Consumption of penicillins in the community, European Union/European Economic Area, 1997–2017

Abstract Objectives Data on consumption of penicillins in the community 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 the main subgroups of penicillins. Methods For the period 1997–2017, data on consumption of penicillins, i.e. β-lactam antibacterials, penicillins (ATC group J01C), in the community 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 penicillins 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 penicillins in the community expressed in DDD per 1000 inhabitants per day varied by a factor of 4.9 between countries with the highest (Spain) and the lowest (the Netherlands) consumption. An increase in consumption of penicillins, which was not statistically significant, was observed between 1997 and 2003 and up to 2010. A decrease, which was not statistically significant, was observed from 2010 onwards. Proportional consumption of combinations of penicillins, including β-lactamase inhibitors (J01CR) increased during 1997–2017, which coincided with a decrease in the proportional consumption of extended-spectrum penicillins (J01CA) and narrow-spectrum penicillins (J01CE). Conclusions Considerable variation in the patterns of consumption of penicillins was observed between EU/EEA countries. The consumption of penicillins in the EU/EEA community did not change significantly over time, while the proportional consumption of combinations of penicillins increased.


Introduction
This article presents data from the European Surveillance of Antimicrobial Consumption Network (ESAC-Net, 1 formerly ESAC) on consumption of penicillins for 30 EU/European Economic Area (EEA) countries in 2017 (Table 1). It updates previous ESAC studies published in 2006 and 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. 2,3 In 2017, penicillins represented 42.3% of antibiotic consumption in the community. 4 The objective of this study was to analyse temporal trends, seasonal variation and the presence of change-points in consumption of penicillins in the community (i.e. primary care sector) for the period 1997-2017, as well as to analyse the composition of consumption of penicillins 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 penicillins, i.e. b-lactam antibacterials, penicillins (ATC group J01C, referred to as penicillins in this manuscript) aggregated at the level of the active substance, were collected using the WHO ATC/DDD methodology (ATC/DDD index 2019) 5  There are 46 unique ATC codes for penicillins in the ATC/DDD index 2019. Compared with previous descriptions of the consumption of penicillins in the community, two additional substances, i.e. nafcillin (J01CF06) and aspoxicillin (J01CA19), 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 in consumption of penicillins 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 were assessed through a compositional data analysis modelling yearly data expressed in DDD per 1000 inhabitants per day from 1997 to 2017. 7

Results
An overview of consumption of penicillins (ATC J01C) 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 penicillins in the community in 2017
In 2017, four substances accounted for 90% of the consumption of penicillins in the community expressed in DDD per 1000 inhabitants per day: amoxicillin/clavulanic acid (45.9% in 2017 compared with 42.4% in 2009), amoxicillin (34.8% in 2017 compared with  Figure 1 shows the consumption of penicillins in the community subdivided in the four main subgroups expressed in DDD per 1000 inhabitants per day for 30 EU/EEA countries in 2017. Consumption of penicillins in the community varied by a factor of 4.9 between the countries with the highest (14.23 DDD per 1000 inhabitants per day in Spain) and the lowest (2.92 DDD per 1000 inhabitants per day in the Netherlands) consumption in 2017, which was higher than in 2009 (factor of 3.6, from 10.91 DDD per 1000 inhabitants per day in France to 3.03 DDD per 1000 inhabitants per day in Estonia).
In 2017, NSP represented 10.1% (compared with 15.1% in 2009) of penicillin consumption in the community. Large variations in NSP consumption were found, ranging from 3.8 DDD per 1000 inhabitants per day in Denmark to 0.0006 DDD per 1000 inhabitants per day in Italy. Phenoxymethylpenicillin, commonly known as penicillin V, was the most widely reported NSP in most countries. It represented >50% of penicillin consumption in the community in Sweden (53.8%) and Norway (53.5%), but represented <1% of penicillin consumption in Belgium, Cyprus (total care data, i.e. community and hospital sector combined), Greece, Luxembourg, the Netherlands and Spain, with no consumption reported in Italy, Portugal and Slovenia. A wide variety of other NSPs was reported, which varied depending on the countries. For example, benzathine phenoxymethylpenicillin was mainly consumed in Austria, Croatia, Germany and Slovenia, and benzylpenicillin in Greece and Romania (total care data). Pheneticillin was exclusively consumed in the Netherlands, and procaine benzylpenicillin was exclusively consumed in Spain.
In 2017, PRP represented 5.0% (compared with 4.3% in 2009) of penicillin consumption in the community. PRP consumption ranged from 1.56 DDD per 1000 inhabitants per day in Denmark to 0.0001 DDD per 1000 inhabitants per day in    Figure 2 shows consumption of penicillins in the community expressed in packages per 1000 inhabitants per day for 20 EU/EEA countries in 2017. Based on this indicator, France showed the highest consumption (2.7 packages per 1000 inhabitants per day) while Sweden showed the lowest consumption (0.5 packages per 1000 inhabitants per day). Denmark shifted from rank 6 for its consumption of penicillins in DDD per 1000 inhabitants per day (among the highest prescribing countries) to rank 12 in packages per 1000 inhabitants per day (among the lowest prescribing countries), Spain moved from the 1st to the 4th rank and Sweden from the 15th to the 20th rank, while Italy moved from the 8th to the 3rd rank ( Table 2). The lowest mean number of DDD per package was observed for France (4.9 DDD per package) and the highest for Spain and Sweden Based on the final fitted model, consumption of penicillins in the community in 1997 was significantly above average in Iceland and Slovakia, and significantly below average in Estonia, Finland, Germany, the Netherlands and the United Kingdom (observed profiles shown in Figures S2 and S3). The seasonal variation was significantly larger than average in Belgium, Italy, Lithuania, Luxembourg and Slovakia, and significantly smaller than average in Austria, Estonia, Finland, Germany, Greece, the Netherlands, Sweden and the United Kingdom. The increase in consumption of penicillins in the community between 1997 and the first quarter of 2003 was significantly larger than average in Croatia, Denmark and Poland. The increase in consumption of penicillins between the second quarter of 2003 and the second quarter of 2010 was significantly larger than average in Belgium, Italy and Luxembourg. The increase in consumption of penicillins between the third quarter of 2010 and the last quarter of 2017 was significantly larger than average in Ireland. Compositional data analysis, 1997-2017 The proportional consumption of ESP and COP significantly increased over time relative to that of NSP and PRP. In addition, the proportional consumption of COP significantly increased over time relative to that of ESP ( Table 3).

Discussion
Penicillins (J01C) were the most frequently consumed antibiotics in the community in the EU/EEA in 2017. 4 Consumption of penicillins in European countries that are not part of the ESAC-Net but covered by the WHO Europe Antimicrobial Medicines Consumption Network also was substantial, ranging from 28% in Kazakhstan to 52.5% in Uzbekistan. 8 In the EU/EEA, consumption of penicillins in the community remained high and stable between 1997 and 2017. Inter-country variability of consumption of penicillins in the community expressed in DDD per 1000 inhabitants per day was substantial, and increased when compared with data from 2009. Seasonal variation was high and remained stable over time.
Among the 46 penicillins with an ATC code, more substances were no longer prescribed in 2017 (21 out of 46) than in 2009 (12 out of 44). Overall, COP was the most frequently consumed subgroup of penicillins. Proportional consumption of COP increased in most countries at the expense of consumption of NSP or ESP. Given that total consumption of penicillins did not change significantly over time between 1997 and 2017, this implies that consumption of antibiotics from one subgroup was merely replaced by consumption of antibiotics from another subgroup, rather than being reduced overall. In Belgium, a 10% decrease in consumption of COP was accompanied by a 10% increase in consumption of ESP, most likely as the result of successful multi-faceted campaigning in the country. 9 Countryspecific consumption of COP showed great variability, ranging from 7.85 DDD per 1000 inhabitants per day in Spain to 0.01 Bruyndonckx et al.
ii18 DDD per 1000 inhabitants per day in Norway. The most frequently reported substance was amoxicillin/clavulanic acid, which represented >50% of penicillin consumption in the community in 11 countries. This finding, once again, raises concern about the appropriate prescribing of amoxicillin/clavulanic acid for respiratory tract infections (RTIs). 10 Given that the mean number of DDD per package varied considerably and increased over time, 11 and that antimicrobial resistance best correlates with consumption expressed in packages, 12 we recommend evaluating antibiotic consumption expressed both in DDD per 1000 inhabitants per day and in packages per 1000 inhabitants per day.
With the exception of four penicillins that represented <1% of penicillin consumption in the community in EU/EEA countries, the penicillins used were listed in the 2019 WHO Access, Watch or Reserve (AWaRe) classification list. 13 Most penicillins are listed as antibiotics belonging to the Access group. The antibiotics listed in the Watch group, which include pheneticillin (J01CE05), piperacillin (J01CA12), piperacillin and b-lactamase inhibitor (J01CR05) and temocillin (J01CA17), are mainly used in the hospital sector (consumption of these antibiotics in the community was <1% in the EU/EEA in 2017). Care should be taken by countries to optimize the availability of penicillins that belong to the Access group.
The continued seasonal variation in consumption of penicillins in the community found in this study confirms that penicillins are still prescribed for seasonal RTIs, which represent nearly 60% of antibiotic prescriptions in the community, even though  RTIs are mostly viral in origin. [14][15][16] Based on this finding, acute RTIs remain an ideal opportunity for antimicrobial stewardship activities in the community in EU/EEA countries.
For a detailed discussion on the limitations of the collected data, we refer to the article on antibacterials for systemic use, included in this series. 17 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, neither the consumption nor the seasonal variation in consumption of penicillins in the community changed over time. However, the proportional consumption of COP increased significantly over time during 1997-2017.  Bruyndonckx et al. ii20