Environment International 92–93 (2016) 183–188

Contents lists available at ScienceDirect

Environment International

journal homepage: www.elsevier.com/locate/envint

Review article

Removal of trace organic contaminants from domestic wastewater: A

meta-analysis comparison of sewage treatment technologies
Steven D. Melvin ⁎, Frederic D.L. Leusch
Smart Water Research Centre, Australian Rivers Institute, Griffith School of Environment, Griffith University, Southport, QLD, 4222, Australia

a r t i c l e i n f o a b s t r a c t

Article history: Trace organic contaminants (TrOCs), such as endocrine disrupting compounds (EDCs) and pharmaceuticals and
Received 9 February 2016 personal care products (PPCPs), represent global threats to aquatic animals and ecosystems. A major source of
Received in revised form 24 March 2016 TrOCs in the aquatic environment is via the discharge of treated sewage, so there is an urgent need to evaluate
Accepted 24 March 2016
the comparative efficiencies of the most widely used sewage treatment technologies as regards elimination of
Available online 20 April 2016
these compounds from wastewater. To address this need, 976 published articles were compiled focusing on es-
Keywords:
timates of removal (%) for 20 common environmental TrOCs, from five major sewage treatment technologies:
Sewage conventional activated sludge (CAS), oxidation ditch (OD), membrane bioreactor (MBR), ponds and constructed
Removal efficiency wetlands (PCW), and trickling biological filters (TBF). A quantitative meta-analysis was performed to compare
Organic contaminant standardized relative removal efficiencies (SREs) of the compounds amongst these technologies, and where pos-
Endocrine disrupting compound sible potential sources of heterogeneity were considered (e.g., flow rates and chemical sorption potential). The
Pharmaceutical and personal care product results indicate that the most widely used CAS treatment and the less common TBF provide comparatively
Meta-analysis poor overall removal of common organic micropollutants. Membrane bioreactors appear to be capable of achiev-
ing the greatest overall removal efficiencies, but the sustainability and economic viability of this option has been
questioned. Treatment with OD systems may be more economical while still achieving comparatively high re-
moval efficiencies, and the analysis revealed OD to be the best option for targeting highly potent estrogenic
EDCs. This study offers a unique global assessment of TrOC removal via leading sewage treatment technologies,
and is an important step in the identification of effective options for treating municipal sewage.
© 2016 Elsevier Ltd. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
2. Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
3. Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
3.1. Overview of data collection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
3.2. Comparing SREs and controlling for heterogeneity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
4. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188

1. Introduction animals and ecosystem health (Boxall et al., 2012; Brausch and Rand,
2011; Kaplan, 2013; Pal et al., 2010; Stuart et al., 2012). Many represen-
Trace Organic Contaminants (TrOCs), such as endocrine disrupting tative TrOCs have been shown to elicit a range of adverse toxicological
compounds (EDCs) and pharmaceutical and personal care products effects in aquatic wildlife, often at environmentally relevant sub-lethal
(PPCPs), are now widely considered as emerging threats to aquatic concentrations (Godoy et al., 2015; Pereira et al., 2015). The discharge
of treated sewage is a primary source of such contaminants in the
⁎ Corresponding author at: Smart Water Research Centre, Building G51, Griffith
aquatic environment (Jasinska et al., 2015; Kaplan, 2013; la Farré
University, QLD 4222, Australia. et al., 2008). Ensuring the protection of aquatic ecosystems and the
E-mail address: s.melvin@griffith.edu.au (S.D. Melvin). preservation of global biodiversity therefore demands sewage

http://dx.doi.org/10.1016/j.envint.2016.03.031
0160-4120/© 2016 Elsevier Ltd. All rights reserved.
184 S.D. Melvin, F.D.L. Leusch / Environment International 92–93 (2016) 183–188

treatment technologies that not only target traditional parameters, such 3) the primary source of wastewater was domestic sewage, and 4) per-
as nutrients and pathogens, but which also eliminate broad classes of centage removal was reported for clearly identified organic contami-
organic micropollutants (Malaj et al., 2014). In light of the known nant(s). When percentage removal was not directly reported, studies
risks to water quality and ecosystem health, there is currently signifi- were still included provided removal (R) could be calculated based on
cant international interest in the identification of treatment technolo- concentrations at the influent and effluent (Eq. (1)).
gies that are capable of effectively removing TrOCs from sewage
(Gerbersdorf et al., 2015; Jones et al., 2007).
Numerous approaches exist for treating domestic sewage, ranging C Influent –C Effluent
R ð% Þ ¼  100 ð1Þ
from simple systems that rely on natural ecological processes, to opera- C Influent
tionally intensive engineered treatment technologies. While all systems
tend to achieve at least some degree of chemical removal, conventional
wastewater treatment technologies were never designed to remove An extensive dataset was established through the Phase 1 literature
these categories of organic contaminants (Bolong et al., 2009; Ratola survey, and this was subsequently used to refine the scope of the meta-
et al., 2012; Rojas et al., 2013). Advanced treatment technologies such analysis by directing a more focused systematic literature search (Phase
as membrane bioreactors are currently of particular interest, due to 2). The dataset compiled through Phase 1 was sorted by contaminant
thorough removal of chemicals from wastewater (Petrović et al., and filtered by treatment technology, and those compounds where re-
2003). However, the economic sustainability and the overall environ- moval estimates were available for multiple technologies were short-
mental benefit of advanced treatment have been called into question, listed for a further search of the literature using a revised set of search
since these systems consume high amounts of energy which often re- terms. Five main treatment technologies were identified for inclusion:
sults in CO2 production (Jones et al., 2007). Attempts to experimentally conventional activated sludge (CAS), oxidation ditch (OD), membrane
compare chemical removal efficacy amongst leading treatment technol- bioreactor (MBR), ponds and constructed wetlands (PCW), and trick-
ogies have been lacking, as logistical constraints have precluded any ling/biological filters (TBF). All contaminants where removal data was
sort of robust large-scale comparative analysis. The main limitation is identified for at least three of these systems were included for the
that it is simply not feasible to acquire statistically rigorous data from Phase 2 literature search, which was performed using revised search
a wide range of full-scale plants employing different treatment options. terms including a combination of (i) the contaminant name, (ii) each
As a result, comparative studies have generally been restricted to small- of the 5 identified treatment technologies, and (iii) the keywords ‘re-
scale, geographically isolated cases, making it extremely difficult to ex- moval’, ‘elimination’, and ‘treatment’. The selection criteria for including
trapolate results on a broad (global) scale with any sort of confidence. a study were the same as previously described for the Phase 1 literature
Qualitative reviews have provided some insight regarding the re- search. For all papers that met the selection criteria the contaminant of
moval of micropollutants using different wastewater treatment options interest, the treatment technology, and the reported or calculated re-
(Liu et al., 2009; Luo et al., 2014; Onesios et al., 2008). The foremost con- moval efficiency (% reduction) were recorded, and the sorption coeffi-
clusion taken from such reviews is that chemical removal is extremely cient (Kow) for each compound was obtained from the USEPA
variable, and particularly amongst the different classes of organic con- Estimation Program Interface (EPI) Suite™ (USEPA, 2015). The flow
taminants (Gerbersdorf et al., 2015). This variability makes it extremely rate of sewage entering the system (m3/day) was recorded whenever
difficult, and generally inappropriate, to compare raw removal data this information was provided.
(Bond et al., 2003). However, quantitative assessment is necessary to Following the extensive two-phase literature search, all compounds
truly evaluate and compare the efficacy of different treatment technol- with removal estimates available for each of the 5 treatment technolo-
ogies for the removal of TrOCs from wastewater. Meta-analyses have gies were included for analysis. To compare removal efficiencies
proven extremely useful for the quantitative synthesis and interpreta- amongst the specified treatment technologies, the removal data was
tion of large-scale datasets, and have been widely applied in the medi- converted into a Standardized Removal Efficiency (SRE) for each con-
cal, ecological and toxicological sciences. Conventional approaches to taminant independently (Eq. (2)). Where, x represents each individual
meta-analysis have allowed within-study effect sizes to be effectively removal efficiency (% reduction) data point, μ represents the average re-
compared and tested for heterogeneity amongst sets of pre-defined moval efficiency for each compound across all 5 treatment technologies,
treatment conditions (Field and Gillett, 2010). Analogous techniques and σ is the standard deviation of the removal efficiencies across all 5
have also been successfully applied to compare between-study experi- treatment technologies.
mental outcomes (Melvin and Houlahan, 2012; Melvin and Wilson,
2013). In terms of evaluating sewage treatment technologies, the latter xμ
approach offers a mechanism to control for the known variability in re- SRE ¼ ð2Þ
σ
moval amongst different classes of organic contaminants.
The present study describes a systematic review and meta-analysis
aimed at comparing the overall effectiveness of major sewage treat- The average SRE for each treatment option (for each contaminant)
ment technologies for eliminating a range of common and widespread was calculated and analysed for significant differences in the average
emerging organic pollutants. SREs (of all contaminants) using randomizations without replacement.
This approach is well suited for investigating differences in between-
2. Methods study effect sizes (Adams et al., 1997; Gurevitch and Hedges, 1999),
and has been effectively applied to analyse similarly structured datasets
A comprehensive two-phase literature survey was performed to exploring ecological and toxicological processes (Melvin and Houlahan,
identify studies reporting the removal efficiency of TrOCs from waste- 2012; Melvin and Wilson, 2013). The randomizations were performed
water treatment systems, on a global scale. Web of Knowledge™, Sci- using MS Excel (Microsoft, Inc.), by calculating the absolute differences
ence Direct™, and Google Scholar™ were searched using various in the average of all SREs amongst the 5 treatment technologies, and
combinations of the search terms: ‘wastewater’, ‘sewage’, ‘effluent’, then randomly reassigning the data to the different groups and
‘contaminant’, ‘chemical’, ‘pharmaceutical’, ‘endocrine disrupting com- recalculating the differences. 10,000 permutations of the data were per-
pound’, ‘EDC’, ‘PPCP’, ‘removal’, ‘elimination’, and ‘treatment’. The initial formed to determine the probability of randomly observing differences
search (Phase 1) was purposefully broad, the criteria for inclusion sim- in standardized removal amongst re-sampling groups that were greater
ply being studies where 1) a full-scale wastewater treatment system than or equal to those reported for the different treatment technologies,
was evaluated, 2) the major treatment technology was clearly defined, using α = 0.05 as the significance threshold.
 S.D. Melvin, F.D.L. Leusch / Environment International 92–93 (2016) 183–188 185

3. Results and discussion

3.1. Overview of data collection

The initial Phase 1 literature survey identified removal data for a

wide range of TrOCs, including a preponderance of data for EDCs and
PPCPs. However, after sorting and filtering the data, the comprehensive
Phase 2 search identified only 20 contaminants (Table 1) where re-
moval data was available for all 5 major treatment technologies, for a
total of 976 removal efficiency data points (Supplementary Table 1).
The data was dominated by studies investigating CAS treatment
(49.1%), followed by PCW (25.7%), OD (13.2%), MBR (7.2%) and finally
Fig. 1. Average removal (%) of the studied TrOCs amongst the various treatment
TBF (4.8%). Such disproportion in the spread of the data across treat- technologies. Boxplots display interquartile range, median (horizontal line), min and
ment technologies supports the use of permutation tests (randomiza- max (whiskers), and average (+) of the 20 compounds included in the final dataset.
tions) for data analysis (Schulz and Grimes, 2002), and also highlights
a need for research evaluating organic contaminant removal from
other widely used sewage treatment options. It is also worth highlight- was achieved in studies reporting contaminant removal using MBR sys-
ing the variance in average removal (%) of the studied TrOCs within the tems, although this technology also exhibited the greatest heterogene-
various treatment technologies, since this illustrates the need for stan- ity (Fig. 3a). The cause of increased heterogeneity in MBR systems was
dardized meta-analysis approaches to effectively compare treatment ef- not immediately apparent, but closer inspection revealed very poor
ficacy between technologies (Fig. 1). comparative removal of the synthetic musk tonalide using this technol-
The average flow rates reported for CAS treatment systems were ogy (Table 1). The literature search only identified a single removal es-
markedly greater than those reported for all other technologies (Fig. timate for tonalide using MBR treatment (Supplementary Table 1),
2), which likely reflects the comparatively high level of implementation raising the possibility that anomalous outcomes for poorly studied com-
of this technology for treating domestic waste in major metropolitan pounds could introduce bias into the analysis. This is somewhat compa-
areas around the world. Contrarily, PCW systems tend to be most com- rable to the phenomenon known as availability bias, where outcomes in
monly applied to treat smaller populations and are frequently used in meta-analyses are influenced by the selective inclusion of the most
rural and remote communities, and developing countries (Jasper et al., readily accessible literature (Rothstein et al., 2006).
2013; Kivaisi, 2001). As such, flow rates were correspondingly low for Therefore, the SRE values for each compound was weighted based on
PCW systems, whereas the other treatment options exhibited compara- the number of data points available for each treatment technology
ble flow (Fig. 2). (Eq. (3)), where, n is the number of removal estimates obtained for
each specific treatment technology, and N is the total number of data
3.2. Comparing SREs and controlling for heterogeneity points for that contaminant.

Analysis of the full dataset identified a significant overall difference SRE  n

SRE0 ¼ ð3Þ
in removal of organic contaminants achieved amongst the 5 investi- N
gated treatment technologies (Fig. 3a). The results demonstrate that
treatment with TBF achieves the poorest overall removal for EDCs and Analysis of the weighted data found CAS to be less effective than all
PPCPs, followed closely by CAS and PCW systems. Treatment using OD other treatment technologies (Fig. 3b). It is important to recognize that
was found to achieve significantly greater overall removal compared the number of data points available for any given treatment technology
to TBF, but no significant difference was identified between OD and does not necessarily provide a strong indication of the reliability of a
any of the other treatment technologies. The greatest comparative SRE study, and the true comparative removal efficacies therefore likely fall

Table 1
Average weighted Standardized Removal Efficiencies (SRE′) for each of the studied TrOCs amongst the 5 major treatment technologies. Bolded values indicate the treatment technology
with the highest SRE for each compound.

Sorption Compound Log Kow CAS MBR OD PCW TBF

Low Caffeine −0.1 −0.039 0.054 −0.002 −0.011 −0.002

Metoprolol 1.9 −0.239 0.157 −0.017 0.138 −0.038
Sulfamethoxazole 0.9 0.003 0.017 0.001 −0.035 −0.002
Trimethoprim 0.9 −0.226 0.013 0.016 0.168 0.029
Medium 17β-Estradiol 4.0 −0.100 0.063 0.080 −0.050 0.007
Bisphenol A 3.3 −0.028 −0.019 0.070 −0.019 −0.004
Carbamazepine 2.5 −0.011 0.011 −0.021 0.016 0.007
Ethinylestradiol 3.7 −0.071 0.046 0.127 −0.043 −0.060
Estriol 2.5 −0.036 0.043 0.090 −0.087 −0.010
Estrone 3.1 −0.106 0.036 0.052 0.013 0.005
Ibuprofen 3.9 0.087 0.018 0.011 −0.116 0.000
Ketoprofen 3.1 −0.153 0.055 0.013 0.091 −0.006
Naproxen 3.2 −0.034 0.031 0.026 −0.001 −0.030
Propranolol 3.5 −0.059 0.125 0.042 −0.077 −0.030
High Diclofenac 4.5 −0.129 −0.004 −0.033 0.188 −0.021
Galaxolide 5.4 −0.150 −0.050 −0.035 0.216 0.019
Gemfibrozil 4.8 0.090 0.071 0.028 −0.156 −0.033
Nonylphenol 5.8 0.144 0.006 0.028 −0.119 −0.058
Tonalide 5.8 −0.072 −0.080 0.058 0.077 0.017
Triclosan 4.8 −0.052 0.001 −0.071 0.066 0.055
Overall SRE (total) −1.182 0.594 0.461 0.257 −0.156
186 S.D. Melvin, F.D.L. Leusch / Environment International 92–93 (2016) 183–188

treatment systems (Dong et al., 2011). Nevertheless, in light of the pro-

nounced differences in the proportion of data pertaining to the different
treatment technologies (i.e., significantly more studies related to CAS
systems), hydraulic loading rate was controlled for as a potential source
of heterogeneity in the global dataset. Roughly 30% of studies did not re-
port sewage flow rate, making mathematical correction (i.e., weighting
data based on flow rate) a tenuous approach. Since the average flow rate
reported for CAS systems was notably greater than the flow to all other
treatment options (Fig. 2), a subgroup analysis was performed
(Borenstein and Higgins, 2013) using only those CAS studies with re-
ported flow rate less than or equal to 65,000 m3/day. Unfortunately,
there was no mechanism to effectively adjust the lower flow rates com-
monly reported for PCW systems. However, if flow truly represents an
important source of heterogeneity, re-analysis after exclusion of a
Fig. 2. Average reported loading rates to the different treatment systems (m3/day) from
the full global dataset. Dashed line indicates the 65,000 m3/day cut-off for inclusion in large proportion of CAS studies would be expected to yield a dramati-
the related sub-group analysis. cally different outcome.
This was not the case, and the overall results of the subgroup analy-
sis were highly comparable with that observed with the full unweighted
somewhere between the unweighted (Fig. 3a) and weighted (Fig. 3b) global dataset (Fig. 4a). This suggests that differences in flow rates
data. Nevertheless, this clearly demonstrates that 1) the most widely amongst studies are not a major factor influencing comparative SREs
used sewage treatment technology globally appears to provide the amongst the 5 treatment technologies. For consistency, the SREs of the
poorest comparative removal of a range of important TrOCs, and subgroup data were weighted based on the number of available re-
2) more research is necessary to evaluate treatment technologies moval estimates for each treatment option, and this again revealed a
other than CAS. This is of further importance because it illustrates the similar pattern to the full weighted data set (Fig. 4b). To clarify, the
risk of uncertainty that is associated with making conclusions based fact that flow rate was not found to be a major source of heterogeneity
on individual studies, and demonstrates how consideration of a broader in the present study does not mean that this is not an important factor
dataset allows stronger conclusions to be made. influencing chemical removal amongst different approaches to sewage
In meta-analyses, it is important to consider and account for as many treatment. However, the goal of the present meta-analysis is to provide
sources of heterogeneity as possible, as this lends strength when a broad comparison of contaminant removal efficiencies amongst differ-
interpreting results and forming overall conclusions (Higgins and ent types of sewage treatment systems, and it is somewhat outside of
Thompson, 2002). In the present study, the impact of hydraulic flow the scope of the present dataset to explore removal mechanisms.
rate and the lipophilicity (Kow) of the studied compounds were evalu- Another potential source of heterogeneity could relate to differences
ated as potential sources of heterogeneity. A pH dependent octanol- in sorption coefficients (Kow) of the studied compounds. Numerous re-
water partition coefficient (Dow) can often be calculated provided Kow searchers have indicated that sorption is a key factor influencing chem-
and the acid dissociation constant of the compound is known (pKa), ical removal during wastewater treatment (Cirja et al., 2008; Luo et al.,
and this can also serve as a robust indicator of sorption potential (Sup- 2014; Rogers, 1996), although it has also been suggested that sorption
plementary Table 2). However, it was not possible to consider this as a may be far less important than other factors (Andersen et al., 2005).
possible source of heterogeneity, due to incomplete reporting of water While there is some evidence suggesting similar sorption capacity be-
quality characteristics amongst the included studies. tween CAS and MBR treatment systems (Cirja et al., 2008), differences
The differences in flow rates observed amongst the various treat- in Kow were nevertheless explored as a potential source of heterogene-
ment technologies could be expected to have a significant influence ity in the global dataset. This was again achieved by performing sub-
on removal efficiencies, although this may not hold true for all group analysis, this time by splitting the full dataset into three groups

Fig. 3. Standardized Removal Efficiencies (mean ± 95 CI) for a) for the full unweighted Fig. 4. Standardized Removal Efficiencies (mean ± 95 CI) for a) for the subgroup analysis
dataset (SRE), and b) with data weighted by the number of removal estimates available with CAS loading rates b 65.000 m3/day (SRE), and b) with data weighted by the number
for each of the various treatment option (SRE′). of removal estimates for each treatment option (SRE′).
 S.D. Melvin, F.D.L. Leusch / Environment International 92–93 (2016) 183–188 187

representing compounds with low (4 chemicals with Log Kow b 2.5), for high sorption contaminants fits with expectations of improved re-
medium (10 chemicals with Log Kow between 2.5 and 4), and high (6 moval in CAS.
chemicals with Log Kow N 4) sorption potential (Cirja et al., 2008; Luo Interestingly, PCW treatment systems were identified as being the
et al., 2014; Rogers, 1996). Results indicate that sorption potential is in- most effective at removing the largest number of contaminants from
deed an important consideration that can influence the relative removal wastewater (i.e., trimethoprim, ketoprofen, diclofenac, galaxolide,
of TrOCs, as the outcomes of the analysis changed somewhat when con- tonalide, and triclosan). However, simple addition of the SREs for all
sidering different sub-groups. contaminants supports the results of the overall analysis, with MBR
There were several noteworthy variations in comparative SREs when yielding the greatest total SRE value (Table 1). This indicates that al-
compounds with low sorption coefficients (Log Kow b 2.5) were though PCW effectively removed the greatest proportion of the studied
analysed separately. The greatest overall removal was still achieved contaminants, overall differences in removal were negligible in these
using MBR, but variance was much greater than when the full dataset cases. Thus, it must be considered that the analysis was influenced by
was considered as a whole, and particularly so for TBF systems (Fig. the choice of contaminants, which was ultimately influenced by the
5a). Results were very similar between the subgroup analysis of me- availability of published removal estimates for the 5 studied treatment
dium sorption compounds (2.5 ≤ Log Kow ≤ 4) and the full dataset, technologies. From another perspective, it is extremely important to ac-
with the greatest removal achieved through MBR followed by OD, and knowledge that different contaminants will pose different levels of risk
all other options exhibiting similarly low comparative removal efficien- to aquatic animals and ecosystems. For example, steroidal estrogens
cies (Fig. 5b). Finally, a dramatic deviation from the analysis of the full may arguably represent the greatest risk of all TrOCs included in the
dataset was observed when the high sorption potential subgroup (Log analysis, due to their ability to elicit negative outcomes related to repro-
Kow N 4) was analysed separately. Compounds with high sorption coef- ductive health and population fitness in aquatic vertebrates at low ng/L
ficients exhibited highly variable comparative removal efficiencies for concentrations (Kidd et al., 2007). Bearing in mind the greater risk
all treatment technologies (with the exception of CAS where only mod- posed by estrogenic EDCs, it may therefore be relevant to highlight
erate variance was observed), such that the analysis revealed no signif- that OD treatment most effectively removed all of the estrogenic EDCs
icant differences in SRE amongst any of the 5 treatment options (Fig. 5c). considered in the analysis (i.e., 17β-estradiol, bisphenol A,
This is at first somewhat surprising; however, upon further consider- ethinylestradiol, estriol, and estrone). Thus, despite PCW removing the
ation it is reasonable based on what is known about chemical removal greatest number of compounds and MBR exhibiting the greatest overall
via sorption. For example, compounds with high sorption potential comparative removal, OD may be the most effective treatment option
should be effectively removed via conventional treatment, and while for reducing the ultimate ecological risk from TrOCs in sewage
CAS exhibited poor removal when the full dataset was analysed, it wastewater.
was observed to achieve comparatively good removal based on the re-
sults of the high sorption subgroup analysis (Fig. 5c). This does not
change the fact that the overall analysis revealed CAS to exhibit no 4. Conclusions
major benefit for removing the 20 studied TrOCs compared to the
other treatment technologies, but the shift in relative SREs observed Choosing the most effective sewage treatment technology for any
given situation requires careful consideration of various environmental,
economic, and social implications. The present study addresses the
identified need for research assessing the comparative removal efficien-
cies of various leading sewage treatment technologies, but does not
consider economic or social aspects. One important outcome from the
analysis is the observation that CAS treatment systems are ostensibly
the least-effective options for removing EDCs and PPCPs from wastewa-
ter. Considering that CAS systems are also not considered to be cost-
effective, this highlights the need to identify effective solutions to im-
prove global sewage treatment. The results reveal that MBR systems
may offer the highest level of removal of organic contaminants. How-
ever, issues with membrane fouling continue to present difficulties
and raise operational expenditures, and unless practical solutions are
identified the benefits of MBR treatment may not outweigh the costs.
A wide range of chemical characteristics, operational parameters,
and various other factors can influence the removal of organic contam-
inants from sewage. Due to inconsistencies in reporting, it was unfortu-
nately not possible to effectively control for all possible sources of
heterogeneity through the analysis of the data. Nevertheless, several
factors known to influence chemical removal were controlled for
(i.e., flow rate and sorption potential) with no major impact on the
overall conclusions of the study. It can therefore be concluded that
MBR systems exhibit marginally greater overall removal for a range of
contaminants, compared to other leading treatment options. However,
a consideration of construction and operational costs, relative toxicities
of different contaminants, and the induced environmental threat associ-
ated with the high energy consumption of advanced treatment
(e.g., MBR) are all important when evaluating the benefits of different
treatment technologies. While this falls outside the scope of the present
analysis, this study provides a first of its kind comparative assessment of
leading sewage treatment technologies, and is therefore an important
Fig. 5. Standardized Removal Efficiencies (mean ± 95 CI) of TrOCs with a) low (Log contribution to discussions surrounding the identification of effective
Kow b 2.5), b) medium (2.5 ≤ Log Kow ≤ 4) and c) high (Log Kow N 4) sorption potential. options for the elimination of important TrOCs from wastewater.
188 S.D. Melvin, F.D.L. Leusch / Environment International 92–93 (2016) 183–188

Supplementary data to this article can be found online at http://dx. Jasper, J.T., Nguyen, M.T., Jones, Z.L., Ismail, N.S., Sedlak, D.L., Sharp, J.O., Luthy, R.G., Horne,
A.J., Nelson, K.L., 2013. Unit process wetlands for removal of trace organic contami-
doi.org/10.1016/j.envint.2016.03.031. nants and pathogens from municipal wastewater effluents. Environ. Eng. Sci. 30,
1–16. http://dx.doi.org/10.1089/ees.2012.0239.
Acknowledgements Jones, O.A.H., Green, P.G., Voulvoulis, N., Lester, J.N., 2007. Questioning the excessive use
of advanced treatment to remove organic micropollutants from wastewater. Environ.
Sci. Technol. 41, 5085–5089. http://dx.doi.org/10.1021/es0628248.
Funding was provided through a 2016 Griffith University Postdoc- Kaplan, S., 2013. Review: pharmacological pollution in water. Crit. Rev. Environ. Sci.
toral Fellowship (No. 219059) award to S.D.M. and the Queensland Gov- Technol. 43, 1–44. http://dx.doi.org/10.1080/10934529.2011.627036.
Kidd, K.A., Blanchfield, P.J., Mills, K.H., Palace, V.P., Evans, R.E., Lazorchak, J.M., Flick, R.W.,
ernment Science Funding scheme (No. RRP-11/12-13). The funding 2007. Collapse of a fish population after exposure to a synthetic estrogen. Proc. Natl.
source had no role in the design, collection, analysis and interpretation Acad. Sci. 104, 8897–8901. http://dx.doi.org/10.1073/pnas.0609568104.
of data, or decision to submit the article for publication. Kivaisi, A.K., 2001. The potential for constructed wetlands for wastewater treatment and
reuse in developing countries: a review. Ecol. Eng. 16, 545–560.
la Farré, M., Pérez, S., Kantiani, L., Barceló, D., 2008. Fate and toxicity of emerging pollut-
References ants, their metabolites and transformation products in the aquatic environment.
Trends Anal. Chem. 27, 991–1007. http://dx.doi.org/10.1016/j.trac.2008.09.010.
Adams, D.C., Gurevitch, J., Rosenberg, M.S., 1997. Resampling tests for meta-analysis of Liu, Z.-H., Kanjo, Y., Mizutani, S., 2009. Removal mechanisms for endocrine disrupting
ecological data. Ecology 78, 1277–1283. compounds (EDCs) in wastewater treatment — physical means, biodegradation,
Andersen, H.R., Hansen, M., Kjølholt, J., Stuer-Lauridsen, F., Ternes, T., Halling-Sørensen, B., and chemical advanced oxidation: a review. Sci. Total Environ. 407, 731–748.
2005. Assessment of the importance of sorption for steroid estrogens removal during http://dx.doi.org/10.1016/j.scitotenv.2008.08.039.
activated sludge treatment. Chemosphere 61, 139–146. http://dx.doi.org/10.1016/j. Luo, Y., Guo, W., Ngo, H.H., Nghiem, L.D., Hai, F.I., Zhang, J., Liang, S., Wang, X.C., 2014. A
chemosphere.2005.02.088. review on the occurrence of micropollutants in the aquatic environment and their
Bolong, N., Ismail, A.F., Salim, M.R., Matsuura, T., 2009. A review of the effects of emerging fate and removal during wastewater treatment. Sci. Total Environ. 473-474, 619–641.
contaminants in wastewater and options for their removal. Desalination 239, Malaj, E., von der Ohe, P.C., Grote, M., Kuhne, R., Mondy, C.P., Usseglio-Polatera, P., Brack,
229–246. http://dx.doi.org/10.1016/j.desal.2008.03.020. W., Schafer, R.B., 2014. Organic chemicals jeopardize the health of freshwater ecosys-
Bond, C.F.J., Wiitala, W.L., Richard, F.D., 2003. Meta-analysis of raw mean differences. tems on the continental scale. Proc. Natl. Acad. Sci. 111, 9549–9554. http://dx.doi.org/
Psychol. Methods 8, 406–418. http://dx.doi.org/10.1037/1082-989X.8.4.406. 10.1073/pnas.1321082111.
Borenstein, M., Higgins, J.P.T., 2013. Meta-analysis and subgroups. Prev. Sci. 14, 134–143. Melvin, S.D., Houlahan, J.E., 2012. Tadpole mortality varies across experimental venues:
http://dx.doi.org/10.1007/s11121-013-0377-7. do laboratory populations predict responses in nature? Oecologia 169, 861–868.
Boxall, A.B.A., Rudd, M.A., Brooks, B.W., Caldwell, D.J., Choi, K., Hickmann, S., Innes, E., http://dx.doi.org/10.1007/s00442-012-2260-9.
Ostapyk, K., Staveley, J.P., Verslycke, T., Ankley, G.T., Beazley, K.F., Belanger, S.E., Melvin, S.D., Wilson, S.P., 2013. The utility of behavioral studies for aquatic toxicology
Berninger, J.P., Carriquiriborde, P., Coors, A., DeLeo, P.C., Dyer, S.D., Ericson, J.F., testing: a meta-analysis. Chemosphere 93, 2217–2223. http://dx.doi.org/10.1016/j.
Gagné, F., Giesy, J.P., Gouin, T., Hallstrom, L., Karlsson, M.V., Larsson, D.G.J., Lazorchak, chemosphere.2013.07.036.
J.M., Mastrocco, F., McLaughlin, A., McMaster, M.E., Meyerhoff, R.D., Moore, R., Parrott, Onesios, K.M., Yu, J.T., Bouwer, E.J., 2008. Biodegradation and removal of pharmaceuticals
J.L., Snape, J.R., Murray-Smith, R., Servos, M.R., Sibley, P.K., Straub, J.O., Szabo, N.D., and personal care products in treatment systems: a review. Biodegradation 20,
Topp, E., Tetreault, G.R., Trudeau, V.L., Van Der Kraak, G., 2012. Pharmaceuticals and 441–466. http://dx.doi.org/10.1007/s10532-008-9237-8.
personal care products in the environment: what are the big questions? Environ. Pal, A., Gin, K.Y.-H., Lin, A.Y.-C., Reinhard, M., 2010. Impacts of emerging organic contam-
Health Perspect. 120, 1221–1229. http://dx.doi.org/10.1289/ehp.1104477. inants on freshwater resources: review of recent occurrences, sources, fate and ef-
Brausch, J.M., Rand, G.M., 2011. A review of personal care products in the aquatic environ- fects. Sci. Total Environ. 408, 6062–6069. http://dx.doi.org/10.1016/j.scitotenv.2010.
ment: environmental concentrations and toxicity. Chemosphere 82, 1518–1532. 09.026.
http://dx.doi.org/10.1016/j.chemosphere.2010.11.018. Pereira, L.C., de Souza, A.O., Bernardes, M.F.F., Pazin, M., Tasso, M.J., Pereira, P.H., Dorta, D.J.,
Cirja, M., Ivashechkin, P., Schäffer, A., Corvini, P.F.X., 2008. Factors affecting the removal of 2015. A perspective on the potential risks of emerging contaminants to human and
organic micropollutants from wastewater in conventional treatment plants (CTP) environmental health. Environ. Sci. Pollut. Res. 22, 13800–13823. http://dx.doi.org/
and membrane bioreactors (MBR). Reviews in Environmental Science and Bio/Tech- 10.1007/s11356-015-4896-6.
nology 7, 61–78. http://dx.doi.org/10.1007/s11157-007-9121-8. Petrović, M., Gonzalez, S., Barceló, D., 2003. Analysis and removal of emerging contami-
Dong, Y., Wiliński, P.R., Dzakpasu, M., Scholz, M., 2011. Impact of hydraulic loading rate nants in wastewater and drinking water. Trends Anal. Chem. 22, 685–696. http://
and season on water contaminant reductions within integrated constructed wet- dx.doi.org/10.1016/j.ejca.2003.09.021.
lands. Wetlands 31, 499–509. http://dx.doi.org/10.1007/s13157-011-0176-5. Ratola, N., Cincinelli, A., Alves, A., Katsoyiannis, A., 2012. Occurrence of organic
Field, A.P., Gillett, R., 2010. How to do a meta-analysis. Br. J. Math. Stat. Psychol. 63, microcontaminants in the wastewater treatment process. A mini review. Journal of
665–694. Hazardous Materials 239-240, 1–18. http://dx.doi.org/10.1016/j.jhazmat.2012.05.
Gerbersdorf, S.U., Cimatoribus, C., Class, H., Engesser, K.-H., Helbich, S., Hollert, H., Lange, 040.
C., Kranert, M., Metzger, J., Nowak, W., Seiler, T.-B., Steger, K., Steinmetz, H., Rogers, H.R., 1996. Sources, behaviour and fate of organic contaminants during sewage
Wieprecht, S., 2015. Anthropogenic Trace Compounds (ATCs) in aquatic habitats — treatment and in sewage sludges. Sci. Total Environ. 185, 3–26.
research needs on sources, fate, detection and toxicity to ensure timely elimination Rojas, M.R., Leung, C., Bonk, F., Zhu, Y., Edwards, L., Arnold, R.G., Sáez, A.E., Klecka, G., 2013.
strategies and risk management. Environ. Int. 79, 85–105. http://dx.doi.org/10. Assessment of the effectiveness of secondary wastewater treatment technologies to
1016/j.envint.2015.03.011. remove trace chemicals of emerging concerns. Crit. Rev. Environ. Sci. Technol. 43,
Godoy, A.A., Kummrow, F., Pamplin, P.A.Z., 2015. Occurrence, ecotoxicological effects and 1–36. http://dx.doi.org/10.1080/10643389.2011.644221.
risk assessment of antihypertensive pharmaceutical residues in the aquatic environ- Rothstein, H.R., Sutton, A.J., Borenstein, M. (Eds.), 2006. Publication Bias in Meta-analysis.
ment — a review. Chemosphere 138, 281–291. http://dx.doi.org/10.1016/j. John Wiley & Sons, Ltd, Chichester, UK http://dx.doi.org/10.1002/0470870168.
chemosphere.2015.06.024. Schulz, K.F., Grimes, D.A., 2002. Unequal group sizes in randomised trials: guarding
Gurevitch, J., Hedges, L.V., 1999. Statistical issues in ecological meta-analyses. Ecology 80, against guessing. Lancet 359, 966–970. http://dx.doi.org/10.1016/S0140-6736(02)
1142–1149. 08029-7.
Higgins, J.P.T., Thompson, S.G., 2002. Quantifying heterogeneity in a meta-analysis. Stat. Stuart, M., Lapworth, D., Crane, E., Hart, A., 2012. Review of risk from potential emerging
Med. 21, 1539–1558. http://dx.doi.org/10.1002/sim.1186. contaminants in UK groundwater. Sci. Total Environ. 416, 1–21. http://dx.doi.org/10.
Jasinska, E.J., Goss, G.G., Gillis, P.L., Van Der Kraak, G.J., Matsumoto, J., de Souza Machado, 1016/j.scitotenv.2011.11.072.
A.A., Giacomin, M., Moon, T.W., Massarsky, A., Gagné, F., Servos, M.R., Wilson, J., USEPA, 2015. Estimation Programs Interface Suite™ for Microsoft® Windows, v4.11.
Sultana, T., Metcalfe, C.D., 2015. Assessment of biomarkers for contaminants of
emerging concern on aquatic organisms downstream of a municipal wastewater dis-
charge. Sci. Total Environ. 530-531, 140–153. http://dx.doi.org/10.1016/j.scitotenv.
2015.05.080.