Evaluation of the Correspondence between the Concentration of Antimicrobials Entering Sewage Treatment Plant Influent and the Predicted Concentration of Antimicrobials Using Annual Sales, Shipping, and Prescriptions Data

The accuracy and correspondence between the measured concentrations from the survey and predicted concentrations on the basis of the three types of statistical antimicrobial use in Japan was evaluated. A monitoring survey of ten representative antimicrobials: ampicillin (APL), cefdinir (CDN), cefpodoxime proxetil (CPXP), ciprofloxacin (CFX), clarithromycin (CTM), doxycycline (DCL), levofloxacin (LFX), minocycline (MCL), tetracycline (TCL), and vancomycin (VMC), in the influent of sewage treatment plant (STP) located in urban areas of Japan, was conducted. Then, the measured values were verified in comparison with the predicted values estimated from the shipping volumes, sales volumes, and prescription volumes based on the National Database of Health Insurance Claims and Specific Health Checkups of Japan (NDB). The results indicate that the correspondence ratios between the predicted concentrations calculated on the basis of shipping and NDB volumes and the measured concentrations (predicted concentration/measured concentration) generally agreed for the detected concentration of antimicrobials in the STP influent. The correspondence ratio on the basis of shipping volume was, for CFX, 0.1; CTM, 2.9; LFX, 0.5; MCL, 1.9; and VMC, 1.7, and on the basis of NDB volume the measured concentration was CFX, 0.1; CTM, 3.7; DCL, 0.4; LFX, 0.7; MCL, 1.9; TCL, 0.6; and VMC, 1.6. To our knowledge, this is the first report to evaluate the accuracy of predicted concentrations based on sales, shipping, NDB statistics and measured concentrations for antimicrobials in the STP influent.


Introduction
The emergence and spread of antimicrobial-resistant bacteria (AMRB) is rapidly becoming an ongoing problem on a global scale [1][2][3]. The problem of AMRB has been assessed and addressed in medical facilities such as hospitals and livestock [4][5][6]. However, AMRB are widely detected in aquatic environments such as rivers, lakes, and seas [7][8][9], the amount of antimicrobials used in the study area during the period studied was estimated using three statistics: domestic annual sales, shipping, and NDB statistics, and then the predicted concentration of antimicrobials in the STP influent from each amount used were calculated. Finally, the correspondence between the measured values obtained in the survey and the predicted values estimated from the amount used was verified, and the effectiveness of the method as a method for predicting the environmental impact associated with antimicrobial use was evaluated.

Sampling
STP influent was collected at a representative STP located in the Yodo River basin of Japan, which has a population of 17 million (14% of the Japanese population) [41,42]. At the time of this study, the STP disposed of municipal wastewater from 369,000 people, and the annual mean flow rate was 189,730 ± 5000 m 3 /day. The wastewater in the STP was treated with conventional activated sludge and then chlorine disinfection was applied. Sewage sludge produced in the treatment process is dewatered in a centrifuge and incinerated for disposal.
Sampling was conducted during 2019-2021: October 2019, September 2020, January 2021, and March 2021. This sampling frequency was on the basis of previous reports similar to the present research [31,32,43]. Identical manual sampling was adopted at a fixed sampling frequency because a composite sampler could not be used at the STP. Sampling was conducted on rain-free days, and no rainfall greater than 1 mm was observed during the two days before the sampling day [44].
A stainless-steel pail sampler was used to collect 100-mL samples of STP influent. Water samples were placed in sterilized glass bottle samples, and ascorbic acid (1 g/L) was immediately added as a preservative for antimicrobials [7,45]. All water samples were immediately transported to the laboratory in a cooler box (within 1 to 2 h), stored at 4 • C in darkness, and processed within 24 h.

Analytical Procedures for Antimicrobials
The concentrations of target antimicrobials in the wastewater was determined by using a combination of solid phase extraction (SPE) and ultra-performance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS), as described previously [7]. 85 [54] Antibiotics 2022, 11, 472 5 of 15

Analytical Procedures for Antimicrobials
The concentrations of target antimicrobials in the wastewater was determined by using a combination of solid phase extraction (SPE) and ultra-performance liquid chromatographytandem mass spectrometry (UPLC-MS/MS), as described previously [7].
Briefly, 10 mL of STP influent was filtered through a glass-fibre filter (GF/B, 1-µm pore size, Whatman, Maidstone, UK). A known amount of each antimicrobial was spiked into two of the sample solutions for analysis to make a final concentration of 200 ng/L to account for matrix effects and loss during sample extraction [55]. The solutions were then subjected to two different preconditioned SPE cartridges (OASIS HLB, 200 mg; Waters Corp., Milford, MA, USA) at a flow rate of 1 mL/min; the cartridges had been preconditioned for each sample by being washed first with 3 mL of methanol and then with 3 mL of Milli-Q water adjusted to pH 3 with 1 N HCl. Similarly, the two unspiked sample solutions were removed and processed as above. All cartridges were washed with 6 mL of Milli-Q water preadjusted to pH 3 and then dried by a vacuum pump. Finally, the adsorbed antimicrobials were separately eluted with 3 mL of acetone and 3 mL methanol for one, and for the other with 2 mL of 10% (v/v) formic acid in acetone, 2 mL of 10% (v/v) formic acid in methanol, and 2 mL of 5% ammonia-methanol (v/v).
Each combined eluted solution was evaporated mildly to dryness under a gentle stream of N 2 gas at 37 • C. The residue was solubilized in 200 µL of a 90:10 (v/v) mixture of 0.1% formic acid solution in methanol, and 10 µL of this solution was subjected for the UPLC system coupled to a tandem quadrupole mass spectrometer (TQD, Waters Corp.) equipped with an electrospray ionization source with positive and negative ion modes. Detailed information on the analytical parameters used for UPLC-MS/MS analysis is given in Table 2. Product ions in italics were used for quantification.

Method Validation
Six-point standard calibration curves were prepared for quantification ranging between 0.5 and 200 ng/mL. Individual linear calibration curves for each antimicrobial were obtained in the concentration range of 0.5 to 200 ng/mL (r 2 > 0.99) by selecting the weighting factor of 1/x. Quantification was performed by subtracting the blank data from the corresponding data given by the spiked sample solutions to account for matrix effects and losses during sample extraction [56,57]. The recovery rates of antimicrobials in the STP influent ranged from 52% to 110% (Table 2). These profiles were generally similar to those reported in previous reports of pharmaceuticals in river water and wastewater samples [39,58,59]. The limits of detection (LODs) and limits of quantification (LOQs) were calculated as the concentrations at signal-to-noise ratios of 3 and 10, respectively, according to the methods applied to residual pharmaceuticals in environmental water samples [59,60]. These values are also summarized in Table 2.

Prediction of the Concentration of Antimicrobials in the Targeted STP Influent
The annual sales, shipping, and NDB statistical data are summarized in Table 3. Concentrations of antimicrobials in the targeted STP influent (C (STP)in ) can be calculated by using annual sales volume summarized as an aggregate for all of Japan. (Equation (1)), as described below based on the previously reported estimation [31], assumes that antimicrobials are consumed equally every day [61,62] and that antimicrobial use is proportional to the population in each prefecture nationwide [25,27].
where D (Sales)total is the distribution of antimicrobials based on the annual shipping volumes in Japan considering pharmaceutical price and content (kg/year) ( Table 3), E is the excretion rate of the target antimicrobials after administration as unchanged form (%) ( Table 1), 365 is the number of days per year, P STP is the service population in the STP (number of persons) (369,000), P Japan is the population of Japan (126,227,000 persons) [63], W STP is the amount of influent water at the STP (189,730 m 3 /day) [64], and 10 9 is a factor for unit conversion from kg/m 3 to ng/L.  (7) *: Antimicrobial usage in Japan, **: Antimicrobial usage in the target area in this study, N.A.: Not available.
Concentrations of targeted antimicrobials in the targeted STP influent (C (STP)in ) can be calculated by using annual shipping volumes summarized as an aggregate for all of Japan (kg/year) (Equation (2)) [31] and by assuming that all shipped antimicrobials are consumed equally every day [61,62] and that antimicrobial use is proportional to the population in each prefecture nationwide [25][26][27][28]30]: Furthermore, the concentrations of antimicrobials in the targeted STP influent (C (STP)in ) can be calculated by using the annual NDB volumes summarized as prefecture-by-prefecture aggregate values (kg/year) (Equation (3)) and by assuming that all prescribed antimicrobials are consumed equally every day [37,61,62] and that antimicrobial use is proportional to the population in each prefecture nationwide [25][26][27][28]30]: where D (Receipt)prefecture is the distribution of antimicrobials based on the annual NDB volumes in the prefectures (kg/year) ( Table 3) and P Prefecture is the population of the prefecture (8,842,523 persons) [63].

Distribution of Concentrations of Antimicrobials in the STP Influent
Seven antimicrobials (CFX, CTM, DCL, LFX, MCL, TCL, and VMC) were detected at a wide range of concentrations from ng/L to µg/L levels (11 ng/L to a maximum of 4.3 µg/L) during the survey period in the STP influent (Figure 1). The detected concentrations of antimicrobials were similar to those previously reported [7,39,65]. The average detected concentrations of each antimicrobial were 309 ± 334 ng/L (CFX), 336 ± 223 ng/L (CTM), 42 ± 72 ng/L (DCL), 2536 ± 1617 ng/L (LFX), 13 ± 23 ng/L (MCL), 4 ± 7 ng/L (TCL), and 118 ± 120 ng/L (VMC). These results indicated that the antimicrobials used in the present study were used constantly throughout the year, with less seasonal fluctuation in use. This trend was mostly reasonable when trends in clinical use [61] were considered. In addition, pharmaceuticals detected in the environment, including antimicrobial components, are persistent in wastewater treatment processes and/or in the river environments [66,67] and remain in aquatic environments for long periods of time [68,69], the environmental risk of their inflow into the aquatic environments was considered to be a concern [70,71]. On the other hand, APL, CDN, and CPXP were not detected throughout the survey period. The reason for the lack of detection of these antimicrobials in the STP influent was related to the report that these antimicrobials are β-lactam antimicrobials with the common β-lactam structure in the structural formula and are easily degraded in the environment, being attenuated in a few hours to a few days [69,72], and were thought to be degraded in the sewage system before entering the STP.

Prediction of Annual Concentrations of Targeted Antimicrobials in the STP Inf
The estimated concentrations of each antimicrobial in the targeted STP the basis of annual sales, shipping, and NDB volumes are shown in Table 4.

Prediction of Annual Concentrations of Targeted Antimicrobials in the STP Influent
The estimated concentrations of each antimicrobial in the targeted STP influent on the basis of annual sales, shipping, and NDB volumes are shown in Table 4. Concentrations calculated based on the shipping and NDB volumes were similar, and a significant difference (p < 0.05) was not observed in estimated concentrations between the shipping volumes and NDB volumes. In contrast, a significant difference (p < 0.05) was observed in estimated concentrations between the sales volumes and both shipping and NDB volumes; concentrations predicted on the basis of sales volumes were approximately two times lower than those predicted by using shipping and NDB volumes in the case of CDN and CPXP, seven times lower in the case of CTM, and approximately 20 times lower in the case of LFX and VMC. This difference probably occurred due to the statistics on shipping volumes and NDB volumes covering all pharmaceutical companies in Japan, but the statistics on sales volumes covered only those antimicrobials sold by major pharmaceutical companies, and targeted antimicrobials in this study are all sold by a large number of pharmaceutical companies as generic drugs [34]. These discrepancies in reporting are similar to those found in the previous studies [31,32].

Relationships between Measured and Predicted Concentrations of Antimicrobials
The correspondence ratios between the measured concentrations as described in Section 3.1 and the predicted concentrations estimated on the basis of sales, shipping, and NDB volumes as described in Section 3.2 as a ratio of predicted concentration/measured concentration are summarized in Figure 2.
The correspondence ratio between the predicted concentration calculated on the basis of sales volume and the measured concentration was 0.4 ± 0.4 for CTM, 0.02 ± 0.01 for LFX, and 0.1 ± 0.1 for VMC. The main reason for the differences in the predicted concentration and measured concentration based on the sales volume is probably the fact that CTM, LFX, and VMC are sold by many pharmaceutical companies as generic antimicrobials [34]. This trend supports the reports of a difference between NDB antimicrobial use and antimicrobial sales [38]. Interestingly, the orders of magnitude of the predicted and measured concentrations were generally in agreement for shipping and NDB volumes. The correspondence ratio between the predicted concentration calculated on the basis of shipping volume and the measured concentration was 0.1 ± 0.1 for CFX, 2.9 ± 2.4 for CTM, 0.5 ± 0.3 for LFX, 1.9 for MCL, and 1.7 ± 1.7 for VMC, and that on the basis of NDB volume and the measured concentration was 0.1 ± 0.1 for CFX, 3.7 ± 3.1 for CTM, 0.4 DCL, 0.7 ± 0.4 for LFX, 1.9 for MCL, 0.6 for TCL, and 1.6 ± 1.6 for VMC. The correspondence ratios generally agreed with the results in the present study, with previously reported results conducted in different regions based on the estimation from shipping volume (0.9 to 2.9 for CFX. 1.0 to 2.1 for CTM, and 18.3 for LFX [31,32]), although, in all cases, the effects of household disposal could not be excluded in the present estimations, as noted [73]. On the other hand, the fact that β-lactam antimicrobials APL, CDN, and CPXP were not detected in the effluent was considered to be due to the deviation from the predicted concentration due to their easy attenuation in the environment, as described in Section 3.1. 0.4 for LFX, 1.9 for MCL, 0.6 for TCL, and 1.6 ± 1.6 for VMC. The correspondence ratio generally agreed with the results in the present study, with previously reported result conducted in different regions based on the estimation from shipping volume (0.9 to 2. for CFX. 1.0 to 2.1 for CTM, and 18.3 for LFX [31,32]), although, in all cases, the effects o household disposal could not be excluded in the present estimations, as noted [73]. On the other hand, the fact that β-lactam antimicrobials APL, CDN, and CPXP were not de tected in the effluent was considered to be due to the deviation from the predicted con centration due to their easy attenuation in the environment, as described in Section 3.1. The measured concentration tended to be approximately 10 times higher than th predicted concentration in CFX in this study, even in the estimation from shipping and NDB volume. The low predicted concentration of CFX may originate from veterinary and livestock fields as well as from human use. Practically, the use of antimicrobials in animal is approximately twice as high as that in humans [74,75], and this trend is similar world wide [76]. These results suggest that it would be important to attempt to evaluate veteri nary antimicrobials with regional characteristics in addition to those used for human us  The measured concentration tended to be approximately 10 times higher than the predicted concentration in CFX in this study, even in the estimation from shipping and NDB volume. The low predicted concentration of CFX may originate from veterinary and livestock fields as well as from human use. Practically, the use of antimicrobials in animals is approximately twice as high as that in humans [74,75], and this trend is similar worldwide [76]. These results suggest that it would be important to attempt to evaluate veterinary antimicrobials with regional characteristics in addition to those used for human use to implement of risk assessment and management of the impact of antimicrobials in the aquatic environments.

Evaluation of Relationships between Concentrations of Antimicrobials Predicted by Using Sales, Shipping, and NDB Volumes
The correspondence ratios between the predicted concentrations based on the measured concentration and predicted concentrations on the basis of the sales, shipping, and NDB volumes are plotted against each other in Figure 3.
There was a positive correlation between the two (correlation coefficient r = 0.72 for shipping volume and 0.79 for NDB volume). It is therefore possible to calculate predicted concentrations with better precision by using statistical data on shipping volumes and/or NDB volumes, taking into account the actual conditions of use. On the other hand, in Japan, information on the shipping volumes of new pharmaceuticals used in the forefront of clinical practice is not available publicly for reasons associated with the patents applied to the manufacture and sales of these pharmaceuticals [35]. Therefore, the availability of statistical information on NDB volumes may be irreplaceable in some specific cases. For this reason, for the evaluation of new pharmaceuticals, improving the correspondence between measured and predicted values is an issue that needs to be addressed in the future. NDB volumes, taking into account the actual conditions of use. On the other hand, in Japan, information on the shipping volumes of new pharmaceuticals used in the forefront of clinical practice is not available publicly for reasons associated with the patents applied to the manufacture and sales of these pharmaceuticals [35]. Therefore, the availability of statistical information on NDB volumes may be irreplaceable in some specific cases. For this reason, for the evaluation of new pharmaceuticals, improving the correspondence between measured and predicted values is an issue that needs to be addressed in the future. The present method developed for predicting the concentrations of antimicrobials in the STP influent is simple, accurate, and precise. By taking into consideration the removal rates of pharmaceuticals by wastewater treatment processes at STPs and the dilution and attenuation rates during the river water flow where the influents are to be discharged, the The present method developed for predicting the concentrations of antimicrobials in the STP influent is simple, accurate, and precise. By taking into consideration the removal rates of pharmaceuticals by wastewater treatment processes at STPs and the dilution and attenuation rates during the river water flow where the influents are to be discharged, the use of the method used in this study should improve the accuracy of concentration prediction approaches targeting aquatic environments [77][78][79].
To the best of our knowledge, this is the first report to show the results of an evaluation of the accuracy of and correspondence between predicted concentrations based on sales, shipping, NDB statistics and measured concentrations for antimicrobials in the STP influent. These findings will contribute to a comprehensive understanding of the environmental risks associated with the presence of antimicrobials in the aquatic environments. In the future, the method developed here should be useful for the priority screening of antimicrobials and for conducting environmental risk assessments of antimicrobials that are yet to be studied in the aquatic environments. Furthermore, the introduction of advanced water treatment systems, such as ozonation [80,81], adsorption [82,83], membrane [84,85], electrochemical [86,87], and peracetic acid [88,89] are considered to be addressed in removing large amounts of pollutant loads including antimicrobial-resistant bacteria into the aquatic environments.

Limitations
The following limitations can be noted in this study. First, this study evaluated representative antimicrobials that are used in Japan and are considered to have higher con-centrations detected in wastewater. Therefore, antimicrobials belonging to some medicinal classifications, such as sulfonamides [90] and aminoglycosides [91], were not investigated in this study, and for antimicrobials used for both human and animal use [92], there is a possibility that the discrepancy between the measured and predicted concentrations may be large, as mentioned in this report. Second, grab sampling was used for collecting wastewater at the STP in this study, and the concentration of antimicrobials in wastewater measured represents the concentration at a certain time of the day. Therefore, it would be possible that errors are due to diurnal variations caused by sampling methods compared to the continuous proportional sampling and flow proportional mode sampling [93,94].
Finally, the statistical aggregation of pharmaceutical use has limitations in that it does not include information on pharmaceuticals imported and purchased from abroad by individuals, as the regions where pharmaceuticals are prescribed may differ from the regions where they are actually used by patients, and it would also be affected by medication compliance, which assumes that all prescribed antimicrobials are taken [37,38]. Our results support the need for further, conclusive research by taking experimental, technical, regional customs, bias, and unknown factors into consideration.

Conclusions
The predicted concentrations of the representative antimicrobials in the STP influent of Japan by estimation from antimicrobial shipping and NDB volumes generally agreed with the measured concentrations with good accuracy. Additionally, although the concentrations predicted from the sales volumes were somewhat lower than those predicted from the shipping and NDB volumes, the orders of magnitude of the measured and predicted concentrations were roughly in agreement. These results indicate that the concentrations of antimicrobials in the STP influents can be estimated with high accuracy from two assets, antimicrobial shipping and NDB volumes. The fact that there is evidence that the use of antimicrobials has a direct impact on the environments is significant in terms of taking countermeasures for antimicrobials in the aquatic environments. Consideration of the removal rates during wastewater treatment processes at STPs and the dilution and attenuation rates in the receiving rivers should improve the accuracy of this method in predicting the concentrations of antimicrobials in river waters.
These findings highlight the effectiveness of the prediction of concentrations of antimicrobials in the STP influent. The importance of reducing the amount of antimicrobials before they are discharged into rivers was also evoked for maintaining both a clean environment and for human health. This study provides a new means of assessing the environmental risks associated with antimicrobials in aquatic environments.