UvA-DARE (Digital Academic Repository) Wastewater-based epidemiology for illicit drugs A critical review on global data

Illicit drug use is complex, hidden and often highly stigmatized behaviour, which brings a vast challenge for drug surveillance systems. Drug consumption can be estimated by measuring human excretion products in untreated wastewater, known as wastewater-based epidemiology (WBE). Over the last decade, the application of wastewater-based epidemiology to monitor illicit drug loads increased and WBE is currently applied on a global scale. Studies from over the globe are evaluated with regard to their sampling method, analytical accuracy and consumption calculation, aiming to further reduce relevant uncertainties in order to make reliable comparisons on a global level. Only a limited number is identified as high-quality studies, so further standardization of the WBE approach for illicit drugs is desired especially with regard to the sampling methodology. Only a fraction of the reviewed papers explicitly reports uncertainty ranges for their consumption data. Studies which had the highest reliability are recently published, indicating an improvement in reporting WBE data. Until now, WBE has not been used in large parts of Africa, nor in the Middle East and Russia. An overview of consumption data across the continents on commonly studied drugs (cocaine, MDMA, amphetamine and methamphetamine) is provided. Overall, high consumption rates are confirmed in the US, especially for cocaine and methamphetamine, while relatively low illicit drug consumption is reported in Asia.


Introduction
Illicit drug use is complex, hidden and often highly stigmatized behaviour, which brings a vast challenge for drug surveillance systems.The traditional way of monitoring illicit drug use is through population surveys, consumer interviews and police seizures.These methods yield qualitative data and provide useful information such as age, sex, behaviour and the mode of use (ingestion, injection or inhalation).However, due to several limitations such as self-report bias and limited population coverage, survey-based results have difficulties in reflecting detailed spatial and temporal differences of illicit drug use.To get access to quantitative data, the human metabolic excretion products resulting from drug consumption can be measured in untreated wastewater.The measured parent compound or metabolite can be used to calculate the daily mass loads entering a wastewater treatment plant (WWTP) for these compounds.With the knowledge of human drug metabolism and excretion rates, together with in-sewer transformation rates, the relation to consumption (g/day) can be calculated from these mass loads.When dividing this by the served population at the point of measurement in the sewage system, the amount of illicit drugs consumed in g/day/inhabitant can be estimated allowing for comparisons between locations (van Nuijs et al., 2020).While this approach provides no knowledge about the behaviour of single users, it is useful for identifying the spectrum of substance use for a population serving a particular WWTP.This approach provides near real-time information for a specific city or region and gives insight about temporal trends.It is also a promising tool to estimate the illicit drug market size.Together with survey-based monitoring, wastewater-based epidemiology (WBE) can evaluate drugs abuse more comprehensively and potentially support governments in developing policies to scale down drug abuse (Lai et al., 2016).
Using WBE to detect illicit drugs was first introduced by Daughton et al. in 2001(Daughton, 2001) and first applied by Zuccato et al. (Zuccato et al., 2005) in 2005 to monitor cocaine use and track trends in local illicit drug use in Italy.In 2012 the first international assessment in 19 European cities was published by Thomas et al. (Thomas et al., 2012).A paper discussing the uncertainties within the WBE approach has been published in 2013 by Castiglioni et al. (Castiglioni et al., 2013) and a global overview of results was presented by Feng et al. (Feng et al., 2018) covering publications between 2005 and 2017.WBE has shown to be an innovative and promising tool, and sophisticated methodologies have been published for measuring a great variety of illicit drugs (Choi et al., 2018;F. Hernández et al., 2019;F. Hernández et al., 2018) Despite the fact that WBE provides an evidence-based estimation on illicit drug use for a specific population, it brings a number of uncertainties related to the sampling methodology, the analytical procedure and the consumption calculation (Castiglioni et al., 2013;Zheng et al., 2019).One of the main challenges in WBE is reducing the uncertainty for each of these variables in order to make respectable comparisons on an international level.
Synchronising wastewater sampling and analysis will allow more accurate data (O'Brien et al., 2019).The European network of sewage analysis, Sewage analysis CORe group Europe (SCORE) coordinates and ensures quality control in international WBE studies.This way, research can be done with the same validated methods, which makes results robust and comparable.Data on illicit drugs collected under SCORE supervision is therefore used annually by the European Monitoring center for Drug and Drug Addiction (EMCDDA) (González- Mariño et al., 2020).Wastewater is an important source of public health information and can be applied in many fields of expertise, like for diet and nutrition, pesticide use, public health and doping, and currently it is used to monitor the SARS-CoV-2 virus (Bivins et al., 2020).
The present paper first investigates the potential and reliability of WBE methodologies by critically looking at the literature published in the last decade.The key areas that are evaluated to determine the reliability are; the sampling methodology, the analytical accuracy and the consumption estimation, and overall reliability scores are considered.Then, an overview of consumption data on the most commonly studied drugs (cocaine, MDMA, amphetamine and methamphetamine) is provided, evaluating spatial trends.European-wide studies are frequently published and discussed (Castrignanò et al., 2018;Krizman-matasic et al., 2019;Ort et al., 2014;Thomas et al., 2012), but here we provide the first overview on a global scale, based on a large number of data from published literature on intercontinental level.Finally, suggestions to further improve future WBE studies are provided.

Literature search
One hundred and ten studies reporting on wastewater-based epidemiology of illicit drugs were extracted from the Scopus database and reviewed systematically (Haddaway et al., 2017(Haddaway et al., , 2015;;Xiao and Watson, 2019).Used search strings were 'wastewater-based epidemiology' or 'sewage epidemiology' and 'illicit drug' or 'drugs of abuse' in studies published between 2010 and 2020.Export data were filtered based on relevance; reports that did not provide new data, concerned only new psychoactive substances (NPS), did not calculate the drug consumption, or were not representative for the community, were removed (S.I.1).Searches were performed until March 2020.Fifty-two papers are included and reviewed on the estimated drug consumption and the overall reliability of the methodology.

Assessment of method reliability
The reliability of the selected papers was assessed based on factors introducing uncertainties as described by (Ort et al., 2010a;van Nuijs et al., 2018;Feng et al., 2018;Castiglioni et al., 2013) and (Fernández de Córdova and Medina, 2014).The studies are reviewed with regard to their sampling uncertainty, analytical accuracy and the consumption estimation.Each of the three main criteria was assessed based on specific sub-criteria (Table 1).
For each sub-criterion, a value of 2, 1 or 0 is assigned, which indicates the reliability of each aspect with a maximum of 6 for the sampling uncertainty, a maximum of 10 for the analytical accuracy and a maximum of 6 for the consumption estimation.The overall reliability score per study has a maximum of 22.When no information is reported in the reviewed literature for a specific criterion, '0' will always be assigned.Table 2 lists the cut-off per sub-criterion, which is further explained and justified per sub-criterion in the sections below.

Comparison of international drug consumption
The selected fifty-two papers were reviewed on drug consumption.Median consumptions of cocaine, amphetamine, methamphetamine, and MDMA were used to compare consumption on an intercontinental scale.

Reliability of methods
In this study, the reliability of literature on WBE and illicit drugs published in the last decade is critically reviewed with regard to the sampling uncertainty, analytical accuracy and consumption estimations.Over all 52 papers that are evaluated, the mean score of the studies is 14.3 out of a maximum of 22, ranging from 10 to 18 (Table 3).According to the criteria used in this study, especially with regard to the sampling and its description there are major possibilities for further improvement.Fig. 1 shows the percentages of studies with a positive score per subcriterion, further details on assigned points per sub-criterion per study can be found in S.I.2.Sampling can be a dominant source of uncertainty for the determination of illicit drugs in wastewater as both short-term variations such as diurnal patterns in consumption and human behaviour, festivals or rainfall, and long-term variations such as week-weekend and seasonal changes do occur (Ort et al., 2010(Ort et al., , 2010a)).Extreme weather changes and human behaviour will change the daily flow of wastewater to a WWTP and therefore will cause variability and dilution (Banta-Green et al., 2016;Ort et al., 2010Ort et al., , 2010a)).The concentration of the analyte is also highly dependant on the season, due to wet and dry weather conditions (Loraine and Pettigrove, 2006).However, these sporadically extreme weather (short-term) events do not give the same uncertainty as (long-term) seasonal changes, which are more constant.

3
The concentration of illicit drugs can increase significantly during special events and during the weekends (Bijlsma et al., 2014;Foppe et al., 2018).This brings variability in human behaviour when it comes to drug consumption and will bias the overall consumption of a certain community.A similar bias applies to work-commuters and tourists who only take part a community for a certain amount of time.Therefore, these are important factors regarding the uncertainty (Foppe et al., 2018).

Table 1
Main criteria including sub-criteria to assess the quality of WBE papers.Subcriteria that have a larger weight in the scoring system are underlined.In general, long-term variations are commonly reported (68%) whereas short term variations are reported in only 39% of the papers.This was expected because long-term variations are easier to track since these are more constant factors.To give a reliable estimate of illicit drug use by the local community, preferably measurements should be performed during a 'normal week'.

Mode of sampling.
The systematic sampling error depends on the sampling mode.The use of 24-hour composite sampling is most common in WBE.Composite sampling has a clear advantage over grab sampling with respect to local heterogeneity (Minkkinen and Esbensen, 2009).As influent flow and composition may vary significantly over the timespan of sampling, composite sampling can be challenging (Feng et al., 2018;Ort et al., 2010b).Composite sampling can be performed by three approaches, i.e., flow-proportional, volume-proportional, or time-proportional.Flow-proportional composite sampling takes a subsample volume proportional to the flow in the sewer at a constant time interval, subsamples are weighted individually to form a composite sample.Volume-proportional sampling takes samples more frequently during higher flows and less frequently during lower flows, the sampling volume remains constant.However, this cannot provide a true average concentration since only the frequency changes and individual samples are not weighted properly according to the flow in the sewer.In time proportional sampling both frequency and sampling volume are constant and therefore does not provide a true average concentration as well.The flow-proportional mode of sampling is regarded as most reliable choice for data on illicit drug consumption and is favoured over time-and volume proportional sampling because subsamples are weighted individually.(Ort et al., 2010(Ort et al., , 2010a)).
The majority of reviewed papers use 24 h composite influent samples, while several studies used grab samples (Banta-Green et al., 2016;Feng et al., 2018;Fernández de Córdova and Medina, 2014;Ort et al., 2010Ort et al., , 2010a;;van Nuijs et al., 2018).Whether the composite samples are volume-proportional, flow-proportional or time-proportional is not always reported.When a detailed description of the sampling methodology is specified, flow-proportional is most prevalent.In total, only 21% of the papers specify the use of flow-proportional composite sampling.Thus, lack of (description of) use of flow-proportional composite sampling importantly contributes to sampling uncertainty.

Sample volume.
Low sample volumes reduce the power of a study and will increase error (Einax, 2004;Koelmans et al., 2019).Detection limits benefit from large sample volumes as extracts can reach high concentration factors.Because of the relatively low concentration of certain excreted illicit drugs in wastewater.Most common illicit drugs (cocaine, amphetamines) will be present in wastewater with relatively high concentrations but for NPS the concentration can be much lower, and therefore a larger sample volume is required.A total sample volume of 1 L for influent is considered sufficient (Ort et al., 2010).Subsamples can be taken for sample preparation and chemical analysis.
More than half (58%) of the studies do not report on sample size or do not meet the requirement of collecting 1 L influent.Publishing full information regarding detailed sampling methodologies is crucial to interpret the obtained results.Together with the long-and short-term variations, the sample size was often not reported and therefore a factor in reducing the sampling uncertainty of these methods.

Identification technique.
Analytical uncertainty is commonly well specified in WBE studies (Ort et al., 2010a).The fact that most illicit drugs are of medium to high polarity makes LC-MS/MS the most attractive method for their identification and quantification.All papers report a sophisticated analytical method; analytical technique is described in 100% of the studies, and in 98% of the studies the technique of choice is liquid chromatography coupled to mass spectrometry (LC-MS) or tandem mass spectrometry (LC-MS/MS).One exception is analysing methamphetamine using gas chromatography coupled to mass spectrometry as the main analytical tool (GC-MS/MS) (Wang et al., 2019).LC-MS/MS is commonly used to investigate pharmaceuticals and

Table 2
Cut-off for assigning '1' or '0' for each sub-criterion, when no information is reported '0' will be assigned as indicated with *.  their metabolites because of the robustness and excellent reproducibility (van Nuijs et al., 2011;Zheng et al., 2019), and therefore also a favoured technique to measure concentrations of illicit drugs.Triple quadrupole mass analysers are used most and considered the reference technique to quantify illicit drugs and their metabolites (Hernández et al., 2018).This technique is both sensitive and suitable for quantitative analysis, and selective for identification, and an excellent tool for targeted analysis.New methodologies are expected to increase in the coming years, e.g., based on mixed-mode liquid chromatography-tandem mass spectrometry (Koelmans et al., 2019), or high-resolution mass spectrometers such as orbitrap and time-of-flight (QTOF) analysers in combination with ultra-high-performance chromatography (UHPLC), which are more suitable for untargeted/suspect screening.Combined, these techniques will enhance the sensitivity of the overall method (Baz-lomba, Reid, and Thomas, 2016).
3.1.2.2.Analytical conditions.Specification of the analytical conditions is crucial to make a reliable estimation of the accuracy.The polar/ionic analytes require specific chromatographic separation.Separation modes are usually reversed phase (RP) chromatography and hydrophilic interaction liquid chromatography (HILIC).In RP, a hydrophobic stationary phase is used for retention and a mobile phase consisting of a mixture of organic modifier and a water phase for elution.In HILIC, the analytical column is polar combined with a highly organic mobile phase in which water is introduced as the eluting solvent.The most commonly reported detection technique is mass spectrometry (MS) with electrospray ionisation (ESI).All studies report the analytical methodology used, either in the report, supplementary data or referring to previous work.Analytical conditions are reported in more detail compared to the sampling methodologies.All studies provided a detailed description of the analytical conditions.This can be related to quality control protocols which are well-developed (Krizman-matasic et al., 2019) and the fact that the analytical conditions are easier to control for researches.

Method validation (Internal standards, replicates, positive-and negative controls).
Prior to application, the analytical methodology needs to be completely validated for all analytes in terms of trueness/ accuracy (evaluated by recovery experiments) and precision (as repeatability), selectivity/specificity, and limits of detection (LOD) and quantification (LOQ) (Hernández et al., 2018).A drawback in this field is the lack of guidelines specifically directed toward the analysis of illicit drugs and their metabolites in wastewater (Hernández et al., 2018).
In ESI, ion suppression is a major issue due to the co-eluting matrix constituents of raw wastewater (van Nuijs et al., 2011;Zheng et al., 2019) resulting in a decrease in sensitivity and unacceptable errors in quantification.Matrix effects can be corrected using isotopic labelled internal standards (ILIS), which are commercially available for most common illicit drugs and their metabolites (van Nuijs et al., 2011;Zheng et al., 2019).During method validation, there is an absolute need to thoroughly evaluate if the labelled IS accurately corrects for matrix effects (Hernández et al., 2018).Indeed, 98% of all studies report the use of isotopic labelled internal standards.
When two or more samples from the same sampling location and time carried through identical analytical steps are investigated, the variance of the total method can be assessed.To examine the reproducibility of the analytical method including a standard deviation, a minimum number of three samples is required.Only 35% of the reviewed studies reported on analysing in triplicates.
Positive and negative control samples are commonly used in analytical chemistry and described in almost all quality control protocols.To verify quantitative recoveries during sample collection and preparation and the analytical measurement procedure, representative positive control samples should be analysed.To correct an analytical method for contaminations during sample collection, preparation and the analytical identification procedure, blanks should be analysed.
54% of the reviewed studies reported on positive controls, whilst the use of blanks was reported for 44% papers.The low percentages in these categories might be explained by poor reporting.
3.1.2.3.LOD/LOQ.The limit of detection (LOD) and limit of quantification (LOQ) indicate the capability to reliably and sensitively detect and/or quantify an analyse.Reckoning of LODs and LOQs is usually performed based on a signal-to-noise (s/n) ratio of 1:3 and 1:10, respectively (Hernández et al., 2018).Estimating LODs and LOQs in wastewater is difficult regarding the notable variations in chemical composition (Hernández et al., 2018).For a reliable analysis, the analyte should be at least 10 times the LOD or LOQ for a sufficient sensitivity.In recent years, the low LOD and LOQs reported are mostly due to modern highly sensitive techniques that are, as mentioned in previous paragraphs, used in all reviewed studies.81% of the studies reviewed report LOD and LOQ values that are sufficient to analyse illicit drugs and their metabolites.

Consumption estimation
The relation between known amounts consumed by the population and the consumption as estimated from concentrations of drug residues and metabolites measured in wastewater is well described for a number of pharmaceuticals (Heberer and Feldmann, 2005;Lindberg et al., 2007;Zuccato et al., 2008).The reliability of the consumption estimate can be evaluated when the original measured concentrations of the drug and/or metabolite in the influent are reported.Comparing data amongst studies using different calculation methods based on other approaches or equations is then feasible.For 58% of the reviewed papers, the measured influent concentrations are reported.Other important aspects to reliably estimate consumption are the stability of the molecules in wastewater, the measurement of parent compounds and/or metabolites, the population estimated being served by a WWTP and the exact equation used for the calculation of consumption.

Consumption calculation & metabolites.
The reliability of calculation of drug use depends on the selection of an appropriate drug target residue (DTR), which can be either the illicit drug itself or one of its metabolites.Due to the stability of the drugs in wastewater and human metabolism/excretion, it can be necessary to base consumption estimates upon metabolites.For example, the parent compound of cocaine is excreted in urine only for 1-9% of the original intake, while benzoylecgonine and ecgoninemethylester are excreted in urine for 35-54% and 32-49% of the parent compound intake (van Nuijs et al., 2011).The first formulas for the back calculation of 1) cocaine and 2) amphetamine-like stimulants were proposed by Zuccato et al. in 2008 based on the main excretion products.This could be summarized per DTR specifically as followed: Where, the excretion rate of the DTR is following the use of an illicit drug of interest; stability of DTR is the ratio of DTR concentration after in-sewer losses due to transformation and adsorption to biofilm/sewer sediment, and suspended solids to initial concentration.Biofilm and sewer sediment grow on sewer walls and can potentially change biomarkers due to chemical and microbial processes.This adds to the uncertainty of the back-estimation (Li et al., 2020).MW Drug is the molecular weight of the drug of interest and MW DTR is the molecular weight of the DTR.However, accurate excretion rates of DTRs are not always available, due to the lack of pharmacokinetic data (Feng et al., 2018;Gracia-Lor et al., 2016), or are not determined for the applied administration route (snoring, ingestion or intravenous injection) or dose.

Consumption
The equation used for the consumption estimation is in 16% of the reviewed studies not explained at all.From concentrations of the metabolite (ng/L), the flow rate (L/day) of the wastewater stream and a correction factor, parent compound loads (g/day) can be calculated.The correction factor is established based on the excretion rates and degradation in waste water for the parent drug and its metabolite ( van Nuijs et al., 2011;Zuccato et al., 2008Zuccato et al., , 2005)).

3.1.3.2.
Stability.An important factor to estimate consumption is the stability of the excretion products in the sewer system (Daughton, 2001;Zuccato et al., 2008) (Zheng et al., 2019).Ignoring losses due to sorption/degradation of illicit drugs or the excreted metabolite in the sewer, and presumably microbial degradation of cocaine which results in the formation of BE may lead to a significant underestimation (A.K. Mccall et al., 2016;Thai et al., 2014;Zheng et al., 2019).Methamphetamine and MDMA have a relatively high stability (A.Mccall et al., 2016).Cocaine is unstable in wastewater, the degradation of cocaine could lead to benzoylecgonine, so by using benzoylecgonine as a DTR cocaine consumption could be overestimated.However, only 37% of the reviewed studies discuss sewer stability of the DTR and incorporated this in the consumption calculation.

Population estimation. An estimation of the population being
served by a WWTP is a critical factor in the back calculation of drug consumption (Feng et al., 2018).Drug loads have to be normalized to the population in order to enable comparison of the data between different WWTPs.An accurate population estimate is challenging because the actual number of persons being served by a particulate WWTP might vary in time due to commuting, seasonal variability, holidays, and/ or special events (Feng et al., 2018;Lai et al., 2013).The determination of biological oxygen demand, chemical oxygen demand, total phosphorus and nitrogen in wastewater or the measurement of a stable indicator of human metabolism have been proposed for the estimation of population served by the WWTP (van Nuijs et al., 2011;Castiglioni et al., 2013).Many of these parameters reflect human activities rather than the population size (Feng et al., 2018).Other sources such as industrial and commercial discharges may cause significant variability (Feng et al., 2018).Some studies based consumption as measured at a WWTP on the total population of the country, which is only applicable for countries with a small population like The Maldives (Fallati et al., 2020).In total, fifteen papers reported a rational estimation of the population, based on the amount of people serving a specific WWTP (Baker et al., 2012;Been et al., 2015;Bijlsma et al., 2016;Causanilles et al., 2017;Croft et al., 2020;Devault et al., 2014a;Klupczynska et al., 2016;Krizman et al., 2016;Kumar et al., 2019;Mercan et al., 2019;Reid and Thomas, 2014;Rodríguez-álvarez et al., 2015;Shao et al., 2020;X. X. Zhang et al., 2019;Y. Y. Zhang et al., 2019).

Critical perspective on the scoring system
The systematic scoring system approach described in this paper gives a general overview of the three main sources of uncertainty in WBE and how the literature from the last decade was dealing with these uncertainties.However, even though the criteria should be as independent as possible, some criteria might cross-affect each other.The scoring system does not correct for this issue.An example is that the sampling procedure (accurate measurement of the flow rate) affects the backcalculation, and the mode of sampling affects the sample volume uncertainty.
Another issue that is not recognized by the scoring system is the coconsumption of drugs in an area.This can affect universal correction factors based on urinary extraction.
For extreme weather events are less likely to occur in certain areas, this might explain why some papers lack information on the weather during sampling.Another potential cause of uncertainty is leakage or overflowing of the sewers since this could lead to underreporting.The scoring system did not take this into account.Another limitation of the score system is that no distinguishing is made between analyte-ILIS and non-analyte-ILIS, which is of great relevance for matrix effects correction and accurate quantification.
Some papers intentionally only show temporal trends and thus are not as dependant on the uncertainties as other reports.The scoring system does not correct for this, and these papers will most likely score lower due to underreporting, even though these uncertainties are not as applicable.

Overall reliability of methods
For each of the reviewed studies, quality criteria and the overall reliability is assessed (Table 3).None of the reviewed papers met the maximum achievable score of 22, one study had the highest score of 18 (Foppe et al., 2018) and two studies had the second-best score of 17 (Mastroianni et al., 2017;Skees et al., 2018).These studies were published in 2017 or later, indicating an improvement in the reporting on WBE approaches.
The average scores are 2.6/6 for the sampling method, 7.9/10 for the analytical accuracy and 3.8/6 for the consumption estimation, and 14.3/22 overall.Thus sampling uncertainty scored the poorest, in agreement with conclusions stated by Ort et al. (Ort et al., 2010a).There is inherent sampling variability due to weather variation, sporadic events, and fluctuating flowrates.Still, most studies fail to report sufficient details with respect to the sampling method.Sampling should be based on current knowledge, and not just be based on the available equipment (Ort et al., 2010).The analytical accuracy gave the best score.The analytical method can be controlled by researchers, whereas sample collection usually involves voluntary cooperation of treatment plant staff.For this reason, it was expected that this type of uncertainty would be reported in more detail and result in the best score.

International Drug consumption
In the last decade, the application of wastewater-based epidemiology to monitor illicit drug loads quickly increased, and the approach is currently applied on a global scale.Data on consumption of illicit drugs and their metabolites is reported for each continent except Antarctica.The variability of illicit drug consumption across the globe gives insight in substance abuse and can potentially assist health and law enforcement agencies to register and cut down drug related incidents and substance abuse (Lai et al., 2016).An overview of drug consumption per country from literature published over the last decade is shown in Table 4 for the four substances that are investigated most often.These studies evaluated illicit drug use patterns in urban and rural areas and investigated weekly patterns as well as use on specific events.
The highest number of WBE studies was conducted in Europe, from where we evaluated twenty-eight selected studies reporting in eighteen different countries (Table 4).Studies on a growing number of participating cities are now organized on yearly basis by SCORE and reported to the European Monitoring Centre for Drugs and Drug Addiction (EMCDDA).Therefore, also general time trends on illicit drug use in Europe are well reported based on a certain period of sampling each year.Nowadays around 70 cities are included in this project (EMCDDA, 2020).For these Europe-wide studies by the EMCDDA the protocol to monitor drug loads is fixed, and common quality control is used in all locations which makes true comparison possible.Overall reliability scores for European-wide studies are amongst the highest (Table 3).The consumption of cocaine is most abundant in western European cities, but a new increase in eastern European cities is also reported by the EMCDDA (EMCDDA, 2020).For amphetamine, highest levels are detected in north and eastern Europe and lower in southern Europe.Methamphetamine consumption is relatively low in Europe overall, with the exception of Slovakia.Since 2019, an increase has been reported by the EMCDDA in eastern Germany, Spain and Cyprus (EMCDDA, 2020).Consumption of MDMA is the highest in The Netherlands, Belgium and Germany, but consumption peaks are also observed in England and Scandinavia (González-Mariño et al., 2020) (Table 4).
The majority of the studies in Europe show an increase of drug use during weekends compared to weekdays, with an exception for methamphetamine (EMCDDA, 2020).These trends were also observed and reported by González- Mariño et al. (González-Mariño et al., 2020).
Within the USA, illicit drug use is a major concern (Foppe et al., 2018).The first study using the wastewater-based epidemiology approach in the USA was in 2009 (Banta-Green et al., 2009).More recent studies in the USA (Croft et al., 2020;Foppe et al., 2018;Skees et al., 2018;Subedi and Kannan, 2014) and Canada (Palardy et al., 2015) followed since then.Here, we evaluated five selected studies, reporting on five different states (Table 4).A high cocaine consumption was measured in New York, as well as for use of amphetamine and methamphetamine in urban and rural areas in the midwestern United States (Croft et al., 2020).A high methamphetamine concentration can affect the amphetamine concentration due to biotransformation, the data was corrected for this issue.
MDMA loads were low for urban areas and not detected in rural areas of the midwestern States, but higher in the state of New York.In Tennessee, reported consumption data of methamphetamine and amphetamine are higher than data from New York state and Kentucky, whereas the cocaine consumption was lower (Skees et al., 2018).Illicit drug use in the midwestern United States is measured during week with special events, showing significantly increased consumption during a festive week (Foppe et al., 2018).
Compared to Western countries, WBE is not as often applied in South America.We evaluated six selected studies, reporting on two different illicit drugs in four different countries (Table 4).WBE in South America has been applied in Brazil (Mariño et al., 2019;F. F. Sodré et al., 2018;F. Sodré et al., 2017), Colombia (Bijlsma et al., 2016), Costa Rica (Causanilles et al., 2017) and the island of Martinique (Devault et al., 2014b)  to estimate the overall cocaine consumption for specific areas.Overall, cocaine consumption in South-and Central America is reported to be relatively high compared to Europe/globe, possibly related to the proximity to the production and the relatively high purity of street cocaine (Bijlsma et al., 2016).
During the last decade, the WBE approach became a more popular indicator for drug consumption in Asia.We evaluated thirteen selected studies, reporting on five different countries (Table 4).Studies from China (Khan et al., 2014;Jing Li et al., 2014;Shao et al., 2020), South-Korea (Kim and Oh, 2020;Yong et al., 2015), Vietnam (Nguyen et al., 2018), the Maldives (Fallati et al., 2020) and Malaysia (Du et al., 2020) have been published in the last decade.Four megacities in China (Beijing, Guangzhou, Shenzhen and Shanghai) show significantly lower drug consumption data compared to other continents (Khan et al., 2014).Where cocaine is a popular recreational drug in North-and South America and Europe, it is consumed much less in China.However, consumption of methamphetamine in China is relatively high.Methamphetamine use is concentrated in the centre of urban areas, indicating a relation with the availability to entertainment activities.Even higher methamphetamine rates are reported in Ho Chi Min, Vietnam and Kuala Lumpur, Malaysia (Table 4).Lower drug consumption is reported in South-Korea.During New Year's Day and Christmas day the methamphetamine no noticeable changes in consumption were observed.Consumption of MDMA and amphetamine is relatively low in China and South-Korea, but MDMA consumption for the Maldives and Malaysia is reported to be comparable to the high levels observed in northern Europe (i.e., The Netherlands, Belgium).The Maldives and Malaysia are popular holiday destinations which might explain why synthetic drugs are more popular in these places compared to the rest of Asia (Du et al., 2020), supported by the fact that drug consumption increases during the holiday season.
With regards to Australia, we evaluated four selected studies in three different states (Table 4).The first study that used WBE as a tool to estimate illicit drug use in Australia was in 2011 (Irvine et al., 2011).Similar trends are observed as in Asia, the cocaine consumption is lower than consumption of methamphetamine, while consumption of MDMA is closer to European consumption data.MDMA and methamphetamine are more popular in rural areas whereas cocaine is more frequently used in urban areas.Noticeable is an increase in drug consumption during weekends for popular drugs such as cocaine, amphetamine, methamphetamine and MDMA.Daily illicit drug consumption on an annual music festival in Australia also shows an increase in MDMA compared to the data from a normal week (Lai et al., 2013).A large investigation is described on illicit use of cocaine, methamphetamine and MDMA (Lai et al., 2016), identifying an overall increase in illicit drug consumption in both metropolitan and rural communities over a six year period.In Australia, as in Western Europe, synthetic drugs are more popular.Levels similar to those in Australia of methamphetamine and cocaine  (Kumar et al., 2019), although MDMA consumption is relatively low.
The application of WBE to monitor illicit drug use is still largely lacking in Africa.So far, only two WWTPs located in Johannesburg and Cape Town, South-Africa, have been monitored to estimate drug use patterns (Archer et al., 2018).Methamphetamine is the primary substance detected, at consumption data comparable to USA and Australia.Clandestine manufacturing of illicit methamphetamine results in dumping events of chemical waste, which could be traced via spikes including the precursors used in the production process observed at the WWTP in Johannesburg (Archer et al., 2018).Cocaine consumption is found to be similar to European consumption data, while consumption data for MDMA are low compared to Australia, the USA and Europe.
Fig. 2 illustrates the global consumption distribution for MDMA (purple), methamphetamine (blue), cocaine (red) and amphetamine (orange).So far, WBE has not been used to determine illicit drug consumption in most of the African continent, the Middle East and Russia Overall, high consumption rates are found in the USA, especially for cocaine and methamphetamine (Fig. 3).For Asia, except for methamphetamine, lowest consumption is reported.
Cocaine production expressed as a percentage of the gross domestic product (GDP) is the highest for Colombia (2 to 3%) (INCB, 2002).while for the USA profits from drugs traffic are estimated to be 0.5% of GDP (INCB, 2002).Within Europe, the cocaine market is the second largest illicit drugs market with an estimated retail value of 9.1 billion euros in 2017 (EMCDDA and Europol, 2019).The total value for the retail market for illicit drugs in Europe was estimated at 30 billion euros in 2017 (EMCDDA and Europol, 2019).The consumption of synthetic drugs, such as amphetamine and MDMA, in Europe increased in the last decade, especially for the northern and eastern regions of Europe.Similar trends are observed in Europe's synthetic drug market, which has rapidly grown in the past ten years (EMCDDA and Europol, 2019).Large-scale production of MDMA and amphetamine remains concentrated in the Netherlands and Belgium, while production of methamphetamine is mainly concentrated in central Europe (EMCDDA and Europol, 2019).In Asia, drug consumption is relatively low compared to other continents.According to a report by the United Nations Office on Drugs and Crime (UNODC) in Asian regions, illicit drug production in Asia is increasing for amphetamine-like substances (Liu et al., 2013;UNODC, 2008).This is proportional to the increase in methamphetamine consumption in China and South-Korea.
The reported results should be viewed with an assured degree of uncertainty.Consumption values are frequently reported without calculated uncertainty values (Ort et al., 2010a).Only a handful of the reviewed papers (Causanilles et al., 2017;Deng et al., 2020;Devault et al., 2018;Du et al., 2020;Kumar et al., 2019;Shao et al., 2020;Skees et al., 2018;F. F. Sodré et al., 2018) explicitly reported on uncertainty values for their consumption data (±xy).The complete uncertainty assessment would include the sampling uncertainty, analytical uncertainty and the uncertainty for the consumption estimation as mentioned in the previous chapter.The trueness of the final consumption data should undergo a suitable statistical propagation.Various studies have used Monte Carlo analysis (Been et al., 2015;Croft et al., 2020;Jones et al., 2014;Shao et al., 2020;Y. Y. Zhang et al., 2019;Zheng et al., 2019) to calculate the uncertainties.Fig. 3 shows the distribution of all reviewed papers per drug for each continent including the reliability of the method used to obtain the data points.For many drug/continent combinations, the majority of the data fall within one order of magnitude.

Conclusion
Based on the relatively limited number of studies that could be identified as high quality, further advancement of the WBE approach for illicit drugs is desired especially with regards to the sampling methodology.However, studies carried out in Europe within the SCORE research group, have been using a systematic analytical approach, including annual interlaboratory exercises as an important factor in making data more comparable.Notable is that over the years, the analytical methodology is rapidly improving, and the implementation of high-resolution mass spectrometry techniques will allow more efficient methodologies for wide-scope screening, identification of untargeted/ suspect compounds, or for the detection of metabolites and transformation products.Due to more advanced analytical approaches in the last decades, detection limits are decreasing and became less challenging when measuring low concentrations.Publishing full information regarding detailed sampling and analytical methodologies, even if only in the supplementary data, is crucial for the interpretation of obtained results.It will be necessary to harmonize the criteria for reliable identification/confirmation and accurate quantification of compounds detected in samples, reliable sampling, and less uncertainty in population estimation.This way, data collected from different laboratories in different parts of the world would become more comparable to each other and valid conclusions can be drawn on illicit drug consumption in This study provides an overview of the available literature, reporting data on drug consumption using the wastewater-based epidemiology approach.Overall, high consumption rates are found in the USA, especially for cocaine and methamphetamine consumption, while lowest illicit drug consumption is reported in Asia.These results agree with data reported by the EMCDDA (Europe) and the UNODC (other continents).The WBE approach can identify trends in consumption quickly which makes it extremely useful in generating data on illicit drug use.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Fig. 2 .
Fig. 2. : Average illicit drug consumption per country for MDMA (purple), methamphetamine (blue), cocaine (red) and amphetamine (orange) based on data published over the last decade.

Table 3
Overview of individual and total scores of papers reporting illicit drug concentrations using a wastewater-based epidemiology approach.

Table 4
Overview of drug consumption on a global scale (in mg/day/1000inhabitants).