Wastewater-Based Epidemiology (WBE): Potential Application for Assessing the Use of Conventional and New Generation Tobacco and Nicotine Products

The selection of suitable BMs in wastewater (WW) is of fundamental importance for successful WBE applications. For instance, alcohol consumption can be best assessed with the minor metabolite ethyl sulfate (ES) in WW. Ethyl glucuronide (EG), despite occurring in higher amounts in urine than ES, was found to be unsuitable, because it is split in WW by fecal glucuronidases (23, 30, 45). ES degradation under WW sampling conditions was reported to be 8%/h (30). For application of ES in WBE, a urinary excretion rate (UER) of ~0.05% is usually assumed with a half-life of several hours (93, 94).

Of the 20 caffeine metabolites, paraxanthine (1,7-dimethylxanthine) was found to be suitable for the application of WBE to caffeine use (16, 17, 19, 95, 96). Stability was reported to be sufficient (17, 97) for paraxanthine, which is the major caffeine metabolite (representing 70–80% of which a main part (~50%) is excreted into urine resulting in a UER of 35–40%) (98, 99).

Cotinine (Cot) is a major metabolite of nicotine, absorbed by various routes upon use of tobacco and nicotine products (100). However, in many WBE studies, nicotine consumption is assessed by measuring 3′-trans-hydroxycotinine (OH-Cot) in addition to or instead of Cot (16, 18, 23, 25, 27, 36, 42, 43, 49, 50, 52,53,54, 56, 59, 62,63,64, 95, 96, 98, 101). For the purpose of differentiating the use of tobacco-containing (e.g., CC, oral tobacco) and tobacco-free nicotine (oral nicotine pouches (ONPs), nicotine replacement therapy (NRT) products), anabasine (AB) is used as a BM for tobacco exposure (52, 57, 59, 62,63,64, 101, 102). Cot shows the highest stability, whereas OH-Cot and AB proved sufficiently stable for WBE studies (23, 30, 62). An UER of about 30% can be assumed for Cot, comprising the free and conjugated forms (103, 104). This percentage considers that the glucuronides of cotinine and OH-Cot are not stable under WW sampling conditions and only free Cot is present in WW samples (49). Fecal ß-glucuronidase is assumed to split these conjugates (50). Also the molar sum of Cot and OH-Cot in WW was used as BM for nicotine uptake. In this case an excretion rate of 74% was most frequently applied for back-calculations (50).

WBE has the potential to provide spatial and temporal data on substance use and exposure (24, 36). Therefore, careful selection of time and place of WW sampling is important, particularly with regard to the study aim. Also, detailed information on the eligible wastewater treatment plants (WWTPs) is necessary, in order to select the optimal sampling points and conditions (2). For studying the use of life-style products, it is of significance to deliberately select the year, season and month of WW sample collection (16, 105). Especially the choice of workdays or weekend days, holidays, days of special events (e.g., music festivals, etc.) are of importance (26).

As suitable sampling places, country, city, community, special areas (sub-sewersheds) or buildings (e.g., university campus (58)) can be considered, depending on the problem to be investigated.

Various types of WW samples can be collected. Most suitable are representative 24h-composite WW samples at the WWPT inlet (before any WW treatment takes place). Sampling can conveniently be conducted by means of auto-sampler devices (2, 106,107,108). Also passive sampling (using microporous polyethylene passive samplers) was found to be a promising, simple and cheap method for WW sample collection (109).

Analytical chemistry is regarded as the fundamental driver in the WBE field (108). Determinations of biomarkers (BMs) in WW is very similar to that of BM analysis in urine with respect to sample preparation (liquid-liquid extraction (LLE), solid-phase extraction (SPE) or direct injection) and instrumentation (most frequently LC-MS/MS is used) (110,111,112). As mentioned above, due to fecal glucuronidase activity, BMs usually do not occur as glucuronides. Since concentrations of (for example) cotinine in WW compared to those in urine of smokers are much lower (by approximately a factor of 100 to 1000), analysis of WW samples is more challenging than that of urine from (exposed) people. Enantiomer-selective analysis can improve the specificity for human excretion data (3, 29, 96, 108). For special purposes, also untargeted analysis of WW samples can be of interest (96).

WW analysis accrues data in from of concentrations of BMs (usually of metabolites of the consumed compounds of interest, in addition possibly also of the markers for population size estimates). Back-calculation of these raw data to population-based consumption/exposure results would allow comparisons with other (geographical) places and time points. Information on the size of the wastewater catchment population (the population covered by the WWTP, where WW sampling took place) is crucial for WW-based results (113). General formulae for back-calculations are presented in the following (2, 3, 114):

Daily mass load of biomarker (DML, mg/d): DML=C×F×SF DML=C\times F\times SF with: C

Concentration of BM (mg/L);

Drug consumption (DC, mg/d) of the total catchment population: DC=DML×MWparUER×MWmet=DML×CF DC=\frac{DML\times MWpar}{UER\times MWmet}=DML\times CF with: UER

Urinary excretion rate (%),

Population-normalized daily consumption (PNC, mg/d/1000 persons): PNC=DCPnorm×1000 PNC=\frac{DC}{Pnorm}\times 1000 with: Pnorm

population size used for normalization (e.g., the catchment population of the WWTP)

Units (of a product) consumed (UC, e.g., cig/d): UC=DCAC UC=\frac{DC}{AC} with: AC

Average content in product (e.g., mg nicotine/cig; more precisely would be to take the average amount of nicotine absorbed per cigarette)

Estimate for Pnorm (using a suitable BM for estimating the population size): Pnorm=DMLPEPML Pnorm=\frac{DML\left( PE \right)}{PML} with: PML

Personal daily excretion (mg/d/person);

The usual data sources for tobacco consumption in larger populations are for example tax excise clearances, sales data, and self-report surveys (53). WBE data can be a complementary source for assessing the consumption of nicotine, alcohol or caffeine containing products. Agreement in consumption data between WBE and the conventional information sources was frequently used to show the suitability of WBE for this purpose (16, 26, 37, 43, 48, 50). Deviations between WBE and conventional methodologies can have plausible reasons (e.g., under-reporting, contraband, launch of new products, etc.). This issue will be discussed in more detail for tobacco/nicotine products in the section “Comparison of WBE results with those of ‘classical’ methods”.

With respect to the market introduction of new generation tobacco/nicotine products (NGPs), temporal trends in WBE data could be of particular interest. As mentioned earlier in this section, use of anabasine as a tobacco-specific and cotinine as a nicotine-specific biomarker in WW samples would allow the differentiation of various products (e.g., CCs and ECs) and, as a consequence, the actual use of these products in the population. This topic will be discussed in more depth in the section “Results and discussion of the WBE Review”.

An established biomarker in WW for the consumption of nicotine-containing products is cotinine (Cot), which was a major inclusion criterium for the studies in Table S1. In some studies, additionally 3′-trans-hydroxycotinine (OH-Cot) or the parent compound nicotine (Nic) was determined in WW samples. If possible, the Cot-based nicotine consumption data were evaluated. In some cases, only consumption data based on the molar sum (Cot + OH-Cot) were reported and used in this evaluation. Nic in WW was not included in our evaluation, due to the fact that also non-consumed but disposed nicotine (cigarette butts, ashes) can considerably contribute to the Nic measurable in WW (27, 96). If PNC values reported in the publication were not normalized to the general population (including all age groups), values were adjusted appropriately. Adjustments were also carried out, if the applied urinary excretion rates (UER) for Cot and Cot + OH-Cot substantially deviated from the established values of 30 and 74%, respectively (50). The evaluation in this section is based on WW data sets which can comprise results from WW samples of one sampling point (usually a WWTP) at different time-points or results from WW samples derived from different sampling points at one point in time.

Results by continent of PNC of nicotine (expressed in mg/d/1000 persons) are summarized in Table 1 and Figure 2. PNC data for nicotine were obtained from 170 data sets (Table 1) obtained from 60 studies in 4 continents (no data from Africa were available). Most studies were performed in Europe (N = 107 WW data sets) with investigations in 17 different countries (most frequently in Belgium (5 data sets), Italy (4) and Turkey (4), followed by Australia (10)/NZ (2), Asia (with 9 data sets from China) and America (USA (7) and Brazil (1)). The means for nicotine PNCs for the 4 continents were found to be in a relative narrow range of 2000–3000 mg/d/1000 p. However, the variability within continents, as can be deduced from the min–max ranges, 95 %-CIs and SDs were large. This has to be presumed primarily as an indication of the diversity of the investigated populations, since data known to be influenced by factors such as COVID-19 restrictions, data from WWTPs serving a university campus, special events were excluded from this evaluation. By no means, the presented data are representative for a continent or a country, but should first of all show that WBE for use of nicotine-containing products has been successfully applied on a global basis with comparable results, which mainly depend on the population served by the WWTP from which the WW samples were derived. The following sections provide further support for this statement.

Figure 2.

Mean WBE-derived nicotine consumption in 4 continents. Error bars represent the variation between 5% and 95% quantiles.

In Table 2, results of WBE studies on nicotine consumption covering a time period of at least one year are summarized. Four of the nine studies were conducted in Australia (also reflecting the fact that a very active WBE research group is working in this country).

The other studies were performed in China (2), Italy (1), Spain (1) and Lithuania (1). The studies starting at about 2010 showed consistently a population-normalized decrease in nicotine consumption until about 2018 (18, 53, 59, 63, 102) or remained at a stable level (31). After about 2018, WW-derived nicotine consumption was stable or increased (25, 59, 138). As indicated in Table 2, the study authors attribute different reasons for the decrease. Mainly a decline in cigarette smoking as a result of tobacco control policies is mentioned. Use of NRT products and new generation tobacco/nicotine products (NGPs), particularly ECs, as a replacement source for nicotine in WW were suggested to play a role only after 2018 (25, 59, 138). The decrease in the consumption of tobacco-containing products (mainly CCs) over time could convincingly be demonstrated by the application of AB analysis in WW, which showed a time pattern different from Cot (59, 63). Time courses in WW-based nicotine consumption were predominantly in agreement with tobacco (cigarette) sales and other survey data. Deviations, however, were observed for the absolute consumption data, which were commonly higher in the WW-derived results. As possible reasons, the authors discussed use of illicit tobacco products, too low self-reported consumption, and use of tobacco-free nicotine products (particularly NRTs and ECs).

In many WBE studies on nicotine/tobacco consumption, influence of weekends (Saturday, Sunday) compared to common workdays (Monday to Friday) were investigated (Table S1). The results predominantly show that nicotine consumption is hardly different between workdays and weekends (22, 25, 32, 33, 35, 38, 42, 43, 50, 63, 76, 116, 147, 150). Two studies (11, 16) reported significantly lower nicotine consumptions on weekends: one investigated a university campus, which also found lower alcohol but stable caffeine consumption on weekends (16); the other study mentioned work-related commuting as a reason for lower levels on weekends (11). In contrast to nicotine, more alcohol was found to be consumed on weekends in some studies (42, 43, 63, 139, 144).

Also events such as (religious) holidays, tourist seasons and festivals were reported to have no major influence on WW-based nicotine consumption (21, 39, 140). Increased nicotine consumption was found during a rock festival (143). A significant decrease in WW Cot levels were reported during Ramadan, as was the case for ethyl sulfate (ES), a WW biomarker for alcohol consumption (34).

Several WBE reports deal with nicotine consumption during COVID-19 restrictions (29, 138, 141, 144,145,146). Three studies reported no or only minor impact of restrictions or lockdowns during the pandemic (29, 141, 144), two investigations found a decrease (145, 146). In one study (138) an increase in nicotine consumption was reported.

The findings presented in this section convincingly show that WBE data reflect the use of life-style and other products in almost real-time. Contrary to expectations, smoking is mostly not influenced by special events, in contrast to alcohol consumption.

Since WBE is a relatively young method for assessing changes in habits, exposures and health status in populations, for validation reasons many studies compare the WBE-derived results with those from ‘classical’ methods such as surveys, epidemiological studies, official statistics as well as sales and tax data (Table S1). The outcome of these comparisons can be of interest, not only when WBE and ‘classical’ approaches were in agreement, but also if deviations are consistent and have plausible and accountable reasons (e.g., see Table 2). Comparisons between WBE-derived and survey/sales data of three studies (59, 63, 138) are already presented in Table 2. Most of the studies investigating this aspect show a fair accordance between WBE and ‘classical’ methods in terms of daily cigarette consumption or percentage of smokers in the population (18, 27, 31, 36, 42, 50, 54, 58, 140, 142,143,144, 147,148,149,150). In a couple of studies WW cotinine-based tobacco (mostly cigarette) consumption was found to be higher than observed in surveys or deduced from sales data (15, 27, 35, 53, 64). The most frequently discussed reasons for this discrepancy were other nicotine source (other than cigarettes, such as nicotine replacement therapy (NRT) products, snus, new generation tobacco/nicotine products (NGPs)), use of illicit tobacco products or too low self-reports of the consumption. One study reported a better agreement of tobacco sales data with WW anabasine (AB)-based (only 5% higher) than WW cotinine-based consumptions (2–28)% higher) (64). A few investigations found lower WBE-derived consumption levels than those derived from surveys or sales data (16, 27, 56). The discussed reasons for these deviations were less obvious and mostly explained with demographic differences between the wastewater treatment plant (WWTP) catchment population and that used for the ‘classical’ methods.

In general, there are limitations in the comparisons of the results derived by WBE and classical methods. For instance, cigarette consumption was based on tax data in part of the studies or on retail sales data in other studies. A serious bias for consumption could result from products purchased abroad. This is of particular importance for border regions.

The application of Cot and OH-Cot in WW can assess only the use of nicotine-containing products in populations, but not differentiate between tobacco-containing products such as combustible cigarettes (CCs), pipes, cigars, oral tobacco and tobacco-free nicotine products such as NRT products (nicotine gum, patch, inhaler), electronic cigarettes (ECs), oral nicotine patches. As mentioned earlier in this paper the differentiation between tobacco-containing and tobacco-free products is possible by measuring also anabasine (AB) and anatabine (AT) in WW. This has been done for a number of studies (52, 57, 59, 62,63,64, 101, 102), which are briefly summarized in Table S1 and Table 2. AB and AT concentrations in WW are about 500–1000-fold lower than those of Cot and OH-Cot. AB and AT were found to be less stable in WW reactors than Cot and OH-Cot (62). AB is preferred to AT as a biomarker in WW due to its lower variability in tobacco products (64, 102). Beyond that, AB showed similar per person load in pooled urine and WW (adjusted appropriately), whereas AT load in WW was by 50% higher (62) than expected. The authors alleged AT occurrence in food (tomatoes, eggplants, peppers) as reason for the latter observation (62). In general, the AB levels in WW, particularly the AB/Cot ratios, were found to be a well-suited indicator for the use of tobacco products in contrast to Cot and OH-Cot alone, which reflect the overall nicotine consumption by use of any nicotine-containing product. Reported levels of AB and AT in cigarette tobacco are in the range of 100–200 and 900–1400 µg/g, respectively (154, 155). Tobacco contents of NAB (N′-nitrosoanabasine), NAT (N′-nitrosoanatabine) and NNK (4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone) were found to be in the ranges 15–23, 114–145 and 80–960 ng/g, respectively (156, 157). Additionally, the N′-nitrosamines NAB and NAT may be formed by nitrosation from the respective precursors AB and AT in wastewater under acidic conditions. The issue of tobacco (instead of human urine) as source of AB, AT, NAB, NAT and NNK in wastewater has not yet been investigated in detail and would require further research. The amounts of these compounds in tobacco compared to those taken up and excreted in urine of users of combustible cigarettes (CC) and oral tobacco suggest that a potential contribution of tobacco to the WW levels of AB, AT, NAB and NAT cannot be ignored.

The tobacco-specific nitrosamines (TSNAs), more specifically their urinary metabolites NAB, NAT, NNN, NNK and NNAL in WW have been addressed in two studies (52, 101). In one study (52), all compounds except for NAT were found to be < LOD. In the other study (101), NNK exposure in four populations in Greece, Switzerland and Belgium were reported to range from 0.5 to 1.5 mg/d/1000 persons, which appears uncommonly high, particularly when considering the authors’ explanation for this observation: third-hand (!) exposure to NNK.

Matter of particular interest would be the differentiation between the use of combustible tobacco products (above all CCs) and NGPs (mainly ECs, HTPs and ONPs), which are all non-combustible nicotine products. In a recent review of ours, it was shown that urinary metabolites of tobacco combustion chemicals such as acrolein (3-HPMA), crotonaldehyde (HMBMA), acrylonitrile (2CyEMA), acrylamide (AAMA or GAMA), 1,3-butadiene (MHBMA) and a number of others are suitable BMs to distinguish combustible from non-combustible tobacco/nicotine products (158). This has been confirmed in a controlled clinical study with users and non-users of these products (159, 160). In this context, a recent publication by Kumar et al. (66) is of interest, which applied WBE to quantify the exposure to volatile organic chemicals (VOCs) such as acrolein, crotonaldehyde, acrylonitrile, acrylamide, 1,3-butadiene and several others in two U.S. communities differing in proximity to manufacturing facilities. As biomarkers in WW, predominantly mercapturic acids were used, mostly in agreement with those employed in the investigation mentioned above (159). Although the focus of the Kumar et al. study (66) was exposure of the general population to VOCs emitted by factories, the application to smoking-derived VOCs appears possible and of supreme interest for differentiating the use of combustible and non-combustible tobacco/nicotine products. Of course, the suitability of the mercapturic acids of interest as biomarkers in WW, particularly their stability under WBE conditions, have to be validated prior to application for the intended purpose. A successful distinction between these two product groups by WBE methodology would allow to monitor the progress of tobacco harm reduction (THR) in populations by selecting suitable WWTPs for WW sample collection. This subject will be covered in more detail in the next section.