Thirty-four of the monitored 111 PhACs were found to exceed their respective LOQ value at least once in one of the water samples (Table S3); additionally, 21 of the PhACs were detected in more than one sample. There are significant differences in the frequency of occurrence and concentration levels of the detected PhACs (Table 1).
Theophylline was found to have the highest absolute concentration (max = 7184 ng L−1); however, those were frequent only at a rate of 20%. The average concentration of caffeine, i.e. the most common stimulant, exceeded 1000 ng L−1. It is present in various food items and health supplements (e.g. coffee, energy drinks) as a natural component, and is not related to the consumption of pharmaceuticals; however, it can be an indicator that the spa water has been contaminated by urine and/or other bodily fluids (Teo et al. 2016). As was observed with two antiepileptics (carbamazepine and lamotrigine), three types of hormones (E1, EE2 and testosterone) and the illicit drug cocaine, lidocaine (anaesthetic) and diclofenac (NSAID) the latter is an agent of several non-prescription drugs, were found to have an average concentration above 10 ng L−1. Note that, whether a PhAC is prescription or non-prescription does not impact the frequency of its occurrence in the samples. For some active substances (e.g. carbamazepine, diclofenac, cocaine), the results were generally consistent with what was expected as based on the results for swimming pools or different types of polluted waters frequented by tourists; however, until this study, most of the above-mentioned PhACs had not been measured in bathing waters. It should be noted that 17α-estradiol (aE2) and E2, which are generally detected in all types of environmental monitoring assessments (Aus der Beek et al. 2016), were detected, but the frequency of detection of the former was much higher in the sampled thermal spa water.
These findings can be used to compare the proportions of the different groups of PhACs (Fig. 1). Although hormones were the most frequent, the occurrence of hallucinogenic drugs was also higher than their ratio within then monitored 111 PhACs. The increased proportion of antiepileptics (e.g. carbamazepine, lamotrigine) is also significant; conversely, the proportion of antidepressants decreased in the found and frequent groups. Moreover, as with swimming pool water (Ekowati et al. 2016; Fantuzzi et al. 2018), this group of PhACs, particularly carbamazepine, was most frequently found in thermal spa water. In contrast, anxiolytics and anaesthetics were not observed in high concentrations, although some of their representative compounds, such as tramadol and lidocaine, were detected in many of the samples. The latter two PhACs were also found to be highly persistent, as reported by Bollmann et al. (2016), Wood et al. (2017), Malchi et al. (2014) and López-García et al. (2018). It should be noted that neither the occurrence or concentration of PhACs was found to be related to the chemical properties of the spa water (Table S1). This also suggests that PhAC content is independent of the water source, and that the analysed PhACs are chemically stable enough to not interact with the high solute components of the thermal water.
Regarding the PhACs that were detected only once, Spa B, which is very popular with foreign visitors, exhibited the highest rate of occurrence of single detection (four of six samples). These PhACs include antidepressants (trazodone), NSAIDs (paracetamol) and hallucinogenic drugs (ketamine and norketamine). Accordingly, the lowest occurrence (three of 10 samples) was found at Spa A, which also has a high number of domestic and international visitors; one antidepressant (bupropion) and two cardiovascular drugs (betaxolol, verapamil) were detected only once there. Alternatively, one antidepressant (tiapride) and one anxiolytic (cinolazepam) were found to be unique agents in three samples from Spa C. For the local spas, there was no single occurrence at Spa D, and, in the case of Spas E and F, the single-occurrence PhACs were in the group termed ‘other’, e.g. papaverine and anxiolytics (alprazolam, cinolazepam). Although there were single occurrences of antidepressants at all of the international baths, this was not the case for any of the local spas. Note that the above-mentioned PhACs were not included in further analysis since they were only found in one sample.
Seasonal and geographical analysis
Seasonal analysis of the 17 samples collected from all outflows revealed hormones to be the most prevalent group in the summer. Of the eight detected hormones, only testosterone was found to occur in every season; hormones related to contraceptives were detected in all of the summer samples (Fig. 2, Table S3). This is consistent with the empirical fact that young women tend to visit thermal baths more often during their summer holiday (HCSO 2017).
Alternatively, drugs used for the treatment of cardiovascular disorders (e.g. bisoprolol, metoprolol, perindopril) were most prevalent in the tourism off-season. Two of the four cardiovascular PhACs were not detected in the high-tourist season, and the remaining two were only found in a few samples. As a hypothesis, this may indicate that older generations prefer to visit thermal spas in the off-season, as it was also reported by Löke et al. (2018).
Various PhACs, such as the anaesthetics tramadol and lidocaine, were found in most of the water samples, regardless of the season. The possible reasons for the high proportion and persistence of antiepileptics have been discussed above. It should be noted that the absence of lamotrigine in the summer samples was unexpected. This phenomenon cannot be explained using the currently available results; thus, its interpretation necessitates further investigation. As was observed in the results from swimming pools in Italy (Fantuzzi et al. 2018), in this study, cocaine was detected in every season; furthermore, it was found in every sample in the pre-season. This finding indicates that cocaine consumption is also widespread among the local population, as Thomas et al. (2012) and Mackul’ak et al. (2016) have also revealed. However, the absolute peaks were observed in the summer at spas frequented by tourists (Table S3).
Regarding geographical variation, research has shown that, as compared to spas mainly visited by the locals, nearly all PhACs occur more frequently at international thermal spas within the city centre, and that the average concentrations of the detected PhACs are higher, especially in the cases of cocaine and certain hormones. The exceptions are the two forms of oestrogen and the cardiovascular drug propafenone, which occur more frequently at spas located outside of the city (Fig. 3).
Eight water samples were collected for Spa A diurnal analysis, and only 15 of the 111 possible PhACs were detected (Fig. 4). It should be noted that nearly half of the identified PhACs were hormones, and that the occurrence (and non-occurrence) of many other agents were atypical.
EE2 was identified at high concentrations (average: 23.1 ng L−1), and with a wide range (4-98 ng L−1, in seven of the eight samples (coefficient of variation, CV = 140%). The oestrogens were found to dynamically fluctuate, exhibiting no apparent patterns. Different types of hormones (testosterone, progesterone, levonorgestrel) were only occasionally measured at low concentrations.
Of the anaesthetics, lidocaine was dominant in terms of frequency and concentration. Nevertheless, the concentration of this substance relevantly fluctuated (CV = 130%), and it was absent in three samples, indicating fast water replacement. Additionally, the steadily high concentration of EE2 indicates persistent contamination throughout the entire day. It is also noteworthy that the typically frequently detected carbamazepine (Aus der Beek et al. 2016; Heberer 2002) was detected only once (in the afternoon sample), and that the concentration of diclofenac was found to be zero. Although they were detected in only three samples, alkaloids were found to have the highest concentration, i.e. >1 μg L−1 in each case; this also indicates fast water replacement and no accumulation. Additionally, although cocaine was detected in only two of the eight samples, its metabolite (benzoylecgonine) was present in five samples.
Regarding daily distribution, an absolute peak was found in the number of PhACs measured in sample of 15:00 at Spa A, when nine PhACs were found. The next highest peaks occurred in the noon and midnight samples. The 9:00 sample, and the sample containing the murky water observed at 18:30 (according to general water chemistry, this sample was due to pool rinsing), were found to have the fewest PhACs, even though both samples also contained lidocaine and EE2.
Regarding the diurnal analysis for Spa B, which has international visitors, of the 15 PhACs that were found, only one-third of them were hormones (Fig. 5). Some of the detected hormones were also found at Spa A (e.g. E1, E2, estriol); however, the summer diurnal analysis for Spa B did not yield EE2; furthermore, it was only detected once at Spa B (26 July 2018). Although several types of hormones were found, their concentrations were not high; specifically, with the exception of the testosterone measured in one sample, all hormones remained below 1 ng L−1. Additionally, only the concentration of E1 was stable, as the concentrations of the other hormones fluctuated throughout the day; specifically, their occurrence was inconsistent.
The concentrations of the anti-inflammatory drug diclofenac and anaesthetic lidocaine were found to be high (typically >10 ng L−1), moreover, the time of day did not relevantly affect the concentrations of these compounds (CV = 40% and 30%, respectively). Because there was no accumulation, the persistent presence of these compounds indicates continuous and largely invariable levels of contamination. The concentration of the antiepileptic drug carbamazepine (CV = 14%) was also found to be high and very stable; specifically, the concentration was considerably higher (36.1 ng L−1) than the swimming-pool-water average (1.1 ng L−1) measured by Fantuzzi et al. (2018). Although Fantuzzi et al. (2018) detected carbamazepine metabolites at concentrations up to 62 ng L−1, their accumulation resulting from water recirculation should be taken into account. Thus, the persistently high concentration as a result of continuous contamination is rather relevant. Additionally, although it fluctuated (CV = 63%), the concentration of cocaine was found to be the highest; furthermore, the concentration remained high throughout the day. The concentration of cocaine measured at Spa B within a single day (46-194 ng L−1, average: 104.2 ng L−1) was found to be higher, by two orders of magnitude, than the corresponding swimming pool measurement by Fantuzzi et al. (2018) (average: 1.29 ng L−1), and the average of the data used for geographical analysis (4.8 and 1.3 ng L−1 for international and local spas, respectively). It should be noted that, although the cocaine metabolite benzoylecgonine was detected in several samples from Spa A under the condition of low cocaine occurrence, this metabolite was not detected at Spa B. This is unexpected, as benzoylecgonine is much more stable than cocaine and the concentration of the former is generally higher (McCall et al. 2016; Thiebault et al. 2019). The background of this finding has been unknown yet, presumably, some other sources of cocaine (other than human metabolism) might have been present in the thermal water of Spa B.
Regarding daily fluctuation, the number and concentrations of PhACs were found to reach their peak in the early afternoon. Weng et al. (2014) and Teo et al. (2016) reported that caffeine could be a good indicator of urination and other types of excrement in swimming pools because it was consistently present at a high concentration, which was related to the number of visitors. However, this theory is not fully supported by the findings of this study; although one sample was found to have a high concentration of caffeine, it was absent in four of the samples (note: eight total samples). This supports the view that the measurements from swimming pools with strongly chlorinated water and water recirculation systems can only be indirectly compared to thermal spa systems.
Analysing all of the samples (off-season, pre-season, main season and diurnal monitoring) from the two international spas, which are similar in size and target the same type of visitors, revealed that the frequency and concentration of carbamazepine are constantly low at Spa A, unlike those at Spa B. However, as compared to Spa B, the frequent occurrence of hormones and constant presence and high concentration of EE2 at Spa A are relevant.