Hospital wastewater treatment and the role of membrane filtration – removal of micropollutants and pathogens: A review

List of abbreviations and meaning.

Abbreviation

Word/Phrase

HWW

Hospital wastewater

WWTP

Wastewater treatment plant

MBR

Membrane bioreactor

GAC

Granular activated carbon

PAC

Powdered activated carbon

AOP

Advanced oxidation processes

NF

Nanofiltration

RO

Reverse osmosis

CSO

Combined sewer overflows

CAS

Conventional activated sludge

BBR

Biofilm biological reactor

ECR

Entrapped cells reactor

EO

Electrochemical oxidation

MF

Microfiltration

UF

Ultrafiltration

MP

Micropollutant

MO

Microorganism

ARG

Antibiotic resistance gene

SI

Supplementary information

MPN

Most probable number

CFU

Colony-forming unit

LOD

Limit of detection

Wastewater emitted by hospitals contains high concentrations of pharmaceutical compounds (Verlicchi et al. 2010b), pathogens, (multi-)resistant bacteria and their resistance genes (Schwartz & Alexander 2014; Sib et al. 2020). As concentrations of the aforementioned components in hospital wastewater exceed the respective concentrations in municipal wastewater (Verlicchi et al. 2010b; Chonova et al. 2016; Dinh et al. 2017; Exner et al. 2018; Paulus et al. 2019; Verburg et al. 2019), hospitals can be classified as ‘hot-spots’ (Schwartz & Alexander 2014; Hocquet et al. 2016). Ecotoxicological assessments of hospital wastewater (HWW) show higher risks to the aquatic environment and human health than municipal wastewater, due to an additional wide variety of other toxic substances like disinfectants, heavy metals and detergents (Verlicchi 2018). While there are different national regulations regarding the discharge of hospital effluents, the most common way is the discharge of HWW to municipal wastewater treatment plants (WWTP) or to surface water bodies after decentralized wastewater treatment (Exner et al. 2018).

The removal of bacteria, resistance genes, pharmaceutical compounds and their metabolites by conventional WWTPs is often insufficient (Reinthaler et al. 2003; Luo et al. 2014) and the resulting consequences for the aquatic environment and human healthcare can be detrimental. Endocrine disrupting substances can effect aquatic wildlife in low ng/L – μg/L concentrations, cause shifts to sex ratios of fishes (Länge et al. 2001) and amphibians (Pettersson & Berg 2007), alter their behaviors and disrupt the reproductive success (Hoffmann & Kloas 2012). An emerging global concern for human healthcare is posed by antimicrobial resistances, as a modelling analysis by Cassini et al. (2019) estimates that over 33,000 deaths in the European Union in 2015 were caused by infections with antibiotic-resistant bacteria. The discharge of antibiotics, (multi-)resistant bacteria and their respective resistance genes to the aquatic environment further worsens this threat, as selective pressure on bacteria and horizontal gene transfer promotes the formation of antimicrobial resistances in the environment (Adler et al. 2018; Abe et al. 2020). A significant advantage of on-site treatment at hospitals is the reduction of released bacteria and antimicrobial resistance genes to the environment by combined sewer overflows (CSO) (Verlicchi 2018), which represent a major pathway of pathogens into the aquatic environment (Schreiber et al. 2016).

To mitigate the release of micropollutants and pathogens, different advanced wastewater treatment processes are considered. The most common methods can be characterized by their removal mechanism: adsorption, oxidation, biological and physical processes. In adsorption processes, powdered activated carbon (PAC) and granular activated carbon (GAC) are effective adsorbents for micropollutant removal (Worch 2012), in particular for uncharged and apolar substances (Verlicchi et al. 2015). The elimination of micropollutants by oxidation processes is most commonly realized by ozonation, UV irradiation or advances oxidation processes (AOP) and relies on photodegradation and reactions with highly reactive radicals. Biological processes for advanced wastewater treatment show enhanced removal of micropollutants by biodegradation, the most common adaptation are membrane bioreactors (MBR). The removal of micropollutants by physical processes is based on the rejection due to size exclusion or electrostatic interactions, which is foremost accomplished by nanofiltration (NF) and reverse osmosis (RO).

Membrane filtration processes, either as stand-alone treatment steps after pre-treatment (NF, RO) or integrated in hybrid systems (e.g. PAC + Ultrafiltration (UF), AOP + UF, etc.), are investigated in several studies for the treatment of HWW. However, there is no systematic review article regarding the treatment of hospital wastewater by membrane filtration processes. In this review, an overview of published articles regarding the treatment of hospital wastewater and the removal of micropollutants and microorganisms by stand-alone filtration processes and hybrid processes involving membrane filtration steps is presented and discussed.

The literature databases used in this study were chosen by accessibility and the option to search in headlines and abstracts. In this systematic review the following databases were used: dimensions.ai, Scopus, PubMed.gov and ScienceDirect. The search strings used for this review are shown in the Supplementary Information (Table S1). The last update of the database search was conducted on 22nd of March 2022. Additionally, four project reports that are not listed in the chosen databases were included as well. Only English and German publications were eligible for further analysis to guarantee a high-quality assessment of the respective studies and avoid translation related errors. For assessing the search comprehensiveness, the identified records were cross-checked with a general review about hospital wastewater treatment from 2015 (Verlicchi et al. 2015). As a result, all of the database available studies involving stand-alone membrane filtration processes or hybrid membrane filtration processes were found in the respective databases given the selected search strings.

Eligible studies include publications and reports about the treatment of hospital wastewater by membrane filtration methods that are suitable to reject micropollutants and/or pathogens in hospital wastewater. Studies in which municipal wastewater or synthetic wastewater was used were dismissed. Furthermore, only studies with specific information about the removal efficiency of micropollutants and/or pathogens or with enough data to calculate the removal efficiency were considered.

For data extraction, the publications as well as their Supplementary Information (if available) were screened. Removal efficiencies depicted in diagrams were analyzed and extracted by the plot digitization software WebPlotDigitizer (Rohatgi 2021). If elimination efficiencies were published with lower thresholds, further data analysis was conducted with these values. If study results were disclosed in multiple publications, elimination efficiencies were extracted only once to avoid a distortion in the data analysis.

An overview of eligible studies that were evaluated in this review for the removal of micropollutants (MP), pathogenic microorganisms (MO) and antibiotic resistance genes (ARG) is shown in Table 1. Besides information about the location of the clinical facilities whose wastewater is treated, simplified process configurations of treatment steps, the process scale and relevant removal parameters are displayed. Table 1 shows that membrane filtration technologies in hospital wastewater treatment are mainly focused on membrane bioreactors. In general, MBR processes coupled with oxidation processes were the most frequent examined process configuration found for the advanced treatment of HWW in this review. In comparison with activated carbon technologies, oxidation processes offer high levels of disinfection (Sousa et al. 2017), albeit micropollutants are not necessarily mineralized and oftentimes broken down into transformation products (Völker et al. 2019). Investigations on NF and RO for treatment of hospital wastewater are scarce, even though their high removal of pharmaceuticals has been known for a long time (Nghiem et al. 2005). For improved treatment, several studies combine multiple treatment options to enhance micropollutant and/or pathogen removal (Nafo et al. 2012; DHI 2016; Paulus et al. 2019). Overall, pilot scale processes were investigated in most of the studies (18/34), followed by laboratory scale applications (13/34) and full-scale processes (9/34).

Advanced full-scale HWW treatment was evaluated in several projects whose results were published in multiple publications. Full-scale HWW treatment in the two European cities Gelsenkirchen (Germany) and Zwolle (Netherlands) was investigated within the PILLS project (Nafo et al. 2012), in which various HWW treatment technologies were applied to determine appropriate methods for HWW treatment. In the evaluation, the ecotoxicological risk of HWW before and after wastewater treatment was assessed as well. The project also included HWW treatment at hospitals in Esch-sur-Alzette (Luxembourg) and Baden (Switzerland) at pilot-scale level (Kovalova et al. 2012; Nafo et al. 2012; Kovalova et al. 2013).

In Waldbröl (Germany), a district hospital was upgraded with a full-scale MBR with either GAC, ozonation or NF/RO as subsequent treatment steps. Similar to the PILLS project, ecotoxicological effects of HWW were studied before and after treatment (Pinnekamp et al. 2009; Beier 2010; Beier et al. 2010, 2011, 2012).

In Herlev (Denmark), HWW treatment was investigated at a full-scale plant by Grundfos Biobooster A/S, at which two nearly identical process lines were compared (Line 1: MBR + GAC + O₃ + UV, Line 2: MBR + O₃ + GAC + UV). The project also conducted experiments on the ecotoxicity of the HWW and also studied the removal of pathogens by the process lines (DHI 2016).

In Delft (Netherlands), the ‘Pharmafilter’ system was installed for full-scale treatment with a process line comparable to the treatment plant in Herlev (Denmark), consisting of MBR + O₃ + GAC + UV. In addition to micropollutant and pathogen removal, this project also studied the removal of antibiotic resistance genes (ARG) (Batelaan et al. 2013; Paulus et al. 2019).

Wen et al. (2004) investigated a full-scale MBR system for HWW treatment in Beijing (China) and pathogen removal by the MBR was assessed. However, no subsequent treatment steps were installed and removal of micropollutants by the MBR was not studied.

Combining conventional activated sludge (CAS) processes with a porous membrane as a physical separation for suspended solids (SS), membrane bioreactors offer several advantages for hot-spot treatment in comparison to CAS processes. To limit the formation of a fouling layer on the membrane surface, backflushes and air scouring are common techniques in membrane bioreactors with submerged membrane modules. In MBR processes, no clarifiers are needed to remove suspended solids from the effluent of the biological stage, thus offering a more compact wastewater treatment (Arévalo et al. 2009). Because the formation of settable biomass flocs for the sedimentation in clarifiers is not required in MBR processes, hydraulic retention time (HRT) is decoupled from sludge retention time (SRT), allowing operation of MBR processes at higher SS concentrations, which additionally reduces the necessary reactor size for the biological treatment. Moreover, several studies (Clara et al. 2005; Maeng et al. 2013; Achermann et al. 2018) report higher elimination of pharmaceuticals at higher SRT due to more diverse microbiological communities with slow-growing bacteria that improve biotransformation of various micropollutants and higher sorption capacities due to smaller flocs and higher SS concentrations (Verlicchi et al. 2015). In comparison with CAS processes, MBR processes offer higher effluent quality, indicated by lower SS concentrations and lower turbidity (Wisniewski 2007; Park et al. 2015). However, MBR systems are more expensive than CAS processes, as increased biomass concentrations in the MBR demand higher aeration rates for aerobic biotreatment (Judd 2008). Moreover, membrane fouling is considered to be a major drawback of MBR systems, as performance and lifespan of the membranes decrease and costs for membrane replacement and anti-fouling strategies increase (Iorhemen et al. 2016).

Depending on the pore sizes of the membrane, a complete rejection of bacteria (Verlicchi et al. 2015) can be achieved, which makes MBR suitable for a broad range of hybrid systems that can tackle the challenges of micropollutant and pathogen removal altogether. Various studies also report high rejection efficiencies for viruses and bacteriophages for MBR systems with ultrafiltration membranes (Melin et al. 2006; DHI 2016). Many advanced wastewater treatment technologies also benefit from the high quality MBR effluent, as low concentrations of suspended solids are preventing negative interferences with these technologies (Verlicchi et al. 2015).

While most studies investigate extensive removal of various pharmaceuticals by MBR systems, the disinfection in terms of pathogen removal in hospital wastewater treatment is seldom evaluated. In theory, due to their low pore sizes, ultrafiltration membranes are capable of complete pathogen removal. With pore sizes ranging from 0.1 to 1.0 μm in microfiltration membranes and 0.01–0.1 μm in ultrafiltration membranes (Eloffy et al. 2022), ultrafiltration offers complete rejection of protozoa (3–15 μm), bacteria (0.5–2 μm) and viruses (0.02–0.07 μm), while microfiltration completely rejects protozoa and bacteria depending on the membrane pore size (Bodzek et al. 2019).

Table 3 gives an overview of the membrane information and results of the pathogen removal in this literature review. The determinations of bacteria concentrations were conducted by different methodologies (most probable number and colony-forming unit), hence a direct comparison between effluent data is limited (Cho et al. 2010). If concentrations or removal efficiencies in non-logarithmic units were reported, the respective logarithmic removal values were calculated based on average influent and effluent concentrations of the MBR.

Even though complete removal of bacteria by ultrafiltration membranes should be achievable, most studies show bacteria in the effluent above the limit of detection. Pauwels et al. (2006) state that elevated bacterial concentrations after the membrane filtration can be attributed to microbiological regrowth in the effluent vessel. Moreover, defects in the membrane material can lead to further elevated concentrations (DHI 2016). Nevertheless, most studies showed higher pathogen removal efficiencies than conventional wastewater treatment (1–2 log units (van der Drift et al. 1977)). A direct comparison between a CAS and an MBR system treating HWW was conducted by Pauwels et al. (2006) and showed 2 log units of increased bacteria retention in the MBR system. The average removal value of E. coli across all studies is 3.7 log units, which is in line with a literature-based determined log removal value of 3.8 (5% percentile of 62 studies) in MBR treatment of municipal wastewater by Branch et al. (2021). It should be noted that in some studies, the removal efficiency was limited by effluent concentrations below the LOD and therefore an underestimation of the removal efficiency is possible. High removal values of E. coli of about 5 log units were reported by Preecha et al. (2009) and Verlicchi et al. (2010a). Given its high E. coli rejection, the published pore size of the membranes used by Preecha et al. (2009) (0.22 mm) seems unlikely and could be explained by a wrongly reported pore size unit.

Removal of antibiotic resistant genes (ARGs) was reported by Paulus et al. (2019) and is the only study in this review that evaluated ARG reduction. The study found that hospital-related ARG were removed below the LOD in the effluent by a process line of MBR + O₃ + GAC + UV.

Removal efficiencies of viruses in MBR treatment of HWW were not reported in the studies considered for this review, which is surprising given the relevance of hospital wastewater disinfection. For municipal MBR treatment, the statistical literature assessment by Branch et al. (2021) determined a virus removal value of 1.8 log units (5% percentile of 539 studies).

Due to their high effluent quality, membrane bioreactors are ideal pre-treatment processes for further oxidation, as the reduced amount of organic carbon in the effluent increases the efficiency of oxidation processes. While oxidation processes can reach high levels of micropollutant removal, mineralization is not always guaranteed and formation of toxic transformation products can increase the toxicity of the oxidation effluent (Nafo et al. 2012; Wang et al. 2018; Völker et al. 2019). Therefore, Völker et al. (2019) suggests additional post-treatment to reduce the amount of transformation products.

In this review, it was found that several different oxidation processes have been used in combination with MBR for HWW treatment (ozonation (+H₂O₂), UV (+H₂O₂/TiO₂), ClO₂) and the most abundant were ozonation processes which achieved the highest elimination efficiencies. Table 4 gives an overview of various elimination efficiencies by ozone treatment. It should be noted that ozonation is highly dependent on the dose and contact time, which differ from study to study. Hence, these operational conditions are displayed as well. Due to the organic impact on ozonation efficiency, most studies standardize the ozone dose in regard to the dissolved organic carbon (DOC) of the effluent. Nielsen et al. (2013) and Kovalova et al. (2013) also investigated the combination of ozone with additional H₂O₂, however, no improvement over the regular ozonation was determined. Overall, ozonation shows high removal for most of the investigated pharmaceuticals, but contrast media (Iopromide, Iopamidol, Diatrizoate) show low median removal efficiencies of <50%.

Table 4

Removal efficiencies of pharmaceuticals in MBR effluents by ozonation (n > 4)

.

.

The substances are arranged by the median removal efficiency, which is calculated using the removal efficiencies of the individual studies.

View Large

The oxidation of MBR effluent by UV irradiation was assessed in the PILLS project (Nafo et al. 2012). Additional results from the hospital wastewater treatment in Switzerland were published in detail by Kovalova et al. (2013). UV irradiation was not successful at the Swiss pilot plant and showed an average removal efficiency of 33%. This was verified by other investigations of the PILLS project, in which additional oxidation agents were necessary to reach high removal efficiencies of >70% (Nafo et al. 2012). However, even though the average removal efficiency was not as high as ozonation, UV treatment was able to remove contrast media to higher degrees than ozonation or activated carbon processes. In direct comparison with ozonation and activated carbon processes, the PILLS project concluded that UV treatment is more expensive, which further explains the more common adaptation of ozonation processes in hospital wastewater treatment (Nafo et al. 2012).

Further oxidation of MBR effluents with ClO₂ was investigated by Nielsen et al. (2013). However, the treatment showed high fluctuations for micropollutant removal and was overall the least effective process of the investigated treatment options (ozonation (+H₂O₂) and PAC) (Nielsen et al. 2013).

Oxidation processes are commonly used for disinfection of drinking water, thus high removal efficiencies are anticipated. However, pathogen removal of HWW by coupled MBR and oxidation processes was only assessed via ARG concentrations by Paulus et al. (2019). In this study, no consistent ARG removal by ozonation was assessed. An increase in bacterial load between MBR effluent and ozonation effluent was monitored, which was linked to regrowth before ozonation. In comparison, a study that investigated disinfection of urban wastewater by oxidation processes showed removal of facultative pathogenic bacteria and ARG of 2 log units (Hembach et al. 2019).

For adsorptive advanced wastewater treatment, activated carbon processes are the most frequently used technologies and can be categorized based on the particle size of the activated carbon. Granular activated carbon (GAC) (0.5–4 mm (Worch 2012)) is commonly adapted in fixed-bed columns, while powdered activated carbon (PAC) (<0.04 mm (Worch 2012)) is usually either dosed into the biological stage or in an external filtration tank (Gutiérrez et al. 2021). Due to its low particle size, the separation of PAC is more complex and requires the use of coagulants and flocculants and/or a separation by different filtration technologies, like deep-bed filtration, pile cloth filtration or membrane filtration (Krahnstöver & Wintgens 2018).

Due to the different configurations, a direct comparison of PAC and GAC is not suitable, hence the results of the studies in this review are compared based on their adsorption process, which is documented in Table 5.

Table 5

Implementation of activated carbon processes in hybrid configurations for micropollutant removal in hospital wastewater

Study .	Activated carbon (manufacturer, product) .	Process configuration .
Kovalova et al. (2013)	PAC (Norit, SAE Super)
Nafo et al. (2012)	PAC (N/A)
Nielsen et al. (2013)	PAC (Calgon Carbon, Filtrasorb F400 /Norit, 830W PAC)
Alvarino et al. (2020)	PAC (N/A)
Nafo et al. (2012)	GAC (Norit, Row 0.8 Supra)
Mousel et al. (2021)	GAC (N/A)
Pinnekamp et al. (2009)	GAC (CSC, PHC 900) (Beier 2010)

Study .	Activated carbon (manufacturer, product) .	Process configuration .
Kovalova et al. (2013)	PAC (Norit, SAE Super)
Nafo et al. (2012)	PAC (N/A)
Nielsen et al. (2013)	PAC (Calgon Carbon, Filtrasorb F400 /Norit, 830W PAC)
Alvarino et al. (2020)	PAC (N/A)
Nafo et al. (2012)	GAC (Norit, Row 0.8 Supra)
Mousel et al. (2021)	GAC (N/A)
Pinnekamp et al. (2009)	GAC (CSC, PHC 900) (Beier 2010)

View Large

The change of ARGs during GAC treatment was investigated by Paulus et al. (2019) and while some ARGs were significantly reduced, a significant increase of other ARGs was detected as well. Similar results were obtained by other studies in which activated carbon was used for municipal wastewater or drinking water treatment. The abundance of ARGs in the effluent is linked to biofilm growth on the surface of the activated carbon and horizontal gene transfer of ARGs (Bai et al. 2015; Hu et al. 2019; Sun et al. 2020).

The treatment of MBR effluents by nanofiltration and reverse osmosis was investigated by Beier et al. (2010), Nafo et al. (2012) (PILLS project) and Lan et al. (2015). All studies were conducted with spiral-wound modules in pilot scale filtration units and high removal efficiencies were consistently achieved in all studies. However, only Lan et al. (2015) investigated retentate treatment, while Beier et al. (2010) recirculated and discarded the retentate. No further information about retentate treatment was given by Nafo et al. (2012), hence recirculation and discard of retentate is most likely. In nanofiltration experiments of MBR effluent, Beier et al. (2010) report average rejection efficiencies of 95% at a yield of 70% and average rejection efficiencies of 94% at a yield of 90%. Comparably, low differences between removal efficiencies at different yields were obtained for reverse osmosis as well. At 70% yield, an average removal of 98% was achieved, while 90% yield resulted in a marginally higher removal efficiency of 99% (Beier et al. 2010). The results are in line with the other studies, Lan et al. (2015) report the removal of spiked Ciprofloxacin from MBR effluent with >97% at a yield of 80%. The investigations of the PILLS project with two reverse osmosis pilot plants located in Zwolle (Netherlands) and Esch-sur-Alzette (Luxembourg) showed that pharmaceutical concentrations were reduced below the LOD at applied yields of 50–75% (Nafo et al. 2012).

One of the main drawbacks of dense membrane filtration is the retentate treatment, which was investigated by Lan et al. (2015), who applied electrochemical oxidation to the retentate of a spiked nanofiltration experiment. In this study, a complete removal of Ciprofloxacin was monitored after 150 minutes. A study investigating NF brine treatment by other oxidational processes (solar photo-Fenton, ozonation) of municipal secondary effluent showed high removal efficiencies (>90%) as well (Miralles-Cuevas et al. 2016). Economically, Miralles-Cuevas et al. (2016) evaluated that the combination of NF and solar photo-Fenton or ozonation is more favorable than the stand-alone oxidation due to a reduced volume that needs to be treated by oxidation. In a municipal application, RO brine treatment with GAC of a wastewater reclamation plant receiving biologically treated wastewater and stormwater was investigated by Jamil et al. (2020) and nearly complete removal efficiencies (95–100%) for 17 micropollutants were achieved within 2880 BVT treated.

As NF and RO membranes reject smaller molecules than microfiltration and ultrafiltration membranes, these membranes show a complete rejection of microorganisms as well as high rejection efficiencies of extracellular ARG. This was demonstrated by Lan et al. (2019), whose study of membrane filtration investigation for swine wastewater treatment showed an increase in extracellular ARG removal of several log units for NF and RO membrane filtration after MF and UF treatment. In this study, NF/RO treatment was capable of removing 4.98–9.52 log units of ARG when compared to the raw sewage influent. However, none of the eligible studies found in this review process has reported ARG removal in dense membrane filtration of hospital wastewater. Even though similar results of ARG removal are overall anticipated, thorough investigations are needed to determine dissemination of crucial ARG that are linked to last resort antibiotics.

Many disadvantages of advanced wastewater treatment processes can be mitigated by multi-process coupling. For example, the release of toxic transformation products generated by oxidation processes can be minimized by subsequent membrane filtration or adsorption. Furthermore, because activated carbon processes are no barrier for pathogens, dissemination of resistant bacteria can be avoided with subsequent membrane filtration or oxidation processes. Due to the many synergistic effects of combining multiple treatment steps, several pilot scale and full-scale projects have incorporated multiple barriers for micropollutants and pathogens.

A full-scale multistage treatment was realized with a MBR coupled with oxidation and adsorption processes at the hospital in Herlev (Denmark) (DHI 2016). The private-public project was the first full-scale HWW treatment facility in Denmark and was operated by Grundfos BioBooster A/S. Two process lines of MBR + GAC + O₃ + UV (Line 1) and MBR + O₃ + GAC + UV (Line 2) were compared regarding their removal of pharmaceutical compounds as well as their removal of bacteria and viruses. During the operation, line 2 was determined to be the more efficient and 99.9% of pharmaceuticals (without contrast media) and 99% of contrast media were removed after O₃ and GAC treatment when compared to the MBR influent. Concentrations of pathogens were removed below the LOD in the effluent during normal operation, however an incident with a cracked membrane of the MBR system led to a contamination of the permeate with E. coli bacteria.

A similar project with comparable process steps was realized in Delft (Netherlands). The system called Pharmafilter treats solid waste and wastewater generated by the hospital by a combination of digester + MBR + O₃ + GAC + UV (Paulus et al. 2019). During the anaerobic digestion, biogas is being produced, which in turn generates energy for the wastewater treatment (Batelaan et al. 2013). In the study of Paulus et al. (2019), high average removal was achieved for many micropollutants: 27 out of 34 substances were detected below the LOD after the last treatment step and three micropollutants, which were above the LOD in the effluent, showed high removal efficiencies of >97%. Negative to moderate total removal efficiencies were measured for Metronidazole (−75%) and Sulfadiazine (33%). In previous investigations a complete removal of about 100 substances was determined (Batelaan et al. 2013). High removal of ARGs was achieved as well, as nine out of 13 ARGs were reduced below LOD and the remaining four out of 13 ARGs were significantly reduced (Paulus et al. 2019). Overall, the on-site treatment showed higher ARG removal than conventional wastewater treatment.

The European project PILLS investigated the treatment of hospital wastewater at two full-scale plants located in Gelsenkirchen (Germany) and Zwolle (Netherlands) and two pilot scale plants in Baden (Switzerland) and Esch-zur-Alzette (Luxembourg) (Nafo et al. 2012). While process combinations were investigated in parts of the project, only single-stage processes after MBR treatment were evaluated regarding the elimination of micropollutants (Nafo et al. 2012). However, ecotoxicity assessments of different single- and multi-stage treatment steps highlighted the need for process combination. While MBR with subsequent GAC and PAC treatment decreased the toxicity of the raw wastewater, it may still contain algal toxic substances. MBR with subsequent ozonation reduced antibiotic and estrogenic effects of the wastewater, but toxicity increased after oxidation by ozonation and UV treatment, presumably due to formation of toxic by-products. Combining activated carbon processes with oxidation processes can further reduce adverse effects significantly, as shown by a combination of UV irradiation and GAC treatment (Nafo et al. 2012).

While conventional wastewater treatment is not sufficient in terms of micropollutant and pathogen removal from hospital wastewater, a wide variety of studies have investigated different advanced wastewater processes involving different kinds of membrane filtration.

On the biological stage, membrane bioreactors achieve high efficiencies of pathogen removal, but the biodegradation of recalcitrant substances is not always sufficient. Hence, hybrid systems with additional treatment steps are needed for further micropollutant removal. In this review, several approaches have been investigated and while studies with subsequent oxidation processes after MBR treatment are the most abundant, information about pathogen removal and the formation of toxic transformation products is lacking. For micropollutant removal in hospital wastewater treatment, hybrid systems with adsorption or dense membrane filtration technologies were seldom reported, even though their high potential has been proven in municipal applications. For adsorption processes, more investigations on the dissemination of antibiotic resistant bacteria and antibiotic resistance genes are needed to thoroughly assess the requirement for additional disinfection. While nanofiltration and reverse osmosis processes show high capabilities for disinfection and micropollutant removal, no full-scale adaptation has been investigated so far. Furthermore, the problematic retentate treatment was only investigated by one study, which highlights the need for further concentrate treatment investigations in nanofiltration and reverse osmosis applications.

In several studies, high quality effluents were achieved after advanced hospital wastewater treatment and indicate that hospital wastewater treatment could contribute towards a circular water economy. The reuse of water is especially crucial for water-scarce regions and various reuse applications are conceivable to minimize water stress. For example, Czuba et al. (2021) suggest, depending on the secondary effluent treatment, the reuse of treated wastewater for groundwater recharge, agricultural irrigation, as a cooling medium, as potable water or as industrial water.

Overall, this review highlights the necessity of combining multiple treatment processes to mitigate the disadvantages of the respective technologies. However, economic assessments on hybrid processes are scarce and more insight is needed for an extensive comparison between different processes and their combinations regarding the cost efficiency for full-scale adaptations.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.