the sludge is incinerated. Using sludge as fertilizer can therefore be considered as another input pathway for various antibiotics into the environment.
The tendency to accumulate in sludge solids can be assessed using the octanolewater partition coefficient (K OW ). Rogers (1996) proposed the following guide to assess the sorption potential of organic contaminants: logK OW < 2.5: low sorption potential (e.g. tetracyclines, sulfonamides, aminoglycosides); 2.5 < logK OW < 4.0 (e.g. b-lactams, macrolides): medium sorption potential and logK OW > 4.0 (e.g. glycopeptides): high sorption potential. However, it should be emphasized that the prediction of the antibiotics sorption onto solids or sludge is mainly possible for non-polar compounds, while the prediction of the behavior of polar or charged compounds is often not correct. In some cases, the use of logK OW values lead to an underestimation of the sorption of e.g. fluoroquinolones (Golet et al., 2003) or tetracyclines  to sludge. For instance, ciprofloxacin (fluoroquinolone) has a K OW value of 1.8, but nevertheless sorbs onto sludge by 80%, indicating that sorption is the main elimination process.
However, antibiotics are mostly hydrophilic and were designed to be biologically resistant; they are therefore expected to mainly remain in the aqueous phase of the wastewater.
The main operational factors that can influence the biological removal of antibiotic residues in wastewater treatment are biochemical oxygen demand (BOD 5 ), existence and size of anoxic and anaerobic compartments, suspended solids (SS) loading, hydraulic retention time (HRT), sludge retention time (SRT), foodemicroorganism ratio (F/M ratio), mixed liquorsuspended solids (MLSS), pH and temperature (Drewes, 2008;Kovalova et al., 2012).
The SRT is related to the growth rate of microorganisms. High SRTs allow the enrichment of slowly growing bacteria and therefore, provide greater diversity of enzymes, some of which are capable of degrading the antibiotic compounds (Jones et al., 2007;Le-Minh et al., 2010). High SRT can be reached with a membrane bioreactor (MBR), where the suspended activated sludge is retained in the reactor by utilizing a membrane for solid/liquid separation instead of a settling tank as used in CAS. Commonly, micro-or ultrafiltration membranes are used in MBRs, which do not retain the antibiotics on the filter. Some studies have been performed to investigate if higher SRTs enhance the elimination of antibiotics, which will be discussed in detail below Gö bel et al., 2007;Radjenovic et al., 2009b;Tadkaew et al., 2011;Kovalova et al., 2012).
The performance (expressed as % removal) of some UWTPs applying biological treatment for removing antibiotics as reported in the literature is summarized in Table 1. The removal is highly variable for many substances (from nearly complete to very little). Frequently, however, operational details are not provided in the studies available in the literature on the fate and transport of antibiotic residues during wastewater treatment or have not been systematically investigated. This poses a major challenge for the comparison and discussion of results. Moreover, differences in reported efficiencies may, in some cases, be attributed to limitations of employed mass balance techniques (Le-Minh et al., 2010). For example, short-term variations of pharmaceuticals loads in influent can be significant Khan and Ongerth, 2005), thus consideration must be taken when comparing influent and effluent concentrations.
Antibiotics can be grouped by either their chemical structure or mechanism of action. The main groups of antibiotics and their potential removal during conventional wastewater treatment are discussed in the following sections.

b-Lactams
b-lactams are not very stable due to hydrolysis of the betalactam ring (Hirsch et al., 1999;Lä ngin et al., 2009). b-lactams have been reported to be significantly reduced during biological treatment with removals higher than 90% (Watkinson et al., 2007(Watkinson et al., , 2009). According to Li et al. (2009) the observed removals at an UWTP in Hong Kong were between 30.4 and 100%. b-lactams were also eliminated significantly at both Shatin and Stanley UWTPs as described in the work of Li and Zhang (2011). Cha et al. (2006) investigated the fate of four blactams (ampicillin, cloxacillin, cephapirin, oxacillin) and the estimated removals were between 17 and 43%. Ampicillin was removed by 82% in an activated sludge process (Li and Zhang, 2011). High removal of ampicillin (>94%) was also achieved in MBR treatment (SRT 3e60 days, Xia et al., 2012). A significant removal (96%) of cephalexin from 2000 ng L À1 to 78.2 ng L À1 has been reported to occur through conventional UWTP processes in Australia (Costanzo et al., 2005). Analysis of amoxicillin conducted by Zuccato et al. (2010) in UWTPs in Italy and Switzerland showed that it is efficiently removed by CAS (100%). Similarly, Watkinson et al. (2009) showed that amoxicillin is quite susceptible to microbial degradation with removal higher than 99% and therefore it is not likely to remain in significant concentration after biological treatment systems. Cephalexin was removed by 53% at the Shatin UWTP, while it was removed by 91% at the Stanley UWTP (Li and Zhang, 2011). Cephalexin was also removed by 36e99.8% in four Taiwanese UWTPs combining biological treatment and disinfection process (UV or chlorination) (Lin et al., 2009a,b) and by 99.6% in an Australian UWTP using CAS (Watkinson et al., 2009). Therefore, cephalexin is relatively easily eliminated in UWTPs with biological processes, whereas cefotaxime, which was only detected in Shatin UWTP, was removed by only 43% (Li and Zhang, 2011). Li and Zhang (2011) reported that roxithromycin was degraded by 40e46% during CAS. Slightly lower removal (33%) was reported for one German UWTP . In the  Xu et al., 2007 (continued on next page) w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 9 5 7 e9 9 5 w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 9 5 7 e9 9 5 (continued on next page) w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 9 5 7 e9 9 5 (continued on next page) w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 9 5 7 e9 9 5 w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 9 5 7 e9 9 5 (continued on next page) w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 9 5 7 e9 9 5 w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 9 5 7 e9 9 5 (continued on next page) w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 9 5 7 e9 9 5 studies of Gö bel et al. (2007) and Joss et al. (2005), roxithromycin was removed at two UWTPs in Switzerland by 38% during secondary treatment and by 38e57% during MBR treatment (SRT ¼ 16, 33, 60e80 days). Moreover, roxithromycin removal was reported to be higher than 53% for four UWTPs in south China . Clara et al. (2005) reported a removal range for roxithromycin of 52e100% during MBR treatment (SRT ¼ 10e55 days). Erythromycin is frequently detected as its main human metabolite, the dehydrated product with an apparent loss of one molecule of water, erythromycin-H 2 O. Erythromycin-H 2 O was degraded by 15% and 26% in activated sludge processes at Shatin and Stanley UWTP, respectively (Li and Zhang, 2011), and up to 10% in two Swiss UWTPs (Gö bel et al., 2007). Higher removals were reported in other studies, that is, 56% in four Taiwanese UWTPs (Lin et al., 2009a,b) and 43.8e100% in an UWTP in USA (Karthikeyan and Meyer, 2006) by secondary wastewater treatment processes both employing activated sludge.
Macrolides may be sorbed to biomass via cation exchange processes due to the fact that under typical wastewater conditions (pH ¼ 7e8), many are positively charged through the protonation of the basic dimethylamino group ( pK a ¼ 7.1e9.2) while the surface of activated sludge is predominantly negatively charged (Le-Minh et al., 2010). Analysis of sludge, however, showed that sorption of macrolides is of minor importance for the elimination in conventional UWTPs with K d of below 400 L kg À1 Kovalova et al., 2012). Abegglen et al. (2009) observed a slightly higher affinity of MBR sludge to macrolides than conventional activated sludge (K d ¼ 1400 L kg À1 for azithromycin).

Sulfonamides
The concentrations of these antibiotics in UWTP influents and effluents vary significantly, depending on consumption patterns and the types of wastewater treatment processes employed. For example, sulfamethoxazole has been reported at concentrations as high as 5450e7910 ng L À1 in sewage influent in China and was completely removed during the treatment (Peng et al., 2006). In a Taiwanese UWTP, sulfamethoxazole was detected in influent at concentration range of 500e10,000 ng L À1 and the removal was 65e96% after the biological treatment ). Sulfamethoxazole has been reported to be removed up to 81% (initial concentration 1090 ng L À1 ) , 69e75% (initial concentration in the range 13e155 ng L À1 ) (Pailler et al., 2009), 68.2e95.7% (initial concentration in the range 146e355 ng L À1 ) ) and 93% (initial concentration in the range 3000 ng L À1 ) (Watkinson et al., 2009). However, in other studies lower removal rates of 20e24% were reported (Brown et al., 2006;. At this point it is worth mentioning that, there is only little knowledge on the environmental fate of humans' metabolites of antibiotics, which are excreted from the human body, often in considerable amounts and can be found predominantly in w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 9 5 7 e9 9 5 the environment (Hollender et al., 2008). Humans' metabolites are often omitted when analyzing antibiotics; a notable exception is the sulfamethoxazole's acetylated metabolite. N 4 -acetylsulfamethoxazole usually accounts for more than 50% of an administered dose in human excretion and can occur in UWTP influents at concentrations of 2.5e3.5 times higher than concentrations of the parent compound (Gö bel et al., 2007). Significant removal efficiencies (81e96% and 68e92%, respectively) of N 4 -acetylsulfamethoxazole during secondary treatment were reported by Gö bel et al. (2007) and Joss et al. (2005). N 4 -acetylsulfamethoxazole can also deconjugate into sulfamethoxazole during wastewater treatment (Gö bel et al., 2007), leading to an underestimation of removal efficiency for sulfamethoxazole if this metabolite is not considered. This might be a reason for the highly varying observed elimination rates. Higher removal rates were observed for sulfadiazine during activated sludge process at Shatin (72.8%, 100%) and Stanley (87%) UWTPs Li and Zhang, 2011). However, the removal rate for sulfadiazine was only 50% in a Chinese UWTP .
The variation of sulfonamides removal may possibly be explained not only by the deconjugation of metabolites, but also by the differences in UWTP operating conditions such as HRT and the presence of an anaerobic compartment. Higher SRT, though, was not found to increase the elimination of sulfamethoxazole and sulfapyridine (Gö bel et al., 2007;Radjenovic et al., 2009b).

Trimethoprim
The presence of trimethoprim can generally be correlated to that of sulfamethoxazole since the two drugs are often administered in combination . The removal of trimethoprim has been reported as 13% and 42% by Li and Zhang (2011). The removal of this compound was found to fluctuate within the same levels in various UWTPs in USA (50e100%), in Germany (69%) and in Taiwan (74%) (Brown et al., 2006;Karthikeyan and Meyer, 2006;Yu et al., 2009). Higher removals were obtained in five UWTPs in Australia yielding 94% (Watkinson et al., 2009) and 93.3% . In contrast, the removal of trimethoprim was negligible as reported in the studies of Lindberg et al. (2005) and Roberts and Thomas (2006).
Some studies have indicated that nitrifying microorganisms appear to be capable of degrading trimethoprim. This suggests an important role for aerobic conditions for the biotransformation of trimethoprim (Perez et al., 2005;Batt et al., 2006). Moreover, trimethoprim elimination was found to be increased at higher SRTs (Gö bel et al., 2007;Radjenovic et al., 2009b;Tambosi et al., 2010;Kovalova et al., 2012).

Quinolones
Removal efficiencies of quinolones during wastewater treatment in Sweden were reported to be 87% for norfloxacin and ciprofloxacin and 86% for ofloxacin . A later study reported the removal of ciprofloxacin (>90%), ofloxacin (56%), and norfloxacin (>70%) during activated sludge treatment followed by chemical coagulation/flocculation ). Sorption to sewage sludge has been suggested by Golet et al. (2003) as the primary removal mechanism for fluoroquinolones (ciprofloxacin and norfloxacin) during secondary wastewater treatment, resulting in the removal of 78e84% of the aforementioned fluoroquinolones from the aqueous phase. High removals of ofloxacin were achieved in UWTPs in Cyprus (>83%)  and in China (100%) (Peng et al., 2006). Removal of ciprofloxacin in an MBR treating hospital wastewater (SRT ¼ 30e50 days) was only 51% (Kovalova et al., 2012). This relatively low removal might have been caused by the lower sludge production in MBR than in conventional activated sludge, leading to lower sorption.

Tetracyclines
Tetracycline is one of the most frequently detected antibiotics in wastewater (Watkinson et al., 2007). According to the study of Yang et al. (2005) tetracycline was removed by 85% in an UWTP in Colorado. Li and Zhang (2011) reported removals of 24e36% at two plants while higher removals (67.9e100%) were reported by Karthikeyan and Meyer (2006) and four Taiwanese UWTPs (66e90%) by Lin et al. (2009a,b) The removal rates for chlortetracycline as reported by Li and Zhang (2011) were in the range of 82% and 85%. Furthermore, for chlortetracycline and doxycycline, after secondary treatment and chlorination, the removal efficiencies were reported to be 78% and 67%, respectively . Choi et al. (2007) reported even higher removal values for minocycline and democlocycline (92 and 89%, respectively). High removal was also achieved for tetracyclines in MBR treatment (SRT ¼ 3e60 days, Xia et al., 2012).
Tetracyclines have complexing properties and can easily bind to calcium and similar ions, thus forming stable complexes, which can bind to suspended matter or sewage sludge (Drewes, 2008). Kim et al. (2005) found no evidence of tetracycline biodegradation during the biodegradability test, but sorption was found to be the principal removal mechanism in activated sludge. These properties might explain why tetracyclines are detected in many cases in low concentration levels (ng L À1 ) in treated secondary effluents.

Other antibiotic groups
Several studies reported the occurrence of lincosamides antibiotics such as lincomycin and clindamycin in wastewater influents and effluents with maximum removal efficiencies of 67% Kovalova et al., 2012). Clindamycin may be transformed back from the main human metabolite w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 9 5 7 e9 9 5 clindamycin sulfoxide in the denitrification process, resulting in increased concentration (Kovalova et al., 2012). A study by Watkinson et al. (2009) showed that removals of polyether ionophores (monensin and salinomycin) in wastewater were up to 95%. Metronidazole, an imidazole antibiotic, was removed up to 23% during CAS (Kasprzyk-Hordern et al., 2009;Jelic et al., 2011) and 45% in an MBR treating hospital wastewater (SRT ¼ 30e50 days, Kovalova et al., 2012). Metronidazole is rapidly transformed into 1-(2-hydroxyethyl)-2-hydroxymethyl-5-nitroimidazole (Mahugo-Santana et al., 2010). Limited information on the behavior of polyether ionophores through UWTP processes is available, due to the less likely occurrence of these antibiotics in urban wastewater except where there is runoff from agricultural lands into sewers. Glycopeptides such as vancomycin was analyzed by Li and Zhang (2011) and the removal after the activated sludge process was found to be as high as 52%. The aminoglycoside gentamicin was found in hospital wastewater, although is a compound that is adsorbed very strongly (Loffler and Ternes, 2003).
In summary, biological treatment cannot completely remove antibiotics in wastewater treatment. Accordingly, alternative treatment processes are considered as necessary in order to provide further elimination of these compounds from wastewater effluents and to better manage environmental and human exposure to these contaminants.
In the following sections, other techniques including membrane filtration, activated carbon adsorption and advanced oxidation processes (AOPs) are discussed. The removal of antibiotics by these processes is depicted in Table 2 along with other relevant and important information. The upgrading of UWTPs and the application of such technologies is regarded as a possible optimization of the biological treatment with regard to antibiotics' removal.

Membrane processes
Removal of antibiotics in membrane processes can occur through multiple mechanisms. First, removal can be governed by adsorption where antibiotics that are hydrophobic or have strong hydrogen-bonding characteristics, readily adsorb to membranes at the initial stages of filtration. In many cases though, removal can occur through steady-state rejection due to either steric effects for uncharged solutes or combined steric and electrostatic effects for charged solutes. These mechanisms are dependent on the physicochemical properties of the compound (molecular weight cut-off (MWCO), pK a , hydrophobicity/hydrophilicity), the solution (pH, ionic strength), and the membrane characteristics (material, surface morphology, pore size) (Le-Minh et al., 2010). While the pores in micro-and ultrafiltration are too large to reject micropollutants, the lower membrane pore size used in nanofiltration (NF, pore size range: 0.001 mm) and reverse osmosis (RO, pore size range <0.001 mm) have been shown in recent years to effectively remove low-molecular-weight pharmaceutical compounds, including antibiotics, during wastewater treatment. Various studies showed up to 90% removal of several antibiotics including quinolones, sulfonamides, tetracyclines and trimethoprim (Kimura et al., 2004;Morse and Jackson, 2004). A study undertaken by Kosutíc et al. (2007) on the treatment of model wastewater of a manufacturing plant producing pharmaceuticals for veterinary use showed that sulfonamides were effectively removed by NF and RO. Zhang et al. (2006) reported a high removal efficiency (98.5e99.7%) for amoxicillin from wastewater, which contains high level of TOC using RO. In a study of Li et al. (2004) oxytetracycline at very high concentration (1000 mg L À1 ) in wastewater from pharmaceutical manufacturing was reduced to 80 mg L À1 (<92% removal).
Given the complementary treatment capacity of MBR and NF/RO membrane filtration, there is significant scope for the coupling of these two treatment processes to achieve an overall enhanced performance (Alturki et al., 2010;Dolar et al., 2012). Excellent overall removal of target antibiotics with removal rates above 99% was achieved with MBR/RO (Dolar et al., 2012).
Some investigations reveal that the fouling of membranes can also lead to improved rejection of many solutes (Schafer et al., 1994;Drewes et al., 2006;Xu et al., 2006). This interesting observation is believed to be due to increased negative surface charge leading to increased electrostatic rejection of ionic species; along with simultaneously increased adsorptive capacity for non-ionic solutes ).

Activated carbon adsorption treatment
Adsorptive treatment with activated carbon can be used for removing many hydrophobic and also some charged pharmaceuticals from water (Le-Minh et al., 2010). The adsorption mainly involves the following steps: (i) solute transport in the bulk-adsorbate movement by the stagnant liquid film surrounding the adsorbent, (ii) film diffusioneadsorbate transport along the film, (iii) pores diffusioneadsorbate diffusion through the porous structure to the active sites (molecular diffusion in the pore and/or in the adsorbent surface), (iv) adsorption-interaction between adsorbate and porous structure (Homem and Santos, 2011). The removal effectiveness of the activated carbon adsorptive treatment system depends on the properties of the adsorbent (e.g. specific surface area, porosity, surface polarity and physical shape of the material), and the characteristics of the compound (e.g. shape, size, charge and hydrophobicity). Moreover, the sorption efficiencies of antibiotics to activated carbon may be significantly altered by the initial concentrations of the target compounds, the pH, the temperature and the presence of other species in the solution (Aksu and Tunç, 2005). Non-specific dispersive interactions (e.g. van der Waals interactions) are the dominant mechanism of removal for organic compounds, including antibiotics, in activated carbon adsorption systems, removing most non-polar antibiotics with logK OW > 2. However, electrostatic interactions between ionic antibiotics and the charged groups on the surface of activated carbon can result in removal of polar antibiotics . The removal of antibiotics by activated carbon has been reported during wastewater treatment in some studies (Adams et al., 2002;Westerhoff et al., 2005;Putra et al., 2009;Rivera-Utrilla et al., 2009;McArdell et al., 2011;Boehler et al., 2012). A post-treatment with powdered activated carbon (PAC) after biological treatment has been mostly investigated. The concentrations of several antibiotics in wastewater with PAC dosages between 10 and 20 mg L À1 have been reduced by 49e99% after 4 h contact time (Adams et al., w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 9 5 7 e9 9 5 w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 9 5 7 e9 9 5   w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 9 5 7 e9 9 5 w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 9 5 7 e9 9 5 w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 9 5 7 e9 9 5 [H2O2] ¼ 250 mg L À1 (57 % of COD); (53% of DOC) CAS effluent (Salerno, Italy) 10 mg L À1 Batch experiments (300 mL), 125W black light fluorescent lamp (300-420 nm; photon flux ¼ 4.7 Â 10 À7 einstein s À1 ) w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 9 5 7 e9 9 5 w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 9 5 7 e9 9 5 2002; Westerhoff et al., 2005). In a study on hospital wastewater treatment, macrolides, fluoroquinolones, trimethoprim and clindamycin were removed well at PAC dosages of 20e40 mg L À1 , while sulfamethoxazole and metronidazole showed poor removals . PAC can also be added directly into the biological reactor, where higher concentrations of carbon are required. Serrano et al. (2011) added 1 g L À1 into a sequential membrane bioreactor and found elimination of 42e64% for erythromycin, 71e97% for roxitromycin whereas no significant removal was obtained for TMP. Putra et al. (2009) compared the adsorption capacity of activated carbon and bentonite and reported that 94.67% of amoxicillin was removed from wastewater using activated carbon at a dose as high as 30 g L À1 .
It should be noted that in the case of the application of the activated carbon adsorption process in wastewater effluents, the natural dissolved organic matter (DOM) in wastewater matrix competes for adsorption sites and decreases the activated carbon capacity for antibiotics and other micropollutants .

Advanced oxidation processes (AOPs)
Advanced Oxidation Processes (AOPs) are quite efficient novel methods for water and wastewater treatment (Legrini et al., 1993;Klavarioti et al., 2009;Malato et al., 2009). These processes involve the use and generation of powerful transitory species, principally the hydroxyl radical (HO ) (Goslich et al., 1997;Andreozzi et al., 1999). HO are powerful oxidizing agents leading to oxidation and mineralization of organic matter (Litter, 2005), while this species is characterized by lack of selectivity of attack. This property is of great importance in wastewater treatment because radicals attack the oxidizable part of organic molecules with rates usually in the order of 10 6 e10 9 M À1 s À1 (Andreozzi et al., 1999). Several studies have reported the effective AOPs treatment for removal of antibiotics in wastewater effluents (Adams et al., 2002;Arslan Alaton et al., 2004;Saritha et al., 2007;Naddeo et al., 2009;Elmolla and Chaudhuri, 2011). It is worth noting the fact that most studies do not include information on the by-products formed during the application of oxidation or any information related to the antibiotic activity of the byproducts. Therefore, AOPs should be carefully monitored and ecotoxicological investigations should be accompanied to investigate the formation of potentially toxic transformation products (Hollender et al., 2009;Rizzo, 2011). The effectiveness of oxidative processes for degrading antibiotics will be largely determined by the specific water matrix. However, the effects of water matrix quality on antibiotics removal are much less well understood than for other technologies. For example, the presence of natural dissolved organic matter (DOM) can result in the formation of oxidation by-products that may cause water quality to deteriorate beyond its initial state of contamination. Similarly, the presence of nitrates, carbonates and DOM, can interfere with the destruction of the target antibiotic(s) and ultimately reduce the effectiveness of the selected AOP. w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 9 5 7 e9 9 5 most common AOPs that have been used and evaluated (mainly at a bench scale but many of the processes are being developed at a pilot-scale as well) are: photolysis under ultraviolet (UV) irradiation; combinations of hydrogen peroxide (H 2 O 2 ), ozone (O 3 ) and UV irradiation; homogeneous photocatalysis with Fenton reagent, heterogeneous photocatalysis with semiconductor materials (e.g. TiO 2 ) and sonolysis under ultrasound irradiation.

Ozonation
Ozone is a powerful oxidant and has been increasingly used for the treatment of wastewater whereas it has been traditionally employed in drinking water treatment (Litter, 2005). Huber et al. (2005) and Hollender et al. (2009) observed that using ozone at a dose of 2 mg L À1 (0.3e0.4 g g À1 DOC) more than 80% of sulfonamides, trimethoprim and macrolides were removed in the effluent of secondary wastewater treatment. Similar results between different wastewater treatment plants are achieved if the dose of ozone per amount of dissolved organic carbon (DOC) is compared. The study by Adams et al. (2002) showed that ozonation removed more than 95% of several sulfonamides and trimethoprim from river water within 1.3 min contact time at ozone dose of 7.1 mg L À1 . Clindamycin was already removed by 95% with an ozone dose of 2 mg L À1 (0.40 g O 3 g À1 DOC) (Hollender et al., 2009) and tetracycline by 100% with an ozone dose of 1.5 mg L À1 (Huber et al., 2005). Balcıoglu and Otker (2003) found that up to 80% of b-lactams removal from wastewater was observed during ozonation treatment after 60 min and ozone dose 2.96 g L À1 h À1 . In a study of Arslan Alaton et al. (2004) the COD of an antibiotic formulation effluent containing penicillin (COD ¼ 830 mg L À1 ) was removed by 10e56% during ozonation process while the addition of small amounts of hydrogen peroxide increased the removal efficiency (83%). In another study of Arslan Alaton and Dogruel (2004) the COD and TOC of the formulation effluent containing penicillin was removed by 49% and 52% respectively under alkaline conditions (pH ¼ 11), whereas the removal efficiency was much lower under acidic conditions (pH ¼ 3) (COD removal max ¼ 15%; TOC removal max ¼ 2%). Many authors (Balcıoglu and Otker, 2003;Arslan Alaton et al., 2004;Andreozzi et al., 2005) suggested that pH is a critical parameter in the ozonation process and a decrease of pH usually affects the reaction rate and also the absorption rates of ozone. During wastewater ozonation, many antibiotics, including b-lactams, sulfonamides, macrolides, quinolones, trimethoprim and tetracyclines, have been shown to be transformed predominantly via direct oxidation by O 3 whereas penicillin G, cephalexin and N 4 -acetylsulfamethoxazole were transformed to a large extent by hydroxyl radicals (Dodd et al., 2006). Ozone and/or hydroxyl radicals deactivate bactericidal properties of antibiotics by attacking or modulating their pharmaceutically active functional groups, such as N-etheroxime and dimethylamino groups of macrolides (Lange et al., 2006;Dodd et al., 2009), aniline moieties of sulfonamides (Huber et al., 2005), thioether groups of penicillins, unsaturated bonds of cephalosporin and the phenol ring of trimethoprim (Dodd et al., 2009). The high removals (>90%) by ozonation were achieved for those compounds with electron-rich aromatic systems, such as hydroxyl, amino (e.g. sulfamethoxazoles), acylamino, alkoxy and alkyl aromatic compounds, as well as those compounds with deprotonated amine (e.g. erythromycin, ofloxacin and trimethoprim) and non-aromatic alkene groups since these key structural moieties are highly amendable to oxidative attack (Dickenson et al., 2009).
Research conducted so far demonstrates that ozonation is a promising approach to degrade antibiotics. According to Table 2, ozonation was found to be an effective process for removing b-lactams, macrolides, sulfonamides and trimethoprim, quinolones, tetracyclines and lincosamides. The energy consumption for upgrading a Swiss municipal wastewater treatment plant with ozonation was evaluated by Hollender et al. (2009). For an ozone dose of 0.6 g O 3 g À1 DOC (effluent DOC w5 g m À3 ), 0.035 kWh m À3 wastewater was consumed, which is 12% of the total energy consumption of a typical nutrient removal plant (0.3 kWh m À3 wastewater). Additionally, 0.01e0.015 kWh m À3 was needed for pure oxygen production. Ozone treatment performance may be enhanced if ozone is combined with UV irradiation, hydrogen peroxide or catalysts (usually iron or copper complexes) (Klavarioti et al., 2009). However, optimal process and operating conditions have yet to be determined for the various water and wastewater types as well as for the different types of antibiotics (Yargeau and Leclair, 2008).

Fenton oxidation
Fenton's oxidation is a homogeneous oxidation process and is considered to be a metal-catalyzed oxidation reaction, in which iron acts as the catalyst (Tekin et al., 2006;Saritha et al., 2007). The main disadvantage of the process is the low pH value required in order to avoid iron precipitation that takes place at higher pH (Melero et al., 2007;Santos et al., 2007). Trovó et al. (2008) observed that amoxicillin degradation was not influenced by the source of the irradiation during the photo-Fenton process and the removals of the antibiotic obtained were 89 and 85% under black light and solar irradiation, respectively. A similar study by Bautitz and Nogueira (2007) showed that tetracycline was removed by 80% during the photo-Fenton treatment using two types of iron and irradiation. Moreover, in a study by Arslan Alaton and Dogruel (2004) adequate COD and TOC removal rates were achieved during the photo-Fenton and photo-Fenton-like treatment of a formulation effluent containing penicillin. Trimethoprim was completely removed during solar-Fenton process in the study of Michael et al. (2012a) and it was found that the presence of organic carbon and higher salt content in the simulated wastewater and real secondary effluent, led to lower mineralization though per dose of hydrogen peroxide compared to ultrapure water. It is important to highlight that a new approach aimed at performing photo-Fenton treatment at neutral pH has been proposed by Klamerth et al. (2010) and De la Cruz et al. (2012). The efficiency of the modified photo-Fenton system is based on the reaction of dissolved organic matter (DOM) present in wastewaters with Fe 2þ leading to the formation of soluble iron-complexes. However, contaminants degradation and mineralization tend to be slower at neutral pH than at pH 3.0. Michael et al. (2012b) investigated the application of a solar photo-Fenton system for the degradation of antibiotics at low concentration level (mg L À1 ) in secondary treated domestic effluents at a pilot-scale. The examined antibiotics were ofloxacin and trimethoprim and the pilot treatment plant w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 9 5 7 e9 9 5 consisted of a compound parabolic collector reactor. The results demonstrated the efficiency of the process in removing enterococci, resistant to these two antibiotics, while the compounds themselves were completely eliminated. The total cost of a full-scale unit for the treatment of 150 m 3 day À1 of secondary wastewater effluent was estimated to be 0.85V m À3 . This value was found to be in agreement with a previous study of the photo-Fenton process in a pilot-scale set-up (Jordá et al., 2011).
Another approach was taken by Lee et al. (2009) who used ferrate (Fe(VI)) to oxidize micropollutants and remove phosphate by formation of ferric phosphates in wastewater. They showed that Fe(VI) doses higher than 5 mg Fe L À1 were capable of eliminating sulfamethoxazole and ciprofloxacin by more than 85%. In comparison to ozone, Fe(VI) was as effective or slightly less effective in terms of micropollutants oxidation, with Fe(VI) having the benefit of phosphate removal. In general, Fenton process has been extensively used with success for the oxidation of many classes of antibiotics including b-lactams, quinolones, trimethoprim and tetracyclines.

Heterogeneous photocatalysis with TiO 2
Heterogeneous photocatalysis by TiO 2 semiconductor is achieved usually by the illumination of a suspension of TiO 2 in aqueous solution with light energy greater than its bandgap energy. This leads to the formation of high energy electronhole pairs (e À /h þ ) which can migrate on the surface of the catalyst and can either recombine producing thermal energy, or participate in redox reactions with the compounds that are adsorbed on the catalyst's surface Schiavello, 1993;Robertson, 1996). The valence holes are strong oxidants and are able to oxidize various contaminants, as well as water, resulting in the formation of HO while the conduction band electrons are good reductants reducing the dissolved oxygen to O 2 À (Munter, 2001).
The study of Elmolla and Chaudhuri (2011) examined the feasibility of using combined TiO 2 photocatalysis (UV/TiO 2 / H 2 O 2 ) and sequencing batch biological reactor (SBR) process for the treatment of an antibiotic wastewater containing amoxicillin and cloxacillin. The complete removal of these compounds was observed at TiO 2 and H 2 O 2 doses of 1000 and 250 mg L À1 , respectively. Amoxicillin was also completely removed from urban wastewater treatment plant effluent using [TiO 2 ] ¼ 0.8 g L À1 after 120 min of treatment as reported by Rizzo et al. (2009). Ofloxacin in wastewater samples was removed by 60% using [TiO 2 ] ¼ 3 g L À1  while Hapeshi et al. (2010) reported that the DOC of a solution contained ofloxacin at 10 mg L À1 was reduced by 79% after 120 min of photocatalytic treatment using [TiO 2 ] ¼ 250 mg L À1 and [H 2 O 2 ] ¼ 0.07 mmol L À1 .
Besides some drawbacks of the heterogeneous photocatalysis (e.g. the rather small quantum yield of the process; the relatively narrow light-response range of TiO 2 ; the need of post-separation and recovery of the catalyst particles from the reaction mixture in aqueous slurry systems), TiO 2 seems to possess some interesting features, such as high chemical stability in a wide pH range, strong resistance to chemical breakdown and photocorrosion, commercial availability and good performance. The catalyst is also cheap and can be reused (Andreozzi et al., 1999;Malato et al., 2009). The properties of antibiotics to be treated such as pK a and molecular structure will determine not only the efficiency of their photocatalytic degradation but also the mechanisms of the oxidation products formation (i.e. contribution of HO radical and valence band holes oxidation pathway).

Sonolysis
Ultrasound irradiation or sonolysis is a relatively new process in water and wastewater treatment and therefore, has unsurprisingly received less attention than other AOPs. This is also reflected by the small number of publications concerning the treatment of pharmaceutical compounds. Ultrasound enhances chemical and physical changes in a liquid medium through the generation and subsequent destruction of cavitation bubbles. These bubbles grow over a period of a few cycles to an equilibrium size for the particular frequency applied. It is the fate of these bubbles when they collapse in succeeding compression cycles that generates the energy for chemical and mechanical effects (Parsons, 2004). The sonochemical degradation in aqueous phase involves several reaction pathways and zones such as pyrolysis inside the bubble and/or at the bubbleeliquid interface and hydroxyl radical-mediated reactions at the bubbleeliquid interface and/or in the liquid bulk. Pyrolytic reactions inside or near the bubble as well as solution radical chemistry are the two major pathways of sonochemical degradation (Emery et al., 2005).
According to the authors' best knowledge, only one paper is available up to now in the literature on the applicability of sonolysis to remove antibiotics from wastewater effluents. Naddeo et al. (2009) evaluated the ultrasonic process on the degradation of amoxicillin spiked in urban wastewater effluent. It was found that the amoxicillin conversion was enhanced at increased applied power densities, acidic conditions and in the presence of dissolved air and the maximum removal observed was 40%.
It is important to note that there is limited literature (Herná ndez-Sancho et al., 2010;Mahamuni and Adewuyi, 2010;Jordá et al., 2011;Hollender et al., 2009;Michael et al., 2012b) dealing with advanced wastewater treatment process economics although this aspect is a very important issue.

2.5.
Effect of disinfection on antibiotics removal

Chlorination
Limited studies have focused on the removal of antibiotics during wastewater treatment with chlorine. Chlorination is by far the most common method of wastewater disinfection and is used worldwide for the disinfection of pathogens before discharge into receiving streams, rivers or oceans. From the chlorinated species, hypochlorite (ClO À ) has the highest standard oxidation potential (E 0 ¼ 1.48 V), followed by chlorine gas (E 0 ¼ 1.36 V) and chlorine dioxide (E 0 ¼ 0.95 V) (Homem and . The two major disadvantages of using chlorine based disinfectants are (i) the safety hazards associated with storage, transportation and handling of chlorine, and (ii) the potential formation of disinfection byproducts.
The effective removal of antibiotics by chlorination from wastewater requires sufficient free chlorine concentration and contact time. For example, cephalexin which was w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 9 5 7 e9 9 5 removed by 91% in activated sludge treatment at the Stanley WWTP was further removed in the following disinfection process by 99%, resulting in a total removal of 100% in the Stanley WWTP whole treatment process (Li and Zhang, 2011). Li and Zhang (2011) also reported that during chlorine disinfection process roxithromycin was eliminated by a further 18% (total removal 53%), erythromycin-H 2 O by 24% (total removal 43%), sulfamethoxazole by 27% (total removal 73%) and trimethoprim by 40% (total removal 65%).

Ultraviolet irradiation
Ultraviolet (UV) disinfection is increasingly finding applications in UWTPs. Photolytic degradation can be either direct or indirect. In direct photolysis, the target contaminant (in this case the antibiotic compound) absorbs a solar photon, which leads to a break-up of the molecule. In an indirect photolysis mechanism, naturally occurring molecules in the system such as dissolved organic matter (DOM) act as sensitizing species which generates strong reactive agents e.g. singlet oxygen ( 1 O 2 ), hydroxyl radicals (HO ) or alkyl peroxyl radicals ( OOR) and hydrate electrons under solar radiation (Arnold and McNeill, 2007;Fatta-Kassinos et al., 2011b). Generally, the degradation of a compound by UV irradiation is affected by the UV energy absorption and the quantum yield of the compound. UV energy absorption is expressed as molar extinction coefficient, which is a measure of how strongly a chemical species absorbs light at a given wavelength that can be used for its degradation (Kim et al., 2009).
Ultraviolet irradiation has been widely used for the treatment of waters and wastewaters worldwide. Several studies have reported the effective treatment of UV irradiation for removal of antibiotics in wastewater effluents (Adams et al., 2002;Ryan et al., 2011;Yuan et al., 2011). It has been recently reported that at high UV doses of nearly 11,000e30,000 mJ cm 2 , an almost complete removal of tetracyclines and ciprofloxacin was achieved (Yuan et al., 2011). Kim et al. (2009) reported that sulfonamides (sulfamethoxazole and sulfadimethoxine) and quinolones (norfloxacin and nalidixic acid) showed high removal efficiency in the range of 86e100% during the UV process. In contrast to this, macrolides (clarithromycin, erythromycin and azithromycin) were removed by 24e34%. Among tetracyclines, chlorotetracycline concentration decreased to less than limit of detection during the UV process while only 15% removal efficiency was achieved for tetracycline. This can be explained by the low molar extinction coefficient of tetracycline (4108 M À1 cm À1 ) comparing to that of chlorotetracycline (18,868 M À1 cm À1 ).
Another study of photolysis was conducted by Arslan Alaton and Dogruel (2004) in which penicillin in the form of formulation effluent with total COD ¼ 1555 mg L À1 was treated under UV irradiation or UV combined with H 2 O 2 . In this study, the removal efficiency was very low compared to the others described above (COD removal max ¼ 22% and TOC removal max ¼ 10% with 30 and 40 mM of peroxide respectively) and this may be attributed to the complexity of the formulation effluent (high COD and TOC values). Zuccato et al. (2010) also reported complete elimination of amoxicillin in Varese WWTP with UV-light treatment. The addition of H 2 O 2 to UV has proven to be more efficient in removing antibiotics than UV alone, and lower fluence doses need to be applied for the same removal (Kim et al., 2009;Rosario-Ortiz et al., 2010;Yuan et al., 2011).
Many of the antibiotics have aromatic rings, structural moieties (such as phenol and nitro groups) heteroatoms, and other functional chromophore groups that can either absorb solar radiation or react with photogenerated transient species in natural waters (e.g. photo excited natural organic matter-NOM) (Fatta-Kassinos et al., 2011b). The organic matter (DOC, COD), UV dose, contact time and the chemical structure of the compound are important factors governing the removal efficiency of antibiotics during direct photolysis. This technology is only applicable to wastewater containing photosensitive compounds and waters with low COD concentrations (e.g. river, drinking waters) (Homem and . Furthermore, wastewater effluents have different organic compounds that may either inhibit or enhance the process by scavenging or generating oxidant species (humic and inorganic substances like dissolved metals) (Jiao et al., 2008). Generally, photolysis has proved to be less effective in degrading antibiotics in wastewater effluents and more energy demanding (Katsoyiannis et al., 2011) than e.g. ozonation.

3.
Concluding remarks and future trends The conventional sewage treatment facilities were never designed to deal with pharmaceutical compounds. Due to their highly variable physicochemical properties (chemical structure, solubility, octanol/water partition coefficient) as well as the operational conditions of the biological process, the efficiencies by which pharmaceuticals are removed vary substantially. Unfortunately, the lack of data concerning the biological treatment processes does not allow comparison among the various studies conducted, and there are only few studies, which comprehensively and systematically investigated operating conditions of the biological treatment. In general, MBR systems have been reported to be equal to or slightly more effective in removing some antibiotics compared to CAS treatment systems (Le-Minh et al., 2010); MBR is more expensive, but provides a more hygienic effluent due to the filtration. As a consequence of the inability of the most commonly applied biological treatments to sufficiently remove antibiotics, the latter are regarded as pseudopersistent contaminants due to their continual introduction into the environment and permanent presence. Advanced treatment, downstream of conventional biological process, can significantly improve antibiotics removal before effluent disposal. Although capital and operational costs of an advanced treatment increase the costs of conventional process, further improvement of micropollutants and other antibiotics removal, in line with possible stringent regulations might be difficult to achieve without advanced treatment. The installation of treatment techniques to remove antibiotics in wastewaters should also be flexible and allow their implementation not only in UWTPs, but also at important source points such as hospitals and the pharmaceutical industry.
More comprehensive studies are required to thoroughly understand the behavior of antibiotics under both conventional sewage treatment and advanced treatment processes and to gain more knowledge on the elimination processes w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 9 5 7 e9 9 5 within the UWTPs including sorption onto sewage sludge. Furthermore, studies should provide all basic treatment plant operational parameters since these are essential for later comparison or assessments.
It is important to underline also the fact that only little information is currently available with regard to transformation products formed in the environment or UWTPs and during oxidative treatment. Future research should include a dedicated focus on the potential formation of pharmacologically active or more toxic products during treatment processes. Additionaly, it is necessary to conduct research on the occurrence, fate and removal of humans' metabolites in UWTPs. Most antibiotics and their metabolites are excreted by humans after administration and therefore discharged to the municipal sewage; however, only little is known about their biodegradability in the aquatic environments.
From a practical point of view, it is necessary to study process integration to maximize the treatment performance in removing antibiotics and for disinfection including those that can use renewable energy resources to power the processes. Moreover, both environmental and economic assessments are considered necessary in the framework of industrial scale applications for the removal of antibiotic residues from wastewater.
Finally, evaluation of the negative impacts (i.e. antibiotic bacteria and resistance genes evolution, toxicity on organisms and plants) caused by the presence of antibiotics in the environment is considered as a necessity in order to reduce the risk for humans.