Surface or groundwater quality is a function of natural processes and human activities [1
]. Main natural processes are weathering of bedrock minerals, evapotranspiration, deposition of dust and salt, leaching of organic matter and nutrients from soil, hydrological factors leading to run-off, and biological processes. Human activities, such as discharging treated or untreated sewage, may determine point and nonpoint sources of pollution in both rural and urban areas, releasing, e.g., nitrates [6
]; metals such as mercury, lead, and cadmium [8
]; organics such as pesticides [9
]; and pharmaceutical compounds (PhCs) [10
]. Several physical, biological, and chemical methods were developed and applied for the removal of inorganic and organic compounds from water. According to the targeted components, treatment processes are generally classified into three main categories: primary, secondary, and tertiary treatments [14
]. Primary treatment removes inorganic and organic suspended solids through floatation, settling, and screening mechanisms, while secondary treatment removes the residual organic matter and, in some cases, dissolved nutrients through trickling filters consisting of bacteria-coated stones and bacterial activated sludges. Finally, tertiary treatment may be required to remove suspended and dissolved materials, such as nutrients and metals, to meet regulatory requirements and can be based on various chemical and biological treatments.
The presence of PhCs and their metabolites in the aquatic environment of several countries has been documented since the 1970s (e.g., [1
]). However, these contaminants have only recently received attention from the scientific community, the institutions, and the general public because of the concern over possible toxicological risks to the microbial community, fish, wild fauna, and humans and the development of microbial resistance to antibiotics [16
A great number of PhCs have been reported in surface water, groundwater, and drinkable water all over the world, at concentrations ranging from few nanograms per liter to hundreds of micrograms per liter [10
]. The direct excretion of PhCs through urine and feces by humans and animals consuming drugs represents the main and widespread source of PhCs released into the environment, while pharmaceutical industries are the secondary point source [24
]. Moreover, the use of sludge coming from wastewater treatment plants (WWTPs), manure from intensive animal farming used as fertilizer, and irrigation with reclaimed water could also bring PhCs into agricultural soils [10
PhC removal in conventional secondary WWTPs is often incomplete, as they are not designed for this purpose [28
]. Moreover, the share of the population connected to wastewater treatment plants varies considerably worldwide and in Europe, as does the level of treatment (primary, secondary, or tertiary) [31
]. In Italy, advanced plants fulfilled the needs of 67% of the total population in 2018 [32
]. As required by the Urban Waste Water Treatment Directive (Directive 91/271/EEC), all agglomerations with a population of more than 2000 should be provided with collecting systems for urban wastewater and urban wastewater entering collecting systems should be adequately treated before discharge. However, the European Commission decided to refer Italy to the Court of Justice of the EU because 620 agglomerations in 16 regions are in breach of EU rules on collection or treatment of urban wastewater [33
At the European level, the Directive 2013/39/EU focused on emerging contaminants and PhCs, and the Decisions 2015/495, 2018/840, and 2020/1161 identified a watch list of substances that pose a significant risk to the aquatic environment [34
]. These substances include biocides and pharmaceuticals. The assessment of PhCs in the environment should be carried out by monitoring programs, providing measured environmental concentrations (MECs), or using predictive models, based mainly on human consumption and excretion/removal and dilution factors [37
], providing predicted environmental concentration (PEC). The European Medicines Agency (EMA) recommended the estimation of PEC for the environmental risk assessment of PhCs for human use [40
]. For any given PhC, the PEC can be calculated from sale volume and human excretion rate, i.e., the percentage excreted as parent compound, and dilution of wastewater by the surface water flow. The PEC of the effluents of WWTs could be refined by applying the removal factor of the specific WWTP. However, assumptions made during the calculation of PEC values, such as an evenly distributed usage over time and space, may not be appropriate and could vary also depending on the hydrology at the local scale. Thus, the relevance of PEC vs. MEC should be evaluated, especially at the local scale, where the pattern of consumption could differ from the regional or national one when the density of population greatly varies, according to the distance to the cities.
So far, the model for PEC estimates of PhCs [40
] was applied in Italy by Verlicchi et al. [41
] in the effluents of a WWTP and by Riva et al. [42
] in both WWTP effluent and surface water bodies. They showed that the PEC was close to and sometimes overlapped the MEC, but differences between PEC and MEC were significant for some PhCs. Conversely, Ong et al. [43
] and He et al. [44
] demonstrated the usefulness of the predictive model in Melbourne (Australia) and Japan, respectively, whereas Neves et al. [45
] successfully slightly modified the model for 10 of the most used PhCs in Belo Horizonte (Brazil). The modification was based on refinements of the parameters (i.e., human consumption excretion/removal and dilution factors) of the formula, allowing a better representation of the city of study. At the same time, Gomez-Canela et al. [46
] refined the model [40
] applied to PEC prediction in Catalonia rivers, using the dilution factor proposed by Keller et al. [47
] from Spain. Accordingly, other authors [48
] highlighted the impact of climate parameters on the dilution factor, especially for carbamazepine, diclofenac, and metoprolol in small and medium rivers downstream WWTPs in Germany.
Thus, the present study aimed at investigating the occurrence of some PhC contaminants in surface water and in the effluent of a WWTP located in the coastal area of Central Italy and at improving the predictive model recommended by the EMA [40
] comparing the PEC of PhCs in surface water with the MEC. An additional objective was to evaluate the environmental risk that the presence of PhCs can produce for the environment. To these aims, in the study area, we defined three categories of anthropic pressure on the basis of the number of inhabitants per unit surface (rural, periurban, and urban areas), and we performed two sampling campaigns of surface water and analyzed the samples for 12 selected PhCs, representing the main therapeutic classes. The PEC was calculated from the sale date and the formula for the calculation of the PEC was modified to take into account the number of inhabitants in the three categories of anthropogenic pressure. PEC and MEC values were compared, and the relevance of PEC values was assessed according to the PEC/MEC ratio. Finally, the environmental risk of the PhCs was assessed by the risk quotient (RQ), which is the ratio between MEC and the corresponding predicted no-effect concentration (PNEC), calculated by dividing the lowest chronic no-observed-effect concentration from standard toxicity tests by an assessment factor.
The analysis of the samples of surface water and effluent of WWTP in the study area confirmed the presence of all selected PhCs or metabolites. Overall, averaged between sampling campaigns, the concentration of retrieved PhCs in surface water were in the range from 1.7 to 1243 ng L−1
(for naproxen and metformin, respectively), while concentrations in WWTP effluents were in the range from 0.07 to 1824 ng L−1
(for estradiol and metformin, respectively). It is noteworthy that several PhCs exceeded the threshold proposed by the European Agency for the Evaluation of Medicinal Products [67
] for environmental risk (10 ng L−1
These results were consistent with the results of monitoring performed by many authors worldwide, e.g., in Italy [28
], Sweden [69
], and China [70
]. A synthesis of research on PhC occurrence in freshwaters at national, regional, and global scales presented in [23
] confirmed that investigations were mainly concentrated in North America, Europe, and the most populous parts of China. In Europe, research efforts were clustered around the high population areas of, e.g., London, Paris, Hamburg, and Frankfurt. Conversely, knowledge of PhC occurrence is poor or absent for large parts of the globe, particularly in developing countries.
Some PhCs, such as carbamazepine, metformin, and sulfamethoxazole in both sampling campaigns and dehydro-erythromycin, diclofenac, and 10,11-dehydro-10,11-dehydroxycarbamazepine in the sampling campaign of September, even increased in the effluent of WWTP compared with the surface water of all the three areas (rural, periurban, and urban). This can be explained by the higher efficiency of the plants, soil, and associated microorganisms (bacteria, fungi, etc.) in PhC degradation compared with WWTPs including primary and secondary treatments [71
]. Thus, the peculiar behavior of these PhCs may depend on the chemical and physical characteristics of their active principles and on the characteristics of the interacting environment. Indeed, for some of these PhCs, the degradative processes, able to remove or break them down into nondangerous and inactive molecules, are known, while for others, no degradation occurs in realistic environmental conditions, and this means that they remain unchanged or form metabolites that can be even more active and harmful than the parent molecule. In the environment where plants, soil, and associated microorganisms interact, the degradation can be physical, chemical (e.g., photodegradation, oxidation), or biological (i.e., carried out by microorganisms such as fungi and bacteria or by plants) [73
], as in constructed wetlands (CWs). Accordingly, CW plants were found to be more efficient in the removal of PhCs than actual WWTPs [75
]. It is noteworthy to cite as an example diclofenac, which is highly photodegradable and was found to be rapidly mineralized in various agricultural soils (with half-lives lower than 5 days) but stable in soils following sterilization by autoclaving, indicating the active role played by microorganisms in its degradation [76
]. Moreover, as regards metabolites, in the present study, we measured those known to be produced and potentially toxic, i.e., dehydro-erythromycin from erythromycin and 10,11-dehydro-10,11-dehydroxycarbamazepine from carbamazepine [77
]. The fact that we detected only the metabolite of erythromycin (i.e., dehydro-erythromycin) in surface water samples of both campaigns can be explained by the rapid degradation of this molecule (50% dissipation time 7–17 days) with respect to other PhCs, such as carbamazepine [78
]. Accordingly, Guasch et al. [79
] reported higher concentrations of dehydro-erythromycin (up to 2.5 μg L−1
) than erythromycin (from very low nanograms per liter to very few micrograms per liter) in various environmental matrices and in effluents.
Overall, our results suggest a growing gradient of concentration from the rural to the urban areas: the samples collected in the urban surface water had concentrations of PhCs of one and in some cases four orders of magnitude higher compared to the concentrations measured in the suburban/rural surface water. This can be explained not only by the higher density of resident population and thus by the higher PhC consumption and release, but also by specific activities normally performed in urban areas, such as medical care and services, tourism, sport, and hospitality [39
]. Thus, our results of increasing PhC pollution with an increasing number of inhabitants confirm the appropriateness of the risk-based approach targeting the receiving waters in densely populated areas [81
Although a consistent pattern of increase in environmental loads of all PhCs was evidenced with the increase in anthropic pressure from rural to urban areas, a great variability within the SwPurb was detected for some PhCs in the first sampling campaign (March), leading to nonsignificant statistical differences (i.e., clarithromycin and 10,11-dehydro-10,11-dehydroxycarbamazepine). This is likely to be due to variations in the suburban area in the numbers of inhabitants and activities or to differences in the flows of surface water in canals, causing an increase or decrease in PhC concentrations. Similar results were found in [37
]. In the former work [37
], similarly to our results, the occurrence of PhCs (i.e., atenolol, lincomycin, erythromycin, clarithromycin, bezafibrate, and furosemide) in sampling sites located along the rivers Po and Lambro in North Italy was different among subbasins and was strictly correlated with the presence of large human settlements and/or animal farms. Similarly, in the latter work [80
], tourism and the associated number of tourist arrivals were demonstrated as a significant contributor to the overall PhC pollution of the Alpine aquatic environment. Indeed, similarly to our results, the potential impact of areas with high population density due to tourism was demonstrated to strongly affect analgesic/anti-inflammatory compounds, such as diclofenac.
The monitoring of the seasonal variation (March vs. September sampling campaign) in the concentrations of the 12 studied compounds showed variations between rural and periurban surface waters, as well as within surface water categories, for some PhCs (i.e., atenolol, carbamazepine, clarithromycin, dehydro-erythromycin, diclofenac, and 10,11-dehydro-10,11-dehydrocarbamazepine). The seasonal variability might be due to the reduction in the consumption of these PhCs in rural and periurban areas in the summertime, and thus the concentration and variability in sample concentrations are lower in September. Conversely, for the other PhCs, no evident seasonal variation was detected either for the surface waters or for the effluents of the WWTP. This is likely to be due to the fact the selected PhCs are highly consumed throughout the year in the urban areas and that the temperature (from March to September) did not have any influence on the improvement of the removal rate, thus not affecting the urban surface water and potentially also the influents of WWTP.
The comparison with predicted and measured concentrations of PhCs in surface water and in WWTP effluent showed that the PECs adjusted with the normalization of the loads to the number of inhabitants of the drainage basin at the sampling stations can provide a good approximation of the MECs in all areas, especially for the urban one. This was supported by the comparison of the frequencies of log(PEC/MEC) classification as acceptable, underestimated, and overestimated estimations using our proposed, refined formula and the unrefined one. However, looking at the pattern of the frequencies using the refined formula, PECs substantially differed from the MECs in urban samples, leading to a consistent underestimation of most PhCs (atenolol, carbamazepine, clarithromycin, diclofenac, estradiol, ethinylestradiol, metformin, naproxen, and sulfamethoxazole). To improve the prediction power in urban surface water samples, we can suggest refining the removal factors using specific estimations based on additional/improved degradation tests or using dilution factors according to the studied areas.
Thus, our method represents a further refinement of the method initially proposed by the EMA and improved in [38
] considering the degradation of the compounds in surface water. In the work of [38
], the refinement of PEC, using the removal factors estimated by information from metabolism, excretion, adsorption in sewage sludge, biodegradation in surface water and in sewage treatment plants, and degradation with hydrolysis and/or photolysis [47
], provided a good approximation of the MEC in 40% of the cases for 13 PhCs (e.g., atenolol, erythromycin, clarithromycin, ibuprofen). However, in other cases (60%), PECs substantially differed from the MECs (from one to two orders of magnitude), indicating that the estimate of the environmental fate of the molecule was not reliable enough. Later, the computation was modified and PECs were refined by considering a further value: the removal rate in WWTPs [21
]. Nevertheless, discrepancies between PEC and MEC still occur, and the uncertainty of prediction may derive from the inaccurate evaluation of dilution effect due to potential variability in the receiving water body flow rate [41
]. In our research, we used a default dilution factor of 10 in all surface waters, but this assumption might be inappropriate, as dilution can change quickly both in space and time. Thus, we can hypothesize that a lower dilution factor should be applied for the urban samples to improve the PECs and thus reach acceptable estimations of the MECs.
The differences in the measured-to-predicted ratios for PhCs are probably also ascribable to their differences in environmental behavior along the surface water body, where various chemical and biological processes of degradation are supposed to occur. As an example, some microorganisms are known to be effective in the degradation of diclofenac (e.g., bacteria: Streptomyces
: 17% in 120 h; Actinoplanes
sp. ATCC 53,771 100% in 5 h; fungi: Trametes versicolor
: 100% in few hours; Cunninghamella elegans
: 100% in 120 h), producing oxidative enzymes (in bacteria), such as lignin peroxidase or laccases [82
], which catalyze the oxidation of various aromatic compounds [83
], or nonspecific enzymes (in fungi) [84
]. Thus, in the process of degradation, the environmental behavior and fate of the PhC might well play significant roles that, however, still need to be clarified for many PhCs across different environments. In our refined calculation of PEC, we hypothesized that the degradation of PhCs occurred in surface water receiving effluents from the WWTP or other primary treatment plants. For this reason, we used the same removal factors taken by literature, but, as previously stated, a more accurate estimate should apply the degradation coefficient of the specific depuration system or of natural systems or differently managed agroecosystems [86
Since the environmental assessment evidenced risks classified into three categories, namely low (i.e., atenolol, carbamazepine, erythromycin, metformin, and naproxen), low/medium (i.e., diclofenac and ibuprofen), and high (i.e., clarithromycin, estradiol, estrone, and sulfamethoxazole), in all samples, and the risk increased progressively in rural, periurban, and urban areas, we could use those data to make PhC prioritizations and, consequently, to address correction actions and serve as a guide for future ecotoxicological research [53
]. According to our results, Grill et al. [59
] predicted high ecotoxicological risk for 2 out of 15 PhCs (i.e., ethinylestradiol and azithromycin). Moreover, similarly to our results, diclofenac was classified as having low to high acute toxicity [93
]. However, the methodology (e.g., organisms, acute or chronic endpoints) used for ecotoxicity tests, and thus for the calculating of the PNEC values, may greatly affect the evaluation of the exposure risk [93
]. Finally, since the risk was consistently high in the effluents of the WWTP for most of the PhCs, this result additionally demonstrated the poor removal capacity of WWTPs having primary and secondary levels of treatments.