Antibiotic use during an influenza pandemic: downstream ecological effects and antibiotic resistance

The years 2009–2010 saw the ﬁrst pandemic virus in several decades. Only in retrospect has the low pathogenicity of the virus been able to be conﬁrmed. The pandemic saw as many deaths per capita as a seasonal inﬂuenza virus, but with the signiﬁcant difference that the young ( 18 years old (Kamigaki and Oshitani, 2009). Pharmaceuticals played an important part of health care during the inﬂuenza pandemic. Many nations implemented huge stockpiles of antivirals in response to the pandemic, but owing to the low pathogenicity of the virus, there was a negligible increase in existing antibiotic use over interpandemic usage. However, current estimates for antiviral and antibiotic use during a moderate and severe inﬂuenza pandemic are without historical precedent (Singer et al., 2011). Here we discuss the environmental and human health implications of a moderate or severe inﬂuenza pandemic with regard to Tamiﬂu itself and the use of antibiotics to treat secondary bacterial infections. Antibiotic use will be framed in the context of existing paradigms of antibiotic treatment and how these practices already contribute to human and environmental hazards and how these hazards might be minimized in the event of a moderate or severe inﬂuenza pandemic.

The years 2009-2010 saw the first pandemic virus in several decades. Only in retrospect has the low pathogenicity of the virus been able to be confirmed. The pandemic saw as many deaths per capita as a seasonal influenza virus, but with the significant difference that the young (,18 years) were atypically impacted over those .18 years old (Kamigaki and Oshitani, 2009). Pharmaceuticals played an important part of health care during the influenza pandemic. Many nations implemented huge stockpiles of antivirals in response to the pandemic, but owing to the low pathogenicity of the virus, there was a negligible increase in existing antibiotic use over interpandemic usage. However, current estimates for antiviral and antibiotic use during a moderate and severe influenza pandemic are without historical precedent (Singer et al., submitted). Here we discuss the environmental and human health implications of a moderate or severe influenza pandemic with regard to Tamiflu itself and the use of antibiotics to treat secondary bacterial infections. Antibiotic use will be framed in the context of existing paradigms of antibiotic treatment and how these practices already contribute to human and environmental hazards and how these hazards might be minimized in the event of a moderate or severe influenza pandemic.

Introduction
The World Health Organization (WHO) has recently documented that infectious diseases are not only spreading faster but emerging at an unprecedented rate of one or more per year, resulting in the addition of B40 infectious diseases that were unknown only a generation ago (WHO, 2007). As evident from the H1N1 pandemic of 2009, we cannot be certain whether highly pathogenic avian influenza A H5N1, H9N2, or another subtype will spark a human pandemic (Wan et al., 2008), but what does appear certain is that another influenza pandemic will occur at some point in the future in the absence of a universal influenza vaccine (Cabinet Office, 2008;U.S. Homeland Security Council, 2007). In response to this pressure, many countries worldwide have published pandemic preparedness plans (European Influenza Surveillance Scheme, 2008;Mounier-Jack and Coker, 2006). The aims of these plans are to maintain essential services, reduce disease transmission, minimize the socioeconomic consequences of a pandemic, and reduce clinical cases, hospitalizations, and deaths (WHO, 2005). Key to the plans is slowing the spread of pandemic influenza through: (i) vaccine development, stockpiling, and distribution (Department of Health and Human Services and Department of Homeland Security, 2007), (ii) nonpharmaceutical measures (U.S. Centers for Disease Control and Prevention, 2007), and (iii) antiviral and antibiotic stockpiling and distribution (Hampson, 2008;NAS, 2008;U.S. Department of Health and Human Services, 2007). While the existence of preparedness plans in many countries highlights the attention that is given to the effects a pandemic on society, possible implications of pharmaceutical therapies on the environment have been addressed much less. This chapter addresses how concentrations of antivirals and antibiotics occurring in wastewater treatment plants (WWTP) and surface water can be simulated, and whether these concentrations could have environmental effects or favor resistance development.

Influenza-Symptoms and Treatment
Influenza in adults is characterized by the abrupt onset of fever, myalgia, headache, malaise, nonproductive cough, sore throat, and rhinitis. Otitis media, nausea, and vomiting are the more commonly reported symptoms of influenza in children. Pronounced elevation of proinflammatory cytokines during H5N1 influenza virus infection, known as the "cytokine storm," is hypothesized to be the main cause of pathology and ultimately of death from uncomplicated influenza (Salomon et al., 2007). Although uncomplicated influenza typically resolves within 7 days, a cough and malaise can persist for .2 weeks. Influenza can cause viral pneumonia, exacerbate underlying medical conditions (e.g., pulmonary or cardiac disease), and lead to secondary bacterial pneumonia.
A vaccine is the first line of defense against influenza virus infections; however, its production is hampered by the inability to predict the antigentic details of the pandemic strain before it arrives. Moreover, a vaccine will only become available roughly 4-6 months after the outset of a pandemic as a result of an outdated vaccine production system (e.g., egg culture) and insufficient production capacity (WHO, 2006a). Owing to this delay, it is estimated that only 14% of the world's population are predicted to have available to them a vaccine for the pandemic influenza strain within the first year (Osterholm, 2005;Uscher-Pines et al., 2006). Given these uncertainties, the WHO has recommended a number of mitigating strategies to help slow the spread of the pandemic, thereby providing additional time for a vaccine to be developed, distributed, and administered. Antivirals are to play a key role in this mitigation strategy.
Treatment for influenza using antivirals must be achieved within 48 hours of the onset of symptoms. Empirical antiviral therapy will be used in the absence of rapid diagnostic tests for influenza infection (Lim, 2007;U.S. Department of Health and Human Services, 2008). Currently, licensed antivirals for therapy and/or prophylaxis of influenza fall into two classes-the adamantanes (amantadine and rimantadine), M2 ion channel inhibitors effective against influenza A viruses only, and the neuraminidase inhibitors [NAI; oseltamivir ethylester phosphate (OE-P; Tamiflu) and zanamivir (Relenza)]-that are effective against both influenza A and B viruses. Although amantadine (Symmetrel) and rimantadine (Flumadine) are approved by the U.S. Food and Drug Administration (FDA) for the treatment and/or prevention of influenza, adamantane resistance is high and growing and is therefore not recommended for use during an influenza pandemic (Barr et al., in press;Hurt et al., 2007;Jefferson, 2007). Neuraminidase inhibitors (NAI), such as Tamiflu, Relenza, and Peramivir, are sialic acid analogs that inhibit the influenza neuraminidase enzyme, which is required for the release of progeny virions from infected cells. An inhibited neuraminidase ensures that the virions remain tethered to the infected cell surface, thereby limiting viral shedding (Ong and Hayden, 2007). Neuraminidase enzyme IC 50 values for oseltamivir for clinically isolated influenza A ranged from 0.1 to 1.3 nM, and for influenza B was 2.6 nM (Roche, 2007).
The WHO has strongly recommended the use of Tamiflu (oseltamivir carboxylate; produced and distributed by Hoffmannn-La Roche) as the primary choice for combating an influenza pandemic as: (i) it is easy to administer (capsule), (ii) it is systemically active, and (iii) it is effective against characterized influenza A and B viruses (Roche Group, 2006;WHO, 2006b).

Antiviral Stockpiling
Many countries have stockpiled antivirals as part of their preparedness plans. Even before the 2009 pandemic, Roche reported orders for Tamiflu from more than 80 countries, equating to approximately 215 million treatments (Tierney and Reddy, 2007). Many countries, including the United States and the United Kingdom, are supplementing their antiviral stockpiles (Reuters, 2007) to include Relenza, (Zanamivir, marketed and distributed by GlaxoSmithKline), with the expectation of further supplementation with drugs soon to be available, such as Peramivir (marketed and distributed by BioCryst Pharmaceuticals) (Morse, 2007;Pharmaceuticals, 2007) and A-315675 (produced by Abbott Laboratories) (Kati et al., 2002), among many others in development.
Chemotherapy guidelines for treating IA-CAP have been recently published with the backing of the British Infection Society, British Thorax Society, and the Health Protection Agency (Lim, 2007). This guidance is tailored toward pandemic IA-CAP, which distinguishes itself from previously published CAP guidelines, for example, American Thoracic Society (ATS) (ATS, 2005). Antibiotic therapy for IA-CAP will be empirical as a result of the anticipated overwhelming surge of patients during a pandemic (Gupta et al., 2008;Nap et al., 2007;UK Department of Health, 2007a).
Patients with IA-CAP infections receive antibiotic therapy and a range of overthe-counter (OTC) medications, for example antipyretics, antiphlogistics, cough medicine, and decongestants, used to relieve some of the discomfort of the illness (U.S. Department of Health and Human Services, 2005). In addition to treating the influenza virus infection, there will be considerable use of OTC drugs including analgesics and antiphlogistics. Personal stockpiles of these drugs have been recommended for all families within the U. S. Pandemic Flu Preparedness Checklist for Individuals & Families (Meltzer et al., 1999;U.S. Department of Health and Human Services, 2006).
The availability of antibiotics within a country is considerably less defined than it is for antivirals. Recently, the UK Department of Health announced the intention to purchase 14.7 million courses of antibiotics, to cover at-risk groups totaling 25% of the population (Department of Health, 2007). However, to date, there was no indication of the nature of the stockpile. Owing to the threat of bioterrorism, many countries maintain a stockpile of antibiotics, primarily ciprofloxacin, for postexposure anthrax treatment (Gupta et al., 2008). The United States stockpiles sufficient antibiotics (ciprofloxacin and doxycycline) to treat 60 million people. Both of these antibiotics could be mobilized to treat some, but not all, of the causes of IA-CAP (Fowler et al., 2005;Gupta et al., 2008).

EXPOSURE ASSESSMENT OF ANTIBIOTICS AND ANTIVIRALS DURING A PANDEMIC
In order to determine risks of pandemic antiviral and antibiotic loads in the aquatic environment, knowledge on the concentrations in WWTPs and surface water as well as on possible effects on WWTP bacterial communities and occurrence of antibiotic resistance is needed. In this section, we summarize how concentrations can be predicted from coupling a spatially structured global epidemic model with a GIS-based watershed model. To our knowledge, the modeling approaches described in (Singer et al., submitted) and in this chapter are the first investigations of possible antibiotic concentrations during an influenza pandemic.
The spatiotemporal pattern of the pandemic and its concurrent secondary infections were simulated by use of the global epidemic and mobility (GLEaM) model (Balcan et al., 2009a), which maps 6 billion individuals and integrates their mobility data. Parameters of the infection dynamics consisted of transmission of influenza under seasonal variation and included secondary infections according to the UK Pandemic Assumptions for complication, hospitalization, and intensive care unit admission rates (Balcan et al., 2009b) (UK Department of Health, 2009). In order to incorporate a wide range of possible influenza transmission potentials, we explored a mild pandemic (R 0 ¼ 1.65), a moderate (R 0 ¼ 1.9), as well as a severe (R 0 ¼ 2.3) scenario. R 0 , the basic reproductive number, is the average number of secondary infections produced by a single infected individual while he or she is infectious in an entirely susceptible population. This is a measure of the degree of transmissibility of an infection (Anderson and May, 1991). No large-scale antiviral treatment was included in the mild scenario. For the other two scenarios, Tamiflu antiviral treatment was foreseen, with the assumption that 30% of the cases were detected and antiviral treated. It was assumed that antiviral treatment reduced the infectious period and the rate of development of secondary infections (Kaiser et al., 2003;Nicholson et al., 2000;Treanor et al., 2000;Whitley et al., 2001).
Within each scenario, we explored the effect of antiviral prophylaxis for 2 or 4 weeks from the start of the outbreak per country, as well as the effect of no prophylaxis. Antibiotics were assumed to be used according to guidelines for treating influenza-associated pneumonia sanctioned by the British Infection Society, British Thoracic Society, and the UK Health Protection Agency (Lim, 2007) (Table 26.1). It should be noted that this is only a small subset of the drugs that have been proposed to be used during a pandemic. Table 26.2 presents data on the environmental fate of a wide range of antibiotics that may be used during a pandemic, reflecting the considerable variation in antibiotic prescribing that exists regionally, nationally, and globally (Coenen et al., 2006;de Neeling et al., 2001;Ferech et al., 2006;Garcı´a-Rey et al., 2002;Muller et al., 2007;Priest et al., 2001); and the possibility that, dependent on the extent of any stockpile, shortages are likely to occur and a range of antibiotics might be used (Gupta et al., 2008). Table 26.3 summarizes data on concentrations in WWTPs and surface water for pharmaceuticals that have been proposed for use during an influenza pandemic.
Simulations of an influenza pandemic (R 0 ) resulted in estimates of the cases of influenza and pneumonia at each stage of disease progression, together with the quantities of drugs used. Simulations were able to project cases and drugs down to the spatial resolution scale of 1 /4 and a time resolution of one day. A water quality model, Low Flows 2000-Water Quality Extension (LF2000-WQX) (Rowney et al., 2009), then predicted the environmental release of antivirals and antibiotics. Values for excretion of active substances within the feces and/or urine were taken from the literature (see Table 26.4). Dilution processes in the river were modeled using the annual mean flow from each river stretch, while a worst-case scenario of no pharmaceutical degradation or loss to sorption in WWTPs and rivers was applied (see Fig. 26.1). The Thames catchment in the southern United Kingdom was chosen as a model test case.

EXPOSURE ASSESSMENT OF ANTIBIOTICS
A survey of the literature as well as an examination of the STPWIN model (U.S. Environmental Protection Agency, 2007) within the Estimation Program Interface (EPI) SuiteTM 4.0 indicates low (,20%) removal for most antibiotics in WWTPs, inclusive of loss due to sorption and biodegradation ( Fig. 26.1). A literature search revealed that most antibiotics, particularly those not containing a β-lactam moiety, are resistant to metabolism in vivo as well as in the environment (Al-Ahmad et al., 1999;Alexy et al., 2004;Benotti and Brownawell, 2009;Brain et al., 2004a;2004b;Gartiser et al., 2007;Junker et al., 2006;Ku¨mmerer et al., 2000), with half-lives of days to weeks (Benotti and Brownawell, 2009;Christensen, 1998). For these reasons, we feel there was justification in assuming a conservative pharmaceutical biodegradation model of zero degradation/sorption.

Interpandemic Antibiotic and Antiviral Use
A very wide range of pharmaceuticals are in constant use in a population and thus will be present in the wastewater during a pandemic (Ku¨mmerer, 2009a(Ku¨mmerer, , 2009b. While the relative contribution of single antibiotic classes to overall use does not greatly vary between countries ( Fig. 26.2), the absolute amount of defined daily Fluoroquinolone with some enhanced pneumococcal activity (e.g., levofloxacin 500 mg od IV or moxifloxacin 400 mg od PO) plus, either macrolide (erythromycin 500 mg qds IV or clarithromycin 500 mg bd IV) or β-lactamase stable antibiotic (i.e., co-amoxiclav 1.2 g tds IV or cefuroxime 1.5 g tds IV or cefotaxime 1 g tds IV)  Total excretion is the sum of the % parent chemical released in the urine and feces. Metabolites were only included in the sum if the metabolite has a known bioactivity.
c There is no official DDD for clavulanic acid. Clavulanic acid is administered in conjunction with amoxicillin in the form of Co-amoxiclav. In this case we used a typical adult dose of 0.125 g and assumed 3 doses per day, equating to a DDD equivalent of 0.375 g.
d Oseltamivir carboxylate (OC) is the active metabolite of the prodrug oseltamivir ethylester phosphate (OE-P); 20% of OE-P is excreted in feces, 80% of OE-P is excreted in urine as OC (He et al. 1999 (Gartiser et al. 2007); 13 (Junker et al. 2006); 14 ); 15 (Ternes et al. 2002); 16 (Lindberg et al. 2007); 17 (Liebig et al. 2006); 18 (Ku¨mmerer et al. 2000); 19 (Rabiet et al. 2006); 20 (Vieno et al. 2005); 21 (Tixier et al. 2003); 22 (Wennmalm 2005); 23 (Gomez et al. 2007); 24 (Trenholm et al. 2006); 25 (Choi et al. 2007); 26 (Yang and Carlson 2003); 27 (Wiegel et al. 2004); 28 (Ashton et al. 2004); 29 (Ahrer et al. 2001); 30 (Ternes 2001); 31 (Carlsson et al. 2006); 32 (Buser et al. 1998); 33 (Stackelberg et al. 2004); 34 (Miao et al. 2004); 35 (Costanzo et al. 2005); 36 (Carballa et al. 2004); 37 (Batt et al. 2006); 38 (Bound and Voulvoulis 2006); 39 (Brown et al. 2006); 40 (Brun et al. 2006); 41 (Brun et al. 2006 dosages per 1000 inhabitants per day ranges between 10 and .20 in Europe (Coenen et al., 2006;Ferech et al., 2006;Muller et al., 2007). If one was to use the average annual antibiotic usage within England during an interpandemic period (NHS BSA, 2008) as a first approximation, 62 μg antibiotics/L would be present in the UK WWTPs, which was assumed to be diluted by the median volume of wastewater in WWTP within the Thames catchment are a, 230 L/head/day (Table 26.5). Table 26.3 and Figure 26.3 give an extensive overview on the concentrations of antibiotics that have been found in WWTP and surface water globally as well as in the United Kingdom. When comparing predicted and actual concentrations of antibiotics in WWTPs (Table 26.5 vs Table 26.3), it should be noted that veterinary use significantly contributes to concentrations in surface water (Hurd and Raef, 2010;Smith et al., 2002) such that WWTP effluent is not the only source of antibiotic residues in the aquatic environment. Over-and underestimations of predicted concentrations of antibiotics in the environment can also be caused by unforeseen losses from sorption and/or degradation, the impact of climate on degradation, difference in antibiotic use between different regions within a country, as well as countryspecific differences, seasonal fluctuations in drug use, as well as seasonal fluctuations on WWTPs and river dilution (ter Laak et al., 2010). Hence, it is very difficult to project the concentration of antibiotics in use during interpandemic periods for any one location at any particular time. For this study we employed annual usage statistics for England (2007) with no further modification or adjustment.
Unlike antibiotics, interpandemic usage of Tamiflu in the United Kingdom been reported to be negligible (Kramarz et al., 2009), implying that any substantial increase will be as a result of the influenza pandemic.

Projected Concentrations of Antibiotics and Tamiflu in WWTP and Rivers During a Pandemic
In WWTPs, a mild pandemic scenario was projected to increase interpandemic antibiotic use of the same antibiotics (see Section ) by only approximately 1% (95%  Mean concentrations of total antibiotics in the Thames catchment were projected to be ,0.09 and ,0.8 μg/L for a mild and moderate pandemic, respectively (Table  26.7). On the other hand, a severe pandemic was projected to achieve ,15 μg/L total antibiotics, with a maximum of 80 μg/L. A mild and moderate pandemic with AVP.0 was projected to generate mean concentrations of Tamiflu in the Thames catchment between 1.1 and 11.5 μg/L (Table 26.7). A more severe pandemic, regardless of AVP, was projected to result in mean concentrations of Tamiflu in the Thames catchment in excess of 100 μg/L, consistent with previous projections of a severe pandemic in southern England (Singer et al., 2007).

EFFECTS ASSESSMENT OF ANTIBIOTICS DURING A PANDEMIC
Elevated concentrations of antibiotics during a pandemic will, in the first instance, affect microbial consortia in WWTPs and surface waters. In order to evaluate such potential effects of antibiotics, we determined the fraction of bacteria that were potentially affected by a given antibiotic exposure [(potentially affected fraction, PAF). This approach projects effects on whole communities from toxicities of a substance to single members of the community compiled in species-sensitivity distributions (SSD) (Newman et al., 2002)]. As experimental data on the toxicity of the chosen antibiotics is almost entirely lacking, we based our analyses on minimum inhibitory concentrations (MIC) of human pathogens from the EUCAST database (EUCAST, 2009 substance on its own). Toxicities were determined for WWTP and river stretches in the Thames catchment area from antibiotic concentrations predicted through exposure modeling. The results of this determination of the potentially affected fraction of bacteria in the community based on simulated antibiotic concentrations are shown in Figures 26.4 and 26.5. For a mild pandemic, projected toxicity in WWTP was well below 1% PAF (less than 1% of the community might be growth inhibited, Fig. 26.4a). During a moderate pandemic, toxicities .5% were predicted in 74% of the WWTP for concentrations at the upper bound of the 95% RR of the stochastic model, while no toxicity was predicted for the lower bound ( Fig. 26.4a). A proportion of growth-inhibited species .5% was chosen as a pragmatic threshold for possible effects on community functioning (European Chemicals Agency, 2008). The severe pandemic was projected to affect between 8 and 32% of the microbial species in WWTP (Figs. 26.4b and 26.5c). a Mean values are inclusive of all excreted antibiotics. Values reflect the median epidemic scenario for each condition and reflect the mean concentration for all 461 river stretches utilized within LF2000-WQX of parent pharmaceuticals investigated in this study excreted in the feces and urine unchanged and/or as a bioactive metabolite. b s1 ¼ where AVP=0, rate of AVT ¼ 30%, limited supply of Tamiflu. s2 ¼ where 2 wk AVP, AVP=1%, rate of AVT ¼ 30%, limited supply of Tamiflu. s3 ¼ where 4 wk AVP, AVP=1%, rate of AVT ¼ 30%, limited supply of Tamiflu. s4 ¼ where 2 wk AVP, AVP=10%, rate of AVT ¼ 30%, limited supply of Tamiflu. s5 ¼ where 4 wk AVP, AVP=10%, rate of AVT ¼ 30%, limited supply of Tamiflu. s6 ¼ where AVP=0, rate of AVT ¼ 30%, unlimited supply of Tamiflu.
Absolute toxicity in rivers was projected to be slightly lower than in WWTPs, with the maximum PAF for any river stretch in a moderate pandemic being B15% (Fig.  26.4e). During a severe pandemic, the 5% threshold of toxicity would be exceeded in about half of the river stretches at the upper bound of the 95% R (Fig. 26.4c), corresponding to about one third of total river length (Figs. 26.4d and 26.5f). Maximum toxicity during a severe pandemic was simulated to reach 30% (Fig.  26.4e).
The same effect models, when applied to the background concentrations of the eight investigated antibiotics, yield toxicities (PAF) in Thames catchment WWTP of between 4 and 17%. At these concentrations, however, no major functional breakdowns have been recorded. Reasons for the tolerance of WWTP communities to chronic exposure to antibiotics might be: (1) reduction of bioavailable concentrations through degradation or sorption, (2) acquired community tolerance through selection for more tolerant or resistant bacteria, (3) differences between the antibiotic sensitivity of human pathogens and WWTP bacteria, and (4) overcapacity of WWTPs, thereby enabling some decline in function to occur with no significant loss in overall sewage treatment. As only the severe scenario increases the PAF substantially, we conclude that a mild or moderate pandemic would not be likely to affect WWTP consortia.
Notably, effects modeling in this study were based on MICs of human pathogens instead of effects observed in WWTP toxicity testing. Arguably, the toxicity of antibiotics to functionally active WWTP bacteria might differ from pathogen MICs. Also, the predominance of biofilms in WWTPs as flocs will influence antibiotic toxicity. There is little experimental data to verify whether modeled toxicity matches experimental data. Also, often, studies are based on test systems that differ from a full-scale WWTP, such as through the use of batch studies or through the use of synthetic sewage as medium.
Erythromycin is the only antibiotic investigated in this study for which substantial data on toxicity in WWTPs exists. For example, in batch studies with activated sludge inoculum from French WWTPs (Louvet et al., 2010a) and raw wastewater as medium, a range of effects of erythromycin were observed. After exposure to the antibiotic for less than one hour, direct toxicity was seen in cell staining experiments. In 24-hour batch experiments with inocula collected at nine different time points, 10 mg/L erythromycin (equivalent to a PAF of 80% according to our models) inhibited COD (chemical oxygen demand) removal by 79% on average. Inhibition strongly varied between different inocula, highlighting the dependence of batch toxicity studies on the initial inoculum. Inhibition of nitrification was 40% on average in similar experiments.
Inhibition was also found at lower concentrations of erythromycin (see Table  26.8). Further, erythromycin interfered with floc structure: flocs divided and biomass was transferred to the foam layer on top of the batch reactors (Louvet et al., 2010a). At longer exposure times of batch reactors (5 days), effects on soluble COD were observed at erythromycin concentrations as low as 4 μg/L (Louvet et al., 2010b),     (Prado et al. 2009); which is equivalent to a PAF of 0.2%. Concentrations higher than 500 μg/L also inhibited nitrification. At 4 μg/L, erythromycin-treated reactors showed higher nitrification as compared to control reactors. This was explained through the extra source of nitrogen provided by dead bacteria. In contrast, Fan et al. (2009) found that erythromycin (at concentrations of 500 μg/L or a single-substance PAF of 46% according to our calculations) did not have pronounced effects on nitrogen and phosphorus removal in sequencing batch reactors fed synthetic wastewater during long incubations (Fan et al., 2009). Still, phylogenetic investigations showed that the diversity of ammonium-oxidizing bacteria and of nitrifying bacteria had declined by up to 80%, highlighting effects of the antibiotic on the composition of the functional community. Further, shortterm toxicity measurements with higher concentrations of erythromycin and H 2 O-erythromycin showed that while the adapted nitrifier and ammonium oxidizer communities were more tolerant to high doses of antibiotics, functional parameters were still affected at concentrations equal or higher than the long-term concentration (Fan et al., 2009). Thus, shock concentrations of antibiotics during the onset of a pandemic might affect WWTP functioning, although almost all treatment plants will be acclimated to lower antibiotic concentrations during the predominant interpandemic period. Further, while relatively low concentrations of single antibiotics are able to select for equally functional, but less diverse, antibiotic-tolerant communities, the question remains as to whether joint effects of several antibiotics might further reduce diversity and therefore functioning.
As seen above, investigations with full-scale treatment plants sometimes show a high functional stability also for other antibiotics. Most relevant to this study, Al-Ahmad et al. (1999) found that a mixture of 12 antibiotics tested in laboratoryscale treatment plants did not have effects on DOC elimination during 84 days of operation, even at concentrations that were 100 times higher than average influent concentrations.. In contrast, effects of a sudden change in wastewater composition were apparent in two WWTPs purifying waste water from a pharmaceutical production site. Failure in ammonium oxidation and changes in the composition of ammonium-oxidizing bacterial communities were found when the plant started producing imidazoles, and problems continued for a few weeks (Wittebolle et al., 2005).
To conclude, congruence of modeled toxicity with experimental data on toxicity of shock loads of antibiotics suggests that a severe pandemic might indeed provide antibiotic concentrations capable of affecting microbial WWTP consortia, at least for a short time. This could compromise vital and obligate microbial functions such as nitrification, phosphorus, and COD removal. With longer exposure duration, changes in community composition, together with an increase in genetically encoded antibiotic resistance (see Section 1.3.2), might preserve WWTP functioning.

Effects on Antibiotic Resistance
The projected widespread use of antibacterials during an influenza pandemic introduces an opportunity for the rapid spread of antibiotic resistance. Bacteria can become resistant to antibiotics by mutations or the acquisition of appropriate genes from other microorganisms (Verhoef and Fluit, 2006 Szczepanowski et al., 2009), showing that the human community discharges bacteria and mobile genetic elements carrying resistance genes to the WWTP in interpandemic times. Further, WWTPs have been discussed as environments for gene transfer between bacteria (Geisenberger et al., 1999), and the presence of residual antibiotics could speed such genetic exchange. In any case, massive use of antibiotics during the pandemic will further resistance development in human microbiota (Barlow, 2009), and a significant proportion of these will reach WWTPs and be discharged with WWTP effluent after incomplete purification. An additional question is whether elevated concentrations of antibiotics during a pandemic might be high enough for selection and recombination of resistant bacteria in the WWTP. There is little experimental evidence to prove or disprove this idea, as most studies so far have investigated treatment plants that also received human waste, presumably containing resistant bacteria. However, recent investigations in WWTPs treating waste from antibiotic production sites showed that high concentrations of antibiotics can indeed select for resistant populations that can subsequently be discharged to the receiving rivers (Li et al., 2009;. In the first production plant, concentrations of penicillin G were around 150 μg/L (PAF of amoxicillin: 34%). Extremely high MICs for β-lactam antibiotics were observed in isolates from the WWTP, while resistance to unrelated antibiotics was also elevated. In the second plant, which produced oxytetracycline (30.5 mg/L in the sludge, equivalent to a doxycycline PAF of 92%), extremely high MICs were again observed in isolates tested for resistance to different tetracyclines. Also, many isolates showed multiresistance to unrelated antibiotics. Isolates often contained class I integrons, genetic elements that can integrate a multitude of resistance genes into bacterial genomes. From these two studies, it indeed appears that antibiotics can select for resistance in WWTPs.

CONCLUSIONS
Through epidemiological simulations and environmental and toxicity modeling, the concentrations and effects of antibiotics during an influenza pandemic were determined. It was shown that a severe pandemic might affect WWTP consortia and lead to increased nitrogen and COD release, at least for short periods. Pandemic concentrations are also likely to increase antibiotic resistance beyond levels already observed in WWTP. In line with modeling results obtained for a mild pandemic, no widespread WWTP failures were observed during the 2009/2010 influenza pandemic. Still, current experimental data is insufficient to disprove possible effects during an influenza pandemic characterized by a higher infectivity.