Consumption of tetracyclines, sulphonamides and trimethoprim, and other antibacterials in the community, European Union/European Economic Area, 1997–2017

Abstract Objectives Data on consumption of tetracyclines, sulphonamides and trimethoprim, and other antibacterials were collected from 30 EU/European Economic Area (EEA) countries over two decades. This article reviews temporal trends, seasonal variation, presence of change-points and changes in the composition of main subgroups of tetracyclines, sulphonamides and trimethoprim and other antibacterials. Methods For the period 1997–2017, data on consumption of tetracyclines (ATC group J01A), sulphonamides and trimethoprim (ATC group J01E), and other antibacterials (ATC group J01X) in the community and aggregated at the level of the active substance, were collected using the WHO ATC/DDD methodology (ATC/DDD index 2019). Consumption was expressed in DDD per 1000 inhabitants per day and in packages per 1000 inhabitants per day. Consumption of tetracyclines, sulphonamides and trimethoprim, and other antibacterials was analysed based on ATC-4 subgroups and presented as trends, seasonal variation, presence of change-points and compositional changes. Results In 2017, consumption of tetracyclines, sulphonamides and trimethoprim, and other antibacterials in the community expressed in DDD per 1000 inhabitants per day varied considerably between countries. Between 1997 and 2017, consumption of tetracyclines did not change significantly, while its seasonal variation significantly decreased over time. Consumption of sulphonamides and trimethoprim significantly decreased until 2006, and its seasonal variation significantly decreased over time. The consumption of other antibacterials showed no significant change over time or in seasonal variation. Conclusions Consumption and composition of tetracyclines, sulphonamides and trimethoprim, and other antibacterials showed wide variations between EU/EEA countries and over time. This represents an opportunity to further reduce consumption of these groups in some countries and improve the quality of their prescription.


Introduction
This article presents data from the European Surveillance of Antimicrobial Consumption Network (ESAC-Net, 1 formerly ESAC) on consumption of tetracyclines, sulphonamides and trimethoprim, and other antibacterials in the community (i.e. primary care sector) for 30 EU/European Economic Area (EEA) countries in 2017 (Table 1). It updates a previous ESAC study published in 2011, and in doing so it provides updated comparable and reliable information on antibiotic consumption that can aid in fighting the global problem of antimicrobial resistance. 2 In 2017, tetracyclines, sulphonamides Table 1. Classification of tetracyclines (J01A), sulphonamides and trimethoprim (J01E), and other antibacterials (J01X; ATC/DDD index 2019) Oxytetracycline J01AA20 Combinations of tetracyclines b J01AA07 Tetracycline J01AA56 Oxytetracycline, combinations

Sulphonamides and trimethoprim
Trimethoprim and derivatives Long-acting sulphonamides J01EA01 Trimethoprim a J01ED01 Combinations of sulphonamides and trimethoprim, including derivatives J01EB08 Sulfathiourea b J01EE01 Sulfamethoxazole and trimethoprim a J01EB20 Combinations  ii46 and trimethoprim, and other antibacterials represented 11.3%, 2.9% and 6.1% of total antibiotic consumption in the community, respectively. 3 The objective of this study was to analyse temporal trends, seasonal variation and the presence of change-points in consumption of tetracyclines, sulphonamides and trimethoprim, and other antibacterials for the period 1997-2017 in the community as well as to analyse their composition over time.

Methods
The methods for collecting and analysing the data are described in the introductory article of this series. 4 In summary, data on consumption of tetracyclines, i.e. Anatomical Therapeutic Chemical (ATC) group J01A, sulphonamides and trimethoprim (ATC group J01E), and other antibacterials (ATC group J01X) in the community and aggregated at the level of the active substance, were collected using the WHO ATC/DDD methodology (ATC/DDD index 2019 5 ) and expressed in DDD per 1000 inhabitants per day. In addition, where data were available, consumption was also expressed in packages per 1000 inhabitants per day. There are 14, 33 and 26 unique ATC codes for tetracyclines, sulphonamides and trimethoprim, and other antibacterials, respectively, in the ATC/DDD index 2019. Compared with previous descriptions of the consumption of other antibacterials in the community, two additional nitrofuran derivatives, i.e. furazidin (J01XE03) and nitrofurantoin, combinations (J01XE51), and one additional other antibacterial substance, i.e. tedizolid (J01XX11), have been assigned an ATC code by the WHO (Table 1). 2 The evolution of the number of DDD per package over time was assessed using a linear mixed model. The temporal trend, seasonal variation and presence of change-points for tetracyclines, sulphonamides and trimethoprim and other antibacterials were assessed using a non-linear change-point mixed model fitted to quarterly data expressed in DDD per 1000 inhabitants per day from 1997 to 2017. 6 The relative proportions of the main subgroups for sulphonamides and trimethoprim, and other antibacterials were assessed through a compositional data analysis modelling yearly data expressed in DDD per 1000 inhabitants per day from 1997 to 2017. 7

Tetracyclines
An overview of consumption of tetracyclines (ATC J01A) in the community, expressed in DDD and packages per 1000 inhabitants per day for all participating countries between 1997 and 2017 is available as Supplementary data at JAC Online (Tables S1 and S2, respectively).

Consumption of tetracyclines in the community in 2017
In 2017, two substances accounted for 90% of consumption of tetracyclines in the community expressed in DDD per 1000 inhabitants per day: doxycycline (75.8% in 2017 compared with 74.9% in 2009) and lymecycline (14.8% in 2017 compared with 10.2% in 2009) ( Table 1). Figure 1 shows the consumption of tetracyclines in the community expressed in DDD per 1000 inhabitants per day in 2017, which varied by factor 16 between countries with the highest (5.0 DDD per 1000 inhabitants per day in Iceland) and the lowest (0.3 DDD per 1000 inhabitants per day in Slovenia) consumption. This was higher than in 2009 (factor of 10, from 5.1 DDD per 1000 inhabitants per day in Iceland to 0.5 DDD per 1000 inhabitants per day in Italy).
Doxycycline was the only tetracycline prescribed in Croatia, Lithuania and Slovenia and it represented >50% of consumption  Tetracyclines, sulphonamides, trimethoprim and other community consumption, 1997-2017 JAC ii47 of tetracyclines in the community in all but four countries (Belgium, Denmark, Sweden and the United Kingdom, where its consumption represented >30%). Lymecycline was the most consumed in Sweden (51.5% of consumption of tetracyclines in the community), and represented >30% in Belgium, Norway and the United Kingdom, and >20% in Austria, Denmark, Ireland and Finland. Minocycline represented >20% of consumption of tetracyclines in the community in Belgium, Luxembourg and Malta. The substance tetracycline (J01AA07) was the most consumed in Denmark (34.6%) and represented >10% in Finland, Norway and Romania (total care data, i.e. community and hospital sector combined). Tigecycline for parenteral use was consumed in the community in six countries ranging from 0.006 DDD per 1000 inhabitants per day in Greece to 0.001 DDD per 1000 inhabitants per day in Malta. Figure 2 shows consumption of tetracyclines in the community expressed in packages per 1000 inhabitants per day for 20 EU/EEA countries in 2017. Bulgaria ranked 11th for its consumption of tetracyclines in DDD per 1000 inhabitants per day and 5th in packages per 1000 inhabitants per day, Lithuania 14th and 8th, France 3rd and 9th, Sweden 6th and 12th, and Denmark 13th and 18th, respectively (Table 2). Consequently, the lowest mean number of DDD per package was found in Bulgaria (9 DDD per package) and Lithuania (10.2 DDD per package), the highest in France Based on the fitted model, consumption of tetracyclines in the community in 1997 was significantly higher than average in Belgium, Finland, Iceland, Latvia and Poland, and significantly lower than average in Austria, Croatia, Denmark, Italy, Portugal, Slovakia, Slovenia and Spain (observed profiles shown in Figure S2 and S3). The seasonal variation was significantly larger than average in Belgium, Finland, Iceland, the Netherlands, Poland and Sweden, and significantly smaller than average in Denmark, Greece, Ireland, Italy, Portugal, Slovenia, Spain and the United Kingdom. The decrease in consumption of tetracyclines in the community between 1997 and the first quarter of 2004 was significantly higher than average in Belgium, Finland, Hungary and Portugal. The decrease in consumption of tetracyclines between  (Table S1). The largest increases were observed for Greece, Spain (including private prescriptions from 2016 onwards) and the United Kingdom, while the largest decreases were observed for Estonia, Germany and Luxembourg. The increase in consumption of tetracyclines was mainly the result of the increase in doxycycline consumption. For Malta, a substantial increase in lymecycline consumption was also observed. As in 2009, Italy had the lowest consumption of tetracyclines in 2017.

Compositional data analysis (1997-2017)
For the tetracyclines, grouping into subgroups is not available from the ATC classification 5 and variation in the consumption among the different tetracycline substances was not assessed.

Sulphonamides and trimethoprim
An overview of consumption of sulphonamides and trimethoprim (ATC J01E) in the community, expressed in DDD and packages per 1000 inhabitants per day for all participating countries between 1997 and 2017 is available as Supplementary data (Tables S3 and  S4, respectively).

Consumption of sulphonamides and trimethoprim in 2017
In 2017, the two most used substances accounted for 90% of the consumption of sulphonamides and trimethoprim in the community expressed in DDD per 1000 inhabitants per day: sulfamethoxazole and trimethoprim (65.8% in 2017 compared with 67.7% in 2009) and trimethoprim (29.4% in 2017 compared with 29.8% in 2009) ( Table 1). Figure 4 shows the consumption of sulphonamides and trimethoprim in the community expressed in DDD per 1000 inhabitants per day in 2017. Consumption of sulphonamides and trimethoprim varied by a factor of 105 between countries with the highest (1.04 DDD per 1000 inhabitants per day in the United Kingdom) and the lowest (0.01 DDD per 1000 inhabitants per day in Lithuania) consumption (Table S3), which was lower than in 2009 (factor of 203, from 1.18 DDD per 1000 inhabitants per day in the United Kingdom to 0.006 DDD per 1000 inhabitants per day in Lithuania).
For six countries (Belgium, Bulgaria, Croatia, Hungary, Italy and Luxembourg), consumption of sulphonamides and trimethoprim in the community was represented by a single substance, namely sulfamethoxazole plus trimethoprim (J01EE01). In another eight countries, combinations of sulphonamides and trimethoprim, including derivatives (J01EE) represented >90% of the consumption of sulphonamides and trimethoprim in the community, and it represented >80% of this consumption in Germany and Poland. In Ireland and Lithuania, consumption of sulphonamides and trimethoprim in the community was only represented by trimethoprim (J01EA01). Trimethoprim also represented >90% of Tetracyclines, sulphonamides, trimethoprim and other community consumption, 1997-2017 JAC ii49 consumption of sulphonamides and trimethoprim in the community in Iceland, >80% in the United Kingdom, >70% in Denmark and Finland and >50% in Austria. Denmark was the only country consuming a substantial proportion of short-acting sulphonamides (>20%, sulfamethizole). The highest consumption of intermediate-acting sulphonamides (sulfadiazine) was found in Malta (>30%), followed by Iceland and France (>5%), and Cyprus (total care data) and Germany (>1%). Consumption of sulfadiazine was <1% in six countries and no consumption of this substance was reported for 14 countries. In 2017, no country consumed long-acting sulphonamides in the community. Figure 5 shows the consumption of sulphonamides and trimethoprim in the community expressed in packages per 1000 inhabitants per day for 20 EU/EEA countries in 2017. In addition, country ranking is depicted according to both DDD and packages per 1000 inhabitants per day in 2017 ( Table 3). The number of DDD per package ranged from 0.2 in Lithuania to 22.9 in Slovakia (2016 data). A high number of DDD per package was associated with a higher ranking of consumption expressed in DDD per 1000 inhabitants per day than in packages per 1000 inhabitants per day. Slovakia (2016 data) and Ireland, for example, had a higher ranking in DDD than in packages per 1000 inhabitants per day because of the high number of DDD per package. Conversely, Iceland and Lithuania, for example, had a lower ranking in DDD than in packages per 1000 inhabitants per day because of low number of DDD per package. In the EU/EEA countries, the number of DDD per package did not change significantly over time during 1997-2017.

Longitudinal data analysis, 1997-2017
The best fit was obtained for a model including two change-points:

Compositional data analysis, 1997-2017
Although consumption of sulphonamides and trimethoprim in the community significantly decreased over time until 2006 and later remained stable, the proportional consumption of trimethoprim and derivatives (J01EA) significantly increased relative to that of all other subgroups of sulphonamides and trimethoprim. The proportional consumption of long-acting sulphonamides (J01ED) significantly decreased relative to that of all other subgroups of sulphonamides and trimethoprim. In addition, the proportional consumption of short-acting sulphonamides (J01EB) significantly decreased relative to that of intermediate-acting sulphonamides (J01EC) and of sulphonamides and trimethoprim, including derivatives (J01EE) ( Table 4).
Trends of proportional consumption in individual countries are shown in Figure S7. When comparing the composition of the consumption of sulphonamides and trimethoprim in 2017 with that in 2009, we focused on countries reporting consumption in both years (i.e. all countries except Belgium and Slovenia). For the proportion of trimethoprim and derivatives (J01EA), both increases and decreases were observed. The largest increases were observed for Iceland (!40.32%), Ireland (!26.40%) and Poland (!15.82%), while the largest decreases were observed for Finland (#28.57%), Norway (#20.52%) and Sweden (#15.39%). These changes were counteracted by decreases (or increases, respectively) in the proportion of combinations of sulphonamides and trimethoprim (J01EE). The proportion of short-acting sulphonamides only changed only for Denmark (#14.50%). The proportion of intermediate-acting sulphonamides remained stable for most countries, but changes >5% were observed for Malta (!28.01%) and Iceland (!6.69%). Long-acting sulphonamides were not

Other antibacterials
An overview of consumption of other antibacterials (ATC J01X) in the community, expressed in DDD and packages per 1000 inhabitants per day for all participating countries between 1997 and 2017 is available as Supplementary data (Tables S5 and S6, respectively).

Consumption of other antibacterials in the community in 2017
In  Table 1). Figure 7 shows the consumption of other antibacterials (J01X) in the community expressed in DDD per 1000 inhabitants per day in 2017. Consumption of other antibacterials varied by a factor of 130 between countries with the highest (4.32 DDD per 1000 inhabitants per day in Poland) and the lowest [0.03 DDD per 1000 inhabitants per day in Ireland; consumption of nitrofurantoin (J01XE01) not included] consumption (Table S5). This was much lower than in 2009 (factor of 892, from 2.85 DDD per 1000 inhabitants per day in Norway to 0.003 DDD per 1000 inhabitants per day in Bulgaria).
Nitrofuran derivatives (J01XE) represented >50% of the consumption of other antibacterials in the community in 17 countries: >90% in Belgium, Croatia, Czechia (2015 data), Estonia, Latvia,  Lithuania, Malta, the Netherlands, Poland and Slovenia; >80% in Portugal and the United Kingdom; >70% in Austria, Germany and Greece and >50% in Cyprus (total care data, community and hospital sector combined) and Iceland. In Spain and Norway, nitrofuran derivatives represented <20% of the consumption of other antibacterials. No consumption of this subgroup was reported for Bulgaria, Hungary and Slovakia (2016 data). In all but two countries, nitrofurantoin was the only nitrofuran derivative consumed [nifurtoinol in Belgium (24.4%) and Luxembourg (6.0%)].
The subgroup 'other antibacterials' (J01XX) represented >70% of the other antibacterials (J01X) in eight countries (Bulgaria, Finland, Hungary, Italy, Luxembourg, Norway, Spain and Sweden), and it represented >20% in Denmark, France, Germany, Iceland, Ireland [consumption of nitrofurantoin (J01XE01) not included] and Slovakia (2016 data). Methenamine hippurate/mandelate and fosfomycin, and to a lesser extent nitroxoline and linezolid, were also consumed, each country using a substantial proportion of only one of these substances. The highest consumption of glycopeptide antibacterials (J01XA, i.e. parenteral vancomycin and teicoplanin) was reported for Cyprus (total care data, i.e. community and hospital sector combined) followed by Greece, Romania (total care data), Czechia (2015 data) and the United Kingdom.     Versporten et al.
ii54 Figure 8 shows consumption of other antibacterials in the community expressed in packages per 1000 inhabitants per day for 20 EU/EEA countries in 2017. The number of DDD per package ranged from 1.6 in Slovakia (2016 data) to 22.5 in Denmark. A high number of DDD per package was generally associated with a higher ranking in DDD per 1000 inhabitants per day than in packages per 1000 inhabitants per day (Table 5). Denmark and Finland had a higher ranking in DDD than in packages per 1000 inhabitants per day because of the high number of DDD per package. Inversely, Greece, France and Italy, for example, had a lower ranking in DDD than in packages per 1000 inhabitants per day because of the low number of DDD per package. In the EU/EEA countries, the number of DDD per package significantly increased over time during 1997-2017, with the steepness of this increase significantly reducing over the years.
Longitudinal data analysis, 1997-2017 Furthermore, the longitudinal data analysis showed no significant seasonal variation for the consumption of other antibacterials (Figure 9).
Based on the fitted model, consumption of other antibacterials in the community in 1997 was significantly higher than average in Belgium, Estonia, Finland, Latvia, Luxembourg and Sweden, and lower than average in Austria, Croatia, Germany, Hungary, Poland, Slovakia, Slovenia, Spain and the United Kingdom (observed profiles shown in Figures S9 and S10). The seasonal variation was significantly larger than average in Czechia. The decrease in consumption of other antibacterials in the community between 1997 and the second quarter of 2006 was significantly larger than average in Czechia, Estonia, Iceland, Latvia and Luxembourg. The increase between the third quarter of 2006 and the second quarter of 2011 was significantly larger than average in Iceland, Lithuania and Portugal. The increase between the third quarter of 2011 and the first quarter of 2015 was significantly larger than average in Lithuania. The decrease between the second quarter of 2015 and the last quarter of 2017 was significantly larger than average in Czechia.

Discussion
This study updates a previous description of the consumption of tetracyclines, sulphonamides and trimethoprim, and other antibacterials in the community in EU/EEA countries. 2 Most substances reported in this article are prescribed to treat a wide range of infections and acne as well as infections with atypical pathogens such as Chlamydophila pneumoniae, Mycoplasma pneumoniae or Legionella pneumophila. 8 While tetracycline consumption did not change significantly over time, its seasonal variation significantly decreased. Four substances represented 99% of consumption of tetracyclines in the community.
In 2017, the proportional consumption of tetracyclines out of all antibacterials for systemic use ranged from 2.51% in Italy to 28.22% in the United Kingdom. 3 In European countries that are not part of the ESAC-Net but covered by the WHO Europe Antimicrobial Medicines Consumption (AMC) Network, there was also considerable variation, i.e. from <0.1% in North Macedonia to 15.3% in Azerbaijan. 9 While the consumption of doxycycline (usually prescribed for respiratory infections), minocycline and tetracycline (usually prescribed for long-term treatment of acne) decreased since 2009, Given that total consumption of sulphonamides and trimethoprim did not change significantly after 2006, this implies that consumption of one subgroup was merely replaced by consumption of another subgroup, rather than consumption being reduced overall.
In the EU/EEA, the average consumption as well as the seasonal variation of the consumption of other antibacterials (J01X) did not change significantly over time. Nevertheless, there were large variations between countries with other antibacterials representing from 0.17% in Ireland to 27.46% in Norway of the consumption of antibacterials for systemic use in the community in 2017. 3 This might be explained by a preference for treating some UTIs with methenamine, which does not drive antimicrobial resistance. 10 In European countries that are not part of the ESAC-Net but covered by the WHO Europe Antimicrobial Medicines Consumption (AMC) Network considerable variation was also observed, from 1.4% in Albania to 10.8% in Georgia. 9  Figure 9. Estimated trend (solid line) of consumption of other antibacterials (ATC J01X) in the community based on quarterly data from 25 EU/EEA countries, 1997-2017. b 0 , predicted consumption in the first quarter of 1997; b 1 , predicted increase (if positive)/decrease (if negative) in consumption per quarter; b 2 , predicted difference in slope after versus before the first change-point; b 3 , predicted difference in slope after versus before the second change-point; b 4 , predicted difference in slope after versus before the third change-point; b 0 S , predicted amplitude of the upward winter and downward summer peak in consumption; b 1 S , predicted increase (if positive)/decrease (if negative) of the amplitude of the upward winter and downward summer peak in consumption per quarter; d, shift in timing of the upward winter and downward summer peak from one year to another. An asterisk indicates that the result was statistically significant at significance level 0.05. Tetracyclines, sulphonamides, trimethoprim and other community consumption, 1997-2017 In addition, there were large variations between countries and over time for specific subgroups. Consumption of parenteral glycopeptide antibacterials (vancomycin and teicoplanin); polymyxins (colistin) and imidazole derivatives for systemic use (metronidazole and tinidazole) was low in the community in most EU/EEA countries. Consumption of fusidic acid for systemic use, available in parenteral and oral formulation, was also low. Countries where these parenteral substances are prescribed in the community tend to report a high consumption of other antibacterials expressed in packages per 1000 inhabitants per day because each package might represent only one DDD of parenteral antibiotic. As such, Greece ranked high for its consumption of other antibacterials expressed in packages per 1000 inhabitants per day because of its high consumption of parenteral metronidazole, colistin, vancomycin and teicoplanin, which could be related to outpatient parenteral antibiotic therapy (OPAT) to treat pneumonia, UTIs and gastro-intestinal tract infections in Greece. 11 Italy is also known for its consumption of parenteral antibiotics in the community, often administered at hospital infusion centres. 12 Teicoplanin and ceftriaxone were the most commonly administered parenteral antibiotics at these centres. 13 Esposito et al. 14 predicted that the demand for OPAT is likely to further increase in Italy due to the potential saving in hospital care costs and improvements in the allocation of limited healthcare resources, with an increased consumption of teicoplanin until 2010. 15 We confirmed that indeed teicoplanin for OPAT increased up to 0.06 DDD per 1000 inhabitants per day in Italy until 2010, but decreased steadily thereafter to 0.03 DDD per 1000 inhabitants per day in 2017.
For France and Italy, the higher ranking for their consumption of other antibacterials expressed in packages per 1000 inhabitants per day was also related to the consumption of fosfomycin, which is dispensed as one dose per package and is commonly prescribed for the treatment of UTIs in these countries. Some guidelines may recommend oral fosfomycin as first-line therapy for the treatment of uncomplicated cystitis. 16 In many countries, resistance to fosfomycin remains low in uropathogenic Escherichia coli, while increasing resistance to trimethoprim/sulfamethoxazole, which is still widely used as the first-line agent for the treatment of uncomplicated UTIs, has been reported. [17][18][19][20] Belgium scored high for its consumption of other antibacterials expressed in DDD per 1000 inhabitants per day and higher when expressed in packages per 1000 inhabitants per day, which is related to the high consumption of nitrofuran derivatives (mainly nitrofurantoin) prescribed to treat uncomplicated cystitis as well as for the long-term prophylaxis of UTIs. 21,22 The absence of consumption of nitrofuran derivatives reported for Bulgaria, Hungary and Slovakia (2016 data) might be explained by the fact that nitrofurantoin had to be imported (due to manufacturing problems) and therefore was not included in the sales data reported by these countries.
Finally, six substances for parenteral use, i.e. tigecycline (J01AA12), colistin (J01XB01), fosfomycin (J01XX01), linezolid (J01XX08), daptomycin (J01XX09) and tedizolid (J01XX11), for which consumption was reported in 2017, are listed as Reserve antibiotics in the 2019 WHO Access, Watch or Reserve (AWaRe) classification list. Six countries, among which two provided total care data, i.e. community and hospital sector combined, reported consumption of tigecycline. There also were large variations in the consumption of colistin among the 22 countries reporting consumption of this antibiotic in the community in 2017. The AWaRe classification can be used in supporting antibiotic monitoring and optimal prescribing. Individual countries should follow up consumption of Reserve antibiotics, which are considered as 'last resort' options for the treatment of confirmed or suspected infections due to multidrug-resistant organisms. These should be prioritized as key targets in national stewardship programmes. 23 For a more detailed discussion on the limitations of the collected data, we refer to the article on antibacterials for systemic use, included in this series. 4 For a discussion on the limitations of the statistical approach used in this study and potential explanations for the common change-points detected through these analyses, we refer to the tutorial included in this series. 6 In conclusion, as for all other main antibiotic groups, there were large variations between EU/EEA countries in the consumption and composition of the consumption for the tetracyclines, sulphonamides and trimethoprim, and other antibacterials in the community. This represents an opportunity to improve the quality of prescription of these antibiotics in many countries. Versporten et al. ii58